4 Ideas for Improving Employee Stock Compensation


  • Stock compensation increasingly matters if a country wants successful startups
  • Stock compensation is partly a tax problem
  • Stock compensation is complicated by having different shares for investors and employees/founders
  • Simpler tax deferral laws for stock compensation would help
  • Other ideas include extending stock option expiry, using a single class of stock, or sharing upside via tokens.

Stock compensation increasingly matters if a country wants successful startups

An increasing portion of jobs in the world are software developer jobs. The financial return you can get from money is now more limited than the financial return from being able to write high quality code.

We are seeing this play out in a few ways:

  1. Software companies like Google and Facebook are highly profitable, so profitable that they could double or triple employee salaries (which are well into six figures) and still be highly profitable. This is putting huge upward pressure on salaries.
  2. Demand for software engineers is so high that high hourly rates are paid for software in countries with very low GDP. For example, I’m now seeing hourly rates of up to $50/hr in countries with less than $5,000 per capita. Slowly, software salaries seem to be globalising.
  3. 50 years ago compared to now, money was of more relative value than founders/coders. Back then, investment terms skewed towards investors. Now, terms (and valuations) skew towards founders. Founders and ideas have become more scarce relative to capital.

As a corollary, software is pulling talent from non-software jobs. This is reducing supply in non-software jobs and putting upward pressure those salaries too. (Why would a mechanical engineer accept $50k if they can get a software job that will train them and pay them $80k [numbers do still depend on geography, but less and less]).

This is where stock compensation comes in for early stage founders. When you run an early stage company, you can’t pay salaries like Google. Stock compensation is your way to compete for employees. A stock compensation framework that is simple to understand and taxed fairly is highly valuable to founders.

The tax problem with stock compensation

Why (deferred) stock compensation should be taxed at a lower rate than income

When employees receive stock, they receive this in return for their work. On the surface, it seems stock compensation should be treated as income and taxed as income tax.

However, the benefit of stock to an employee is typically delayed (if they receive any benefit at all). They can’t cash in on their gains until the stock becomes liquid (which means an acquisition or the company going public). So, there is an inflation aspect to stock compensation that isn’t there for a monthly salary.

This inflation argument also applies to the case of why capital gains should be taxed at a lower rate than income tax. For example, if there is an annual rate of inflation of 2%, the associated tax on stock compensation over a period of 10 years of almost 22%.

There is a separate argument for having a lower tax rate on stock compensation, which is to incentivise entrepreneurship and/or employee ownership. I won’t get into that here and don’t have strong views either way.

Lastly, given capital gains is taxed at a lower rate in most countries, it seems unfair to startup employees to have to pay ordinary income tax on all of their gains, when investors who buy the shares get capital gains treatment. For reasons I’ll outline below, non-investor stockholders in startups that are not yet public often end up paying income tax on stock gains rather than capital gains.

What happens in practise with taxation of employee stock compensation

One specific problem with giving employees stock in a startup is that the shares are not liquid (they cannot easily be sold). If you simply grant shares to an employee, they have to pay income tax when they receive those shares today, even if those shares end up being worthless (in which case they could claim a capital gains loss, but not a write off on their income tax).

To avoid this issue, companies that are not yet public (and whose shares are not trade-able) often give employees stock options. This solves the problem of the employee paying tax when they get the shares, but introduces the problem that they will have to pay tax if and when they elect to exercise those stock options (assuming the value of the company goes up when they elect to exercise them). Further, the rate of tax paid on any paper gain made upon purchase of their shares is the income tax rate, not the lower capital gains rate.

Technically there are ways for employees to avoid paying this income tax on execution if strict rules around holding periods are followed. In the US, it is also possible to get capital gains treatment from the time of share purchase until the point of sale. The strict requirements include holding periods and sometimes that employees be still be employed at the same company when the exercise their options. In short, strict rules mean that most employee stock options end up being taxed in large part as ordinary income.

The 409a valuation and the giraffe

One of the big bureacracies of stock compensation in the US revolves around how to value stock options (or stocks) given to employees – particularly for companies that are not public. This process is determined by tax law, is called a 409A valuation, and – in practise – is well exemplified by this conversation my founder friend had with a company that can do the 409A valuation for you:

My friend to 409A advisor: “Ok, so here’s how you’re saying the 409A valuation works… You walk into a room with a giraffe and the giraffe asks you for your financial statements and all of your previous fundraising information. The giraffe takes a week and comes back with some more questions on your fundraising records. The next week, the giraffe comes back again but this time simply takes the price investors paid for shares in the last round of funding, divides by four, and uses that as the price at which you grant employee stock options… Is that how it works?

409A Advisor: “Yeah, that’s basically right, but it’s not a giraffe, it’s an elephant.”

my founder friend

Note that I’m not proposing a solution here (yet). I also want to be clear that this isn’t an easy problem. Valuing the shares in a company that is not yet liquid/traded is not easy – hence why I later suggest we should think about making it easier for companies to let their stock be tradeable.

What the above story illustrates is that rules around stock compensation – in their bid to provide a lower effective tax rate – have created enormous bureaucracy (and costs) for companies, as well as lots of confusion for employees.

One further point around the factor of four in the story above. The reason employee stock options are valued lower than investor stock is because investors typically have preference shares, which means they get their money back before employees and founders get anything. There is a good rationale to this (that I’ll dive into later), but it adds to the confusion for employees around how to value their options – and – the more confusion for employees, the less they will value the stock compensation, and the weaker the stock compensation becomes as a useful tool for founders.

Ideas for better systems:

  1. Government level: A simple rule allowing for tax deferral until sale

I provide this as an illustrative example because the problem isn’t a simple one, and deeper thought is needed from others that understand taxes and politics better than I.

One simple approach might be to forego all of the current benefits around stock options and instead allow recipients of stock to defer the tax on any stock (or options) received until they sell that stock. When the stock is sold, they pay income tax on any proceeds up to and including the value when they received the stock, and capital gains for anything above.

At the time of issuance, the stock could be valued with current bureaucratic process for valuing employee shares (often common stock), or, ideally some laws are passed that employe some simple rules of thumb for companies at various stages.

2. Founder level: Long expiry dates on employee options

In this article, I’m trying to look holistically at how founders can make their shares more valuable as a tool to incentivise to employees. Tax laws are outside of a founder’s control, so this idea and the next two focus on things a founder could do.

This first idea is something that is already happening with reasonable frequency. The idea is to provide employees with a very long (10 year) expiry period on their stock options, and allow them to keep any vested stock options even if they leave the company. Traditionally, startup employees have been forced to exercise (pay to buy) their stock options within a few months of leaving [both by tax law, and by companies]. Given employees move jobs a lot, and given the high demand for employees, it makes sense to me to make stock option terms more friendly [even if the tax laws in the US still force employees to buy their stocks when they leave to gain the tax benefit].

3. Founder level: All common stock

I mentioned above that investors often have a different type (known as class) of shares than founders and employees. Investors typically have preferred shares which get their money back before excess proceeds are shared around. For example, if investors put in $20M and the company sells for $15M, then investors get $15M back and founders/employees don’t get anything. (In practise, this may differ because there may be an incentive to keep founders/employees on board).

There is a good logic to having investor-specific stock because it protects investors. It means that if a founder raises $20M and then sells the company for $10M in the next twelve months, the founder and employees don’t get a share of that $10M. However, it also means that in a scenario where a company sells for an amount that is close to the total money raised to date, then employees don’t have any assurance of a payoff (of course, management could structure something for employees near the time of sale – but certainly employees have no leverage in the stock agreements). Further – and more importantly – it makes it very complicated (and generally impossible) for employees to to value their shares. Employees don’t have access or knowledge to understand the detailed payout structures with two classes of shares.

So, while having a preferred class of shares for investors does protect investors, it makes the common stock (which is what founders and employees receive) less valuable economically. More critically, two classes of shares makes employee stock much more confusing (which additionally makes it less valuable). Therefore, in a market that is shifting towards founders/employees, founders might consider moving to having just one class of shares for everyone (favouring employees/founders and disfavouring investors).

There are companies that do this, although they are currently largely limited to second time founders with a good track record. Pillar.vc are one VC firm that invests on this basis, and I think this will become more common.

In short, by giving everyone common stock (including employees), one takes out the complexity of employees trying to understand what return they will get in the case of the company being sold.

4. Tokenisation of upside

This is where you roll your eyes, and roll your eyes you should, but I think there is some value in thinking about tokenisation of upside as a way to provide employee compensation. I wouldn’t do it myself right now because I don’t see a very clean way to do it (that is compliant and sensible), but I would consider it in future.

It’s best to understand tokenisation through an example. Uniswap is an exchange for crypto currencies. 0.05% of all trades made on the platform accrue to the owners of Uniswap tokens*, and there is a fixed maximum supply of Uniswap tokens. So, by buying uniswap tokens, you are buying upside (revenue potential) in Uniswap’s protocol. Uniswap dishes out these tokens to platform contributors for the work they do.

So, by creating tokens and attaching revenue streams to them, you can create a separate way to share upside (tokens) with employees/founders. [Side note: The difference between employees and contractors is shrinking, so I see that laws, including stock compensation laws, will need to adapt to this.]

Now, you might say “Aren’t those tokens just unregulated shares?”. The answer to the spirit of that question is “yes”. Token based approaches are largely doing what usually is done via a public stock listing, but in a much simpler way, albeit unregulated. One specific near term regulated path I see for founders is creating tokens that only accredited investors (people with minimum net worth) can buy**. This kind of approach might lend itself to liquidity while potentially being compliant with securities laws.

This brings us back to a deeper point about early stage startup shares – which is that they are not liquid/tradable. All else being equal, something illiquid is less valuable as a form of compensation, and raises the question of why startups can’t be publicly traded earlier.

The reason startups don’t go public earlier is because:

  1. It’s too expensive. The documentation and filing requirements and ongoing cost of listing on an exchange are at a level that it doesn’t typically make sense to go public until your company is worth billions.
  2. They don’t want to. A lot of founders and investors would say they don’t want to go public because they don’t want the pressure of having to provide quarterly reports as well as public scrutiny. Largely, this is associated with the expense and bureaucracy of the current laws for public companies.

The current tokenisation we are seeing in crypto should make us realise that we can come up with cheaper and simpler regulated processes for companies to go public. Afirst step would be to create a new class of simple laws for small companies (worth less than $1 billion market cap) to go public and allow them to trade their tokens on open source exchanges that comply with standards.

*Technically, there is currently an option for the community to turn on this fee, so it’s not there right now.

**I have mixed feelings on current US laws that restrict startup investing to people with a minimum net worth or income. It may protect some people from bad investments but it seems exclusionary. It’s hard to weigh these two factors but I would lean towards removing income and net worth requirements.

Stock Compensation Increasingly Matters

In summary, people remain important in building successful startups. The value of what people can offer is – largely thanks to software – increasing, so workers (especially coders) have more leverage. Stock compensation is one of a founder’s most valuable assets for recruitment.


October 2021 Newsletter: Adversarial views and interviews

In this month’s edition:

  • Adversarial interviews
  • Hiring people you know
  • Who gets the best bang for the buck?

Adversarial interviews

With blogs, podcasts and social media, more people are becoming journalists. Journalists who traditionally worked for companies are increasing working for themselves. This is the first year where I’ll have paid more for subscriptions to independent newsletters/artists/entertainers (e.g. Byrne Hobart, Matt Iglesias, Vivienne Aerts) than to large companies (e.g. Financial Times, Spotify).

This trend gives us all a much wider variety of sources. However, I find there is a tendency for these independent types of content (including this newsletter) to lack a presentation of – and interaction between – opposing viewpoints.

Much content today is either a soliloquy or has a soft interviewer allowing the interviewee to present their views without much pushback. One of the most successful recent podcasters is Joe Rogan, who is highly skilled at interviewing in a supportive manner that makes guests comfortable presenting themselves in a more unfiltered form than they otherwise would. This kind of interview is valuable – and Joe Rogan shouldn’t change his style – there is a place for and a lack of more adversarial content within independent media.

As two pieces of high quality adversarial content (and I mean adversarial in a helpful way), I suggest the following:

  1. Tyler Cowen’s recent interview with Amia Srinivasan on the topic of feminism & sex work.
  2. A recent debate on Bankless between advocates of Bitcoin and advocates of Ethereum (technical).

In addressing a topic, we often look at who stands on each side of the issue and what is their political persuasion (I tend to do this too). Rather than engage with the issue, we assume that the answer to a problem comes down to our preferred philosophy. This makes progress difficult and is where adversarial content can help.

Hiring people you know

Over the last three weeks I’ve been looking for a coder to help build a tax/accounting tool. After interviewing about 10 candidates from a combination of AngelList and http://www.gun.io (a kind of recruiting service), I made a hire based on a recommendation from a Celo community member I’ve gotten to know well over the past year.

It is well known that referral systems are effective for hiring. Hence, many large corporates offer bonuses – sometimes over $10,000 – for employees who make referrals. My takeaway from this recent hiring experience is that – even when you think you have fully mined your network – mining your network even more is probably still more effective than trying to use hiring websites or recruiters. This is because referrals capture a lot of the long-term risks and benefits that interviewing cannot owing to the short term nature of any interview process.

Side note: Recently, hiring people you know has become somewhat of a bug rather than a feature. If a company only hires people its employees know, doesn’t that result in a monolithic pool of employees in terms of views and backgrounds? Yes, I think this can happen. However, a further question to ask is whether it’s the “hiring people you know” that is the problem or whether it’s the existing employee pool or culture (and it’s narrowness of network) that’s the problem. When starting a business, the networks of the initial employees really matter, because this is the starting direction for the future network of the company.

Elon Musk: Who gets the best bang for the buck?

You may have seen this crazy tweet early this week:

The cynical view circulating is that Elon Musk wants to sell some of his Tesla stock at the peak in order to diversify his holdings, and he is using this poll to get public approval to do so. I don’t think that makes a lot of sense. When Musk made his first eight figure fortune from the sale of Paypal (of which he was a founder), he put that money into an electric car company, a solar company and a space rocket company. That’s not the kind of thing you do if you are optimising for diversification.

The more interesting question to me is who gets the best bang for the buck when it comes to spending money? In this case, spending money to tackle the energy transition from fossil fuels.

Elon Musk put $70M into Tesla originally and today Tesla is worth over $1 trillion. One can argue that Tesla is massively overvalued, but it is harder to argue with the impact that Tesla has achieved. In June 2021, the Tesla model 3 was the top selling car in Europe. All of the next nine top selling cars are petrol/diesel.

What Musk can do with $1 is probably 10X to 100X what the US government can do with $1 in terms of impact. This is not a criticism of the US government or of governments in general. Governments and entrepreneurs can be complementary. The point is that there are individuals (and also groups of individuals) that are extremely effective in what they do. These kinds of people will play a very important role in tackling challenges like transitioning from fossil fuel and bringing up the living standards of the poorest parts of the world. As an entrepreneur, I think it’s sad so many people feel Musk should have to sell his shares here. I think it is shortsighted for overall progress. The request, but more-so the sentiment, is a deterrent to entrepreneurship.

Side note: I think there is an argument that governments should be spending more on targeted R&D initiatives, like was done for the Manhattan project (but not for making nuclear bombs). Given the choice between spending $4 trillion on a general welfare program and giving $400 bn each to 10 different project managers (that are given very high autonomy), my sense is that general welfare does better in the long run in the latter scenario.

That’s it for this month! Cheers, Ronan

P.S. Each Friday I post a list of links to interesting articles from the past week. If you’d like to get that list by email each week, you can subscribe here (this is a separate mailing list to my monthly newsletter on ronanmcgovern.com).


Weekend Links: Friday 5th Nov 2021

  1. Five Tweet Tuesdays. I subscribe to this guy’s short weekly newsletter (Shaan Puri). Very self-confident, and can be a bit much, but the brevity and quality of the tweets he shares is excellent.
  2. Beautiful visual explainer of gas pipelines and shortages in Europe. (From the Financial Times, not paywalled).
  3. Interview by an independent journalist of Mark Zuckerberg on Facebook’s rebranding to Meta.
  4. Story about $200M low/no windows student dorm funded by Warren Buffett’s Lieutenant Charlie Munger. I don’t know what to make of it. No windows doesn’t sound fun.
  5. Story from earlier this year on how Tesla uses stock options to avoid unionisation in its German factory.
  6. List of the top selling cars, by number, in Europe in June 2021. 1. Tesla Model 3; 2. Renault Clio; 3. Dacia Sandero (never heard of that one).

I asked coder friends how to get into coding

They told me:

  1. Learn by doing. Find a small project you want to build and just get started, probably with Python.
  2. When you get errors in your code, type them into Google and they will often be answered on sites like StackOverflow.
  3. Start a github account (an online profile where you can store your code). Keep adding your projects here. Employers can look at your profile to see what and how often you build stuff.
  4. Contribute to open source projects. You can find these on Github or by googling around. Help solve issues in open source code and build relationships with other coders that way.
  5. Get involved in Reddit and Discord communities for software projects. This lets you find small projects, including many paid.

Things not to do:

  • Long software courses
  • Reading about coding instead of just going ahead and coding
  • Building stuff you don’t think is fun

Right now, it’s a builders market. There is a lot of money out there – not just salaries and contracting – but even grants from community funds (e.g. Celo Community Fund [I got one!] or Gitcoin).

Lastly, I don’t think you can be too young or too old to code.

Lastly lastly, here is a 10 minute video covering some more detailed tips on learning how to code:

And please do comment with any other tips or ideas.


Weekend links: October 29th 2021

  1. Time lapse of COVID spreading across US states in 2020 and 2021. Very slick.
  2. 10 minute video on how to learn to code.
  3. Am I a Bitcoin Maximalist?

Weekend Links October 22nd 2021

  1. Parker Conrad founded Zenefits (a software HR company) and resigned as CEO. He then started Rippling, which does very much the same thing and is now worth more than Zenefits. An overview here.
  2. Bitcoin ETFs should not exist, by Robert Armstrong. It’s true that buying Bitcoin via an ETF defeats the purpose of it’s decentralisation. Armstrong makes some other good points too.
  3. How frequently is crypto used to finance terrorism? A scratch the surface article from Coinbase here.
  4. Zillow runs into trouble flipping houses itself.
  5. MIT MechE alums building a big industrial separation technology business, Shreya/Brent at Via Separations.


Weekend Links: Friday 15th Oct 2021

  1. Stripe get back into crypto. I’ve written before how Stripe pulling out of crypto some years ago was bearish for crypto. Now they are back in. FOMO? I’m confused, but generally my assumption with Stripe is that whatever they are doing is probably smart.
  2. A time capsule of my sense of humour ten years ago. I was reminded of this by a reader who was digging deep in the ronanmcgovern.com archives.
  3. A recent interview with Lance Armstrong that gets into the biology of doping. Highly recommended.
  4. Is there strong evidence Facebook is doing harm? A pro-Facebook view. My view is that too much social media is probably harmful, but I think it’s a lot more difficult to measure the harm than media headlines imply.
  5. Google/Facebook ban ads that deny climate change. Would my piece discussing climate change be allowed according to their rules? Probably yes, but I’m generally concerned with this direction of blocking discourse – even, and maybe especially, of things that are commonly accepted.
  6. My latest blog describing a better way for governments to support charities.
  7. I’ve got a part-time job opening for a front end engineer on a new startup project. Any referrals are much appreciated.

Job Opening: Front End Engineer

Hi folks,

I’m starting to build some software accounting tools to make life easier for businesses that run on blockchain (as opposed to traditional bank accounts/Quickbooks/Xero etc.).

I have a part time contractor gig open for a front end engineer experienced with React, Vercel and Python. The gig will start at 10 hours per week, with opportunity to ramp up.

The full job posting on AngelList is here, and you can email me at (hello at celo dot tax) with questions or interest in applying.


A Better Way for Governments to Support Charities: Quadratic Charitable Deductions


  • I describe a method for governments to distribute charitable tax subsidies so that they favour smaller charities.
  • Instead of government tax breaks being proportional to the donations a charity receives, they would be proportional to the square root of what they receive.
  • This system means a charity that receives $1,000,000 in annual donations will get 10 times the tax subsidy of a charity that receives $10,000, instead of 100 times.
  • Said differently, with quadratic charitable deductions, government provide a disproportionate boost to small charities (and donors).
  • Mathematically, the approach encourages people to donate proportionally to how much they care about a cause.
  • Practically, the approach will make it a lot more attractive to start a new charity.

How US and Irish tax systems currently work for donations

In the US, you don’t pay federal taxes on income that you directly donate to a charity. In Ireland, you pay taxes on income that you donate, but the charity is allowed to claim back that tax. Mathematically, both approaches are the same.

As an example: In a country with a 50% income tax rate and a tax deduction for charitable contributions (whether US style or Irish style), half of the money received by charities is a subsidy from the government:

  • Charity A, receiving E1,000,000 of (after-tax) donations, gets a government subsidy of E1,000,000.
  • Charity B, receiving E10,000 of (after-tax) donations, gets a government subsidy of E10,000.
  • In total, the government subsidy to these two charities is E1,010,000.

This is roughly how charitable deductions work in the US and Ireland today.

An example of Quadratic Charitable Deductions

Here is an improved way to split the total government subsidy of E1,010,000 given to charity A and charity B:

  • Split the subsidy according to the square root of after-tax donations received by each charity.
  • Charity A: The square root of 1,000,000 is 1,000.
  • Charity B: The square root of 10,000 is 100.
  • So, the total subsidy of E1,010,000 gets split in the ratio of 1,000 to 100 (or 10:1), not $1,000,000 to $10,000 (or 100:1).


  • Charity A (with E1,000,000 of donations] gets a tax subsidy of E918,182.
  • Charity B (with E10,000 of donations] gets a tax subsidy of E91,818.

This is a small percentage reduction in income for charity A, but a large percentage increase in income for charity B. The whole system massively favours smaller charities and greatly increases the incentives to start new small charities.

A caveat on the Quadratic Charitable Deduction

With the above approach, there is an incentive for large charities to reorganise as a group of smaller charities. I think this issue is real, and worthy of consideration. However, I suggest the two following mitigating factors:

  1. There is an administrative burden to running one charity as two (or more) separate charities. This provides a natural (albeit limited) deterrent to charities splitting up.
  2. Governments can take a compromised policy approach and – instead of making subsidies proportional to the square root of after-tax donations – subsidies could be proportional to a 0.6 or 0.7 or 0.8 power of donations received.

Implementing this in practise

I don’t think a government like the US or Ireland would adopt this approach overnight, or even over years, although I am optimistic governments may move this way on the scale of decades. It’s worth noting that, in the software world, these kinds of funds already exist to support the development of open source software. Gitcoin is a significant, and growing example. You can read about Gitcoin grants involving millions of dollars of funding on Vitalik’s website.

In the meantime, there is a way to implement this system privately that I am thinking about. Quite simply, I could set up a charity that donates funds – using a quadratic method – to all registered Irish charities based on the prior year’s reported donations. It would be a little bit like an index fund for charitable giving, but incentivising smaller charities.

Over time, I think it would be possible to move away from a fund that is bounded by national borders, but the list of Irish registered charities provides clear information for a starting point.

Let me know if you would be interested in collaborating on a preliminary analysis of how a quadratic fund for Irish registered charities might work. A first step would be to figure out how to import a list of Irish charities and the total after-tax they each received in 2020. Further reading on Gitcoin’s experience with the approach would also be smart.


  1. Irish charitable tax breaks work a little differently than how I have described above because a flat tax rate is assumed in calculating the charity’s tax back (not the specific tax rate that the donor paid).
  2. In the US, there is a charitable deduction at the federal level, but not always at the state level.
  3. For a theoretical dive into quadratic voting and payments, I highly recommend reading Vitalik’s primer on quadratic payments, and the advantages of that approach over “one dollar one vote” and “one person one vote” as a way to fund public goods.

Weekend Links – Friday 8th Oct 2021

  1. Ginkgo Bioworks, an MIT spinout that recently went public via SPAC gets attacked in a short seller report by Scorpion Capital. Very aggressive language.
  2. Do you ever answer a question by saying: “That’s a great question!“. You probably won’t do that so often after listening to this hilarious podcast with Steven Dubner of Freakonomics.
  3. A scathing review of Eleven Madison Park‘s new vegan menu by Pete Wells: “In tonight’s performance, the role of the duck will be played by a beet, doing things no root vegetable should be asked to do.” (Paywalled).
  4. Modern strategies for charitable donation and public funding, notably “Quadratic Funding” by Vitalik Buterin. Something I’m thinking about doing.
  5. An Econtalk podcast covering two ideas on how to improve public governance and discourse – the second is the idea of creating a “fantasy public intellectual league“, a bit like fantasy football, but public intellectuals are scored according to how self aware and considering they are of counterarguments and empirical evidence.
  6. Quite a confrontational and highly recommended podcast on Utopian Feminism with Amia Srinivasan and Tyler Cowen.

September 2021 Newsletter: Why bother with NFTs?

In this month’s edition:

  • My First Virtual Exhibition – with NFTs!
  • Is Lab-Grown Meat Viable?
  • Stripe is eating Quickbooks

Sagrada Barcelona – My first virtual photography exhibition

Last week, after a trip to Barcelona, I pulled together a virtual exhibition of photos and short videos. You can check out the virtual experience for free here – it’s pretty short. Let me know your favourite!

This virtual art gallery pulls together a few technology trends:

  1. Virtual Reality – a major area of investment for Facebook. Virtual Reality holds a lot of promise because the experience is far more immersive than looking at a flat screen. My virtual exhibition is a bit clunky, but within ten years I think we’ll be viewing these with virtual reality goggles or glasses. Maybe within twenty years we’ll have some kind of eye implant for immersive experiences.
  2. Non-fungible tokens (NFT). I have turned each digital image/video into a digital token that can be sold or auctioned (example of one of mine here). NFTs are a big area of hype and speculation, so let me give a little background on where I think there is value.

My take on NFTs

Art is an alignment among a group of people, however big or small, that something is valuable for aesthetic reasons. Non-fungible tokens are unique digital identifiers for a given piece of art. For example, a digital photograph might have an identifying code of C789AB89D.

The list of different pieces of art along with their owners is called a blockchain. That blockchain/list is stored on the internet and anyone can download the list to see who owns what. If you own a piece of art that is listed, you can transfer ownership to someone else by signing off on a transfer with a passcode (called a private key).

So, NFTs are just digital codes for pieces of art, and a blockchain is just a public list of who owns what. Why is it useful to have a public list of art?

  1. The list* is public and not controlled by any one government or company. The list is accessible to all people in all countries (if they have internet). Nobody can be blocked from reading the list, or contributing by receiving or sending tokens.
  2. Typically when an artist sells a painting, they only make money up front – i.e. when the painting is re-sold, it’s the owner of the painting, not the artist, that benefits. With digital ownership lists, artists can get a percentage cut of the transaction every time their art is sold.

Can other people copy my digital image for free?

Using digital lists/blockchains and NFTs is not a replacement for copyright laws. Yes, somebody can use a digital image that you own, but they would typically be breaking copyright and usage laws in many countries. The list/blockchain doesn’t enforce any of these rights (although they allow you to check if a given image is owned by someone else).

For example, if I buy an NFT for a Mickey Mouse picture, typically I would own the rights to that image (and the list/blockchain can prove that), but for any enforcement of my rights, I would to resort to state laws.

Where is the digital photo associated with my NFT stored?

The art/content/photo/video itself is not stored on the blockchain. The blockchain is just a list of items and owners. The image itself is typically stored by the marketplace where you buy it (e.g. opensea.io). Further, it is common for the same photo to be stored in many places/marketplaces across the internet, and owning the NFT gives you ownership of all of these copies. Here’s how this works:

  1. Lets say you own a photo called BigLabrador.jpeg , which is a digital file.
  2. A digital identifier (i.e. the NFT) is created for your photo by taking the digital file and doing a calculation (called a hash) on that file’s data to come up with a unique code – let’s say BA6789EA. This code is your NFT.
  3. There may be many identical copies of that NFT on the internet. If they are identical, then doing the hash calculation on the digital file will lead to the same code BA6789EA.
  4. By owning an NFT, you own all of the files that have a certain unique code.

In summary, public lists for ownership of things makes good sense because a) they provide the best access to everyone and b) it’s easier to work with digital records that are global than manual or siloed records that are just controlled and/or formatted by one government or country. With public ownership lists, you can have global lists of art, of houses, of cars etc. that are accessible to everyone.

* There are actually a number of lists/blockchains you can choose from if you want to put your art on a public list. The most common right now is called Ethereum. It is possible that different protocols will specialise in different types of list (e.g. art versus property). It is also likely that there will be one protocol that will dominate (currently Ethereum is dominating for NFTs).

Is lab grown meat viable?

I like the taste and nutrition that meat provides. I’m an optimist on technology and feel we have a chance of making lab-grown meat that tastes as good as meat from animals.

This article on the cost of lab grown meat is highly skeptical that it can be anything more than a very high end product, and forced me to reconsider my optimism.

The core argument of the piece is that lab grown meat involves scaling up a pharmaceutical grade process orders of magnitude larger than has been done before and – even under aggressive assumptions – any reasonable estimate at artificial meat costs comes in far above the price of meat we eat today.

One framework for thinking about the cost of lab grown meat is to compare it with the efficiency of animals we breed. Knowing it takes about 10 calories of feed for every calorie of chicken protein or 25 calories of feed for every calorie of beef protein, we can compare these numbers to the calories of energy it takes to make one calorie of artifical protein.

A 2021 study by CE Delft, they considered a baseline scenario for a large scale facility producing 10,000 tons of artificial meat per year. They estimate a need for 220 million kWh of electricity to run this plant each year. This works out to about 22 kWh per kilogram of meat (which is about 19,000 calories). One kg of meat (let’s say beef) has about 2,500 calories. So, the study’s baseline scenario for scaled up artificial meat production is about 7.5 calories of electricity for every calorie of artificial meat delivered. This doesn’t include energy for raw materials or transport, but doesn’t seem entirely awful compared to breeding chickens or cows.

Unfortunately, the baseline price estimate for at-scale artificial meat is well above $10,000 per kg of meat when all costs are included. This begs the question of how companies like Impossible Foods are surviving.

My assumption here is that the capital cost of building factories to make the artificial meat – and possibly the energy and emissions in building these factories – are not fully being built into the footprint and cost of the meat. When impossible burgers is selling burgers at $12 in a sandwich shop, but a 2021 study puts the costs at over $20k per kilogram, there must be something hidden somewhere.

In conclusion, I’m somewhat optimistic we’ll crack artificial meat over a long period of time – maybe 50 years, but in the meantime I suspect early companies may become vegan guinea pig casualties of innovation.

Stripe is Eating Quickbooks

In the US, it is common for startups to use Quickbooks to manage their accounts. Quickbooks is software that connects to bank accounts and allows you to tag spending and revenue in order to generate financial reports. In Ireland it seems that startups tend to use Xero – which is similar. Quickbooks is a big business that is owned by Intuit, which also owns TurboTax – the most popular tax software in the US – and also Mint (a money management tool).

Stripe is a company that allows you to accept credit card payments. It was founded by two Irish lads, John and Patrick Collison. Stripe is offering more and more products to businesses. For example, you can set up a business using Stripe Atlas. You can automate calculation of VAT (or sales taxes) using Stripe. Just a week or so ago Stripe announced that you can now generate accounting reports using Stripe Revenue Recognition. Stripe is including everything Quickbooks does, but their software is easier to use and they are quicker to launch new products. Most of Stripe’s customers started as startups, but those startups have since gotten big, so Stripe is growing with their customers and they are eating up Quickbooks.

All of Intuit (including Turbotax and Mint) is worth $150B today. Stripe is worth just under $100B. So, in some ways Stripe is already worth more than Quickbooks.

It is interesting that Stripe is not supporting cryptocurrencies. They did for a while but then stopped. I feel two ways about this:

  1. The Stripe founders are very smart. If they aren’t building solutions for accounting on blockchains/lists/crypto, maybe blockchains/crypto is less good that I think it could be.
  2. Maybe the biggest risk about Stripe is that they are wrong about crypto.

That’s it for this month! If you’d like, I can send you this newsletter each month. Just sign up with your email below:

Signing you up…
Kindly check your e-mail to confirm your subscription.

My First Virtual Art Gallery: Sagrada Barcelona

I’m just home from a trip to Barcelona and I’ve pulled together some digital photos and short videos into a virtual gallery.

You can view it for free at https://oncyber.io/barcelona .

Which piece is your favourite?


Weekend Links

  1. Clickbait is Unreasonably Effective. This video made me less cynical about clickbait. Top bloggers/youtubers spend half of their time just on the title and thumbnail. Youtuber “Mr. Beast” makes an appearance in this video. He’s just in his early twenties and doesn’t get bogged down in over optimisation: “Dude, just do what makes sense”.
  2. Degrowth is the philosophy is that we need to regress economically in order to solve environmental issues. From the blog: “At its core, I feel like degrowth’s appeal comes from its implicit promise to recast genteel North European decline as some sort of grandiose world-saving moral quest.”
  3. China is hammering the US on technology: The US only leads market share on electric vehicles and semi-conductors.
  4. Apple delays scanning iPhones for child porn. I was surprised Apple said they would make a way to scan content on iPhones – it just seemed to go completely against their approach of privacy for phone users and open up a pandoras box to scanning phones for many reasons. I’m not surprised they have pulled back on this.
  5. Consumer Lending in China. Uncollateralised lending is getting big in China – this means giving small personal and business loans without asking for deeds to a house or the title to a car in return. In many ways, uncollateralised loans were inconceivable until data analysis of transaction data on WeChat and Alipay. The Chinese government are wary, largely because of the disaster that was peer to peer lending some years ago.
  6. Two criticisms of the Gini coefficient. The Gini doesn’t do a a good job of distinguishing two specific issues within inequality, 1) a large group of people being poor, and 2) a small group of people being very rich. One can be true without the other being true.
  7. Steve Jobs Responding to an insult. Steve Jobs had his own vision for the future and did not fully follow customer feedback (which goes against some modern startup philosophies). He explains why he compromises on product features so that the product has a more consistent long term vision.
  8. The EU wants to only allow one phone charger. It doesn’t make sense to specify a single type of charger for electronic devices. Technology moves too fast. At the same time, it is a pain to have different adapters. I think Apple is making a mistake with all of their different chargers – yes, they can earn profits on chargers – but the customer experience (with even a Mac and an iPhone being different) is a bad one. So, I think the problem is real but this solution is bad.

Liked these links? You can sign up for longer form content in my monthly newsletter here:

Signing you up…
Kindly check your e-mail to confirm your subscription.


Where are the Smart People Going?

Tyler Cowen – on Ezra Klein’s podcast – claims that he was a big skeptic of crypto. However, he has recently become convinced because he is seeing a lot of smart people getting into it.

Ezra Klein then remarks that this question is a good heuristic for evaluating new fields.

I’m not so sure… It seems to depend on how you define “smart”. I think smart needs to mean quick thinking and unconventional.

Example 1: Civil engineering in 2006

When I started engineering at University College Dublin (UCD) in 2006, most of the class decided to do civil engineering. By the time 2010 came around, Ireland had suffered a massive crash in housing prices, so there were hardly any civil engineering jobs. Still today, in 2021, there is just a tiny minority of engineering students at UCD that are doing civil engineering. In fact, there are almost as many civil engineering lecturers as students!

In hindsight, the smart move in 2006 was to do computer engineering (and probably still is?). Really though, the unconventional thing would have been to code your own software at home and maybe skip university or leave it early – like John and Patrick Collison did, who eventually started Stripe.

Example 2: Doctors in Ireland

In Ireland there is a standardised national test at age 17/18 that determines which university courses you can get to. A large percentage of those that score the highest choose to study medicine. That is not because medicine is the field with the best potential future. It’s probably a combination of a) doctors having high social regard, b) tangibly helping people, and c) good pay, which is partly tied to the number of training places being limited.

Anyway, the young people going into medicine are certainly smart when measured by school grades, but their choice to do medicine has been there for decades and doesn’t say much interesting about the future of medicine as a field.

Financial Crisis 2008/9

Lots of top students went into banking and finance in the run up to the end of the 2008/9 crisis. A lot of them were probably top of their class and probably quick thinkers.

The problem with my examples

I think my examples are bad because the categories of people I described followed paths that are selecting for being smart, but not selecting for being sufficient unconventional.

For there to be an obvious signal about a new promising field, you need to know enough people that are both quick thinkers but also doing something abnormal. You also need to know enough of them to see a pattern, but not so many that they are just following a crowd.

So, figuring out the next emerging field is still very difficult, but if you can know some smart unconventional people, that will probably help.

Where are the quick-thinking unconventional people you know going?


August 2021 Newsletter: China is Far Ahead

In this month’s edition:

  • Payment Systems: China 1, US/Europe 0
  • Monopolies and Regulation: Which comes first?
  • Growth vs De-Growth and How to Save the Earth
  • Why are there so many primary school teachers on the Irish dating scene?

Payment Systems: China 1, Europe/US 0

I’m currently reading Cashless, by Rich Turrin, about payment systems in China. It has opened my eyes to how far behind the EU and the US are when it comes to using technology to improve access to and the quality of financial services.

In China over the last decade, government and Big Tech (Ali Baba/Alipay and Tencent/WeChat) brought their level of financial inclusion to 87% according to Ernst & Young. By comparison, that same financial inclusion index in the US is stuck at 46%. In short, while the US and Europe try to re-shape old, paper-based systems into digital form, China is creating clean digital solutions from scratch, untethered from legacy interests and embracing, rather than cautious, of technology.

Already, there is a huge gap between the mobile/technology based payment systems in China, and the legacy bank/card based systems in the US and Europe. In 2019, mobile payments in China were 500 times larger than in the US, even though Chinese GDP is only a third of the US! With China’s next generation of currency, the gap between the US and EU is about to get even larger. Two acronyms you’ll soon hear more of are CBDC (central bank digital currency) and DC/EP (digital currency, electronic payment). Inspired by elements of cryptocurrencies like Bitcoin, but centralised and with complete governmental control and identification requirements, CBDCs like the digital Yuan represent a digitally native implementation of currency, i.e. not one that builds software on legacy paper-based systems, but a digital-only payment system that is designed from scratch. The Chinese Central Bank is already testing these systems and they are faster, simpler, cheaper and provide much greater levels of information and security than what is used today.

While China rapidly improves its payment systems, US and EU governments seem stuck in two ways:

  1. The US, and the EU – to a slightly lesser degree – are tethered to regulated banking monopolies that are not natively digital. Governments struggle to overcome lobbying pressures slowing the next generation of technology.
  2. US and EU governments and companies are slowed by constant debate between preventing censorship and preventing illicit activity such as money laundering, child pornography (e.g. Apple stating they would scan phones, only later to delay the policy) and pornography (e.g. OnlyFans recently banned sexually explicit content due to pressure from payment processors, only to make a U-turn). The resultant approach is fractured and polarised rather than practical.

The EU and US are not going anywhere fast when it comes to digitisation.

There is then the third approach to payment systems of trans-national crypto platforms like Bitcoin and Ethereum. These embrace technology but do so with a libertarian streak that is off-putting to US and EU governments – particularly around issues of identification/privacy/anonymity. A significant (although probably falling) portion of the crypto community is dogmatic on issues such as anonymity, which goes against what most of the public find reasonable. Meanwhile, politicians resist engaging with crypto unless it is on the terms of existing financial regulation. So, the US and EU has two factions – traditional finance and crypto, whose interests and intents are probably largely aligned, but have managed to malign each other. It is very important that there are digital currency alternatives to the Chinese digital Yuan, which will be here in just a year or two, so I am frustrated by this fracture between most governments and crypto.

Ideally, I would like there to be a number of strong digital currencies that do have government approval. There are a lot of open technical questions and it seems that trial and error on different approaches is necessary – hence why we need more than one. Already, we are seeing certain countries (Singapore, Hong Kong, El Salvador) taking forward-looking approaches – that leave room for private innovation – on currencies. The European Central bank is progressing plans for a digital currency, but regulation for private entities remains burdensome and there seems to be little appetite, at least in Ireland, for engagement other than with existing regulations as a prerequisite.

Crypto in its current form needs to better prioritise designs for avoiding illicit activities. This would bring them more in line with what a majority of the public considers reasonable. Governments on the other hand need to better understand the technical improvements that digital currencies bring – for financial access, for efficiency, for robustness and for the minimisation of censorship. A positive approach is needed from countries and regulators to engaging with new digital currencies and approaches, including an openness to replacing old regulations.

*This comment is harsh on Robinhood, as there are positives to what that company has done. Robinhood has brought to light the importance of having an easy to use app/interface, over the old and complicated software and paperwork that banks tend to use. It is also healthy for the market that individuals take an interest in individual stocks over just buying passive index funds (although that may be negative for the median individual).

Monopolies and Regulation: Which comes first?

If you’ve lived in America, you probably have spent hours, if not days, of your life talking on the phone to Comcast – one of the phone/internet providers. In many parts of the US, Comcast is the only option for internet and, as with many monopolies, bad customer service is a feature. If you find a company’s service is bad, you might simply suppose a) they will go bankrupt, or b) they are a highly profitable monopoly.

My second example here is Amazon, which I understand has excellent customer service in the US [note: unfortunately not in Ireland] but terrible supplier service (as you can read from my own experience here). Amazon is not (yet) a monopoly from a customer standpoint, but they are often the only place (or the main place) their suppliers sell, which makes them a monopsony.

My third example is Interactive Brokers, a global company that allows you to buy and sell stocks. For Irish businesses, it is possibly one of the only options for buying and selling stocks, along with Davy Stockbrokers. It recently took me three months to set up an account for a company, and the customer service was terrible. Again, a monopoly with bad service.

Ok, that’s enough examples, so let’s move to the point. Which comes first – regulation or monopoly?

Typically, I would say the public thinks of the monopoly coming first and then the government stepping in to apply regulations thereafter. This is the narrative that comes through when we hear of Apple and Google and Standard Oil (to take a historical example) in the news. With technology based companies, this narrative is largely true. The monopoly comes first and then comes the clamour to regulate.

However, it also happens in the other direction (regulation leading to monopoly), in two different ways.

  1. In telecommunications or exploration, it is been common to auction off rights – giving certain companies an official monopoly to provide phone services or extract oil.
  2. In finance, regulations often develop over time that make it hard or impossible for new smaller companies to compete. My sense is that this is the case for Interactive Brokers (and Davy) for the services they offer to businesses. Specifically, neither DeGiro nor Revolut nor Wise (formerly TransferWise) offer stock brokerage solutions to business customers even though they do offer them to private customers. Having seen the amount of paperwork I had to fill out for Interactive Brokers, I suspect this bureaucracy is contributing to a monopoly-type market.

Nuance is important when it comes to monopolies. I take a positive view of monopolies that are not caused by regulation, and a more negative view of monopolies created by regulation. Interestingly, if you look at Apple, Google and Amazon, “monopolies” not created by regulation, customer service is good. In more regulated monopolies, this tends not to be the case.

Growth vs De-growth and How to Save the Earth

There are two opposing ways to look at how we might maintain a liveable environment: growth and de-growth.

“De-growth” is the idea that humanity must reject further economic growth and return to past ways of living if the world’s environment is to be saved (and anthropological climate change to be averted). Noah Smith provides a nice primer on de-growth in his recent freely available essay. Many criticisms of “consumerism” or “globalisation” implicitly contain elements of the de-growth narrative that we have more than we need, and a dialling back of the economy is the solution. A more specific articulation of de-growth is that we don’t need to grow or develop further, we need to redistribute. As Noah Smith points out, this would involve moving each person in the world to an annual income of $17,000 to even things out.

Growth is the idea that, with continuously improving technology, we can always get more for less. David Deutsch is a strong advocate for this approach in his book “The Beginning of Infinity”, which I summarized and reviewed here. Growth is the argument that a rising tide lifts all boats – against which there is the criticism that some have yachts and most have dinghys.

It’s anecdotal and may very well be my reading selection, but I sense a move this year and last away from de-growth and towards growth:

  • COVID vaccines. Although the long term success remains to be seen, the public perception is that this is a win for technology.
  • Lab grown meat. Despite my appreciation for some well reared Irish meat, and my reluctance to settle for less, I have to concede and even support that lab growth meat is quickly improving in quality.
  • Contactless payments. What has happened long ago in China is now happening in Ireland and the US and not just payments but also identity cards (digital COVID passport in Ireland, soon state IDs in the US) are moving to phones.

Why are there so many primary school teachers on the Irish dating scene?

I heard, from a friend, that the prevalence of primary school teachers on dating apps in Dublin is anecdotally about ten times that in Boston. I’ve been curious to understand why this is the case.

  1. Student teacher ratio – this is very roughly the same in Ireland as in the US, certainly not enough to explain a tenfold difference.
  2. Teachers don’t live in Boston? I don’t have data on this, but maybe more of them live in the suburbs and aren’t in my friend’s 50 kilometer radius?
  3. My friend might have an eye for primary school teachers – even though many don’t list it on their profile. Always a possibility.
  4. Primary school teachers are mostly single. I don’t have any data, but it sounds a bit harsh on primary school teachers.

I’m tending to lean towards explanation 2, although it’s not fully satisfying, so please email me with back with other theories so I can put this one to bed for my friend.

If you liked this newsletter, I can send it to you each month by email if you sign up below:

Signing you up…
Kindly check your e-mail to confirm your subscription.

Don’t blame your phone for being distracted: A Review of “Indistractable” by Nir Eyal


Reading this book served as a reminder to me to create more blocks of focused timed with a clear purpose. Even if the book hadn’t been good, it would have been worth it just for this. As it turned out, the book was good, and I give it 7 out of 10.

Here are my three takeaways:

  1. You can’t blame your mobile phone for distraction

Nir Eyal makes a somewhat contrarian point that mobile phones aren’t to blame for distraction. His view is that “you can’t call something a distraction unless you know what it’s distracting you from”. He feels we put too much emphasis on reducing our use of mobile phones instead of addressing what they are distracting us from (bad relationships, boring jobs etc.).

2. Skepticism that technology is making distraction worse

Eyal remarks that technology has always been blamed for distraction:

  • “In 1474, Venetian monk and scribe Filippo di Strata issued a polemic” stating “the printing press is a whore”.
  • In 1936, kids were said to “have developed the habit of dividing attention between the humdrum preparation of their school assignments and the compelling excitement of the [radio] loud-speaker”.

3. Constraints are needed for creativity

When we think of creativity, I think we often think of freedom. Eyal makes the point that we can do better with creativity when there are some constraints. Having some constraints (but not others) provides guidance, inspiration and focus for creativity.

A specific example of this is not just setting a time block for a task, but to think through – in advance – how you will approach the task. When I don’t do this, I get stuck figuring out wht to do during the time block.

How to cut back meetings

Eyal feels you should not be allowed to schedule a meeting if i) you don’t set an agenda, ii) you don’t propose a solution/approach in advance. I agree.


A Local Approach to Philanthropy

The reason I’m writing this blog is because one of my friends has been doing well on the job front lately and is deciding to give a portion of income to charity. They were asking around for thoughts on how best to think about a strategy for donating.

Back in the first half of 2020, I was involved with a small group of friends that started a hand sanitiser business where we decided to give profits to charity. I think there are many good approaches to giving and am writing this blog so you have one more to add to the compendium.

First, the limited amount I know about approaches to giving.

  1. Effective Altruism – The Trendy Approach

Kicked off by William MacAskill, this approach involves analysing which of kinds of giving offer the most impact per dollar donated.

Donation opportunities should be tractable but also underserved. For example, effective altruism might see donating to disaster relief as tractable (i.e. the aid can actually help) but not underserved because natural disasters often already get government funding as well as donations from the public. By contrast, 

run deworming programs would be seen as tractable (they are shown to help in countries with this issue) and underserved because not many other organisations see this as a priority.

Side-note: Another key tenet of the Effective Altruism movement is that it may be more efficient to work at a high paying job (e.g. banking) and donating a percentage of your money, than to work at a low paying non-profit job.

2. Causes I care about – The Classic Approach

This approach does what it says on the tin and somewhat describes the approach MacKenzie is taking with her donations of Amazon shares after divorcing from Jeff Bezos.

Causes I care about can be local or national or international or blended.

There is also the related “Causes I care about and benefit me politically or financially”, which guarantees a healthy debate. We are not going down that rabbit hole in this blog.

3. Fractal localism – The Meta Trendy Approach

This approach says that – in the same way you care more about your family than about your community, and more about your community than your country, and more about your country than other countries – you should care more about first donating locally before donating more broadly. It is more important to support your own daughter or your own neighbourhood, than someone else’s daughter or neighbourhood in a foreign country.

People generally have a better understanding of what is local – they feel the upsides and the downsides. Once donations go wider afield it is hard to appreciate the full impacts (positive and negative) of donations – and indeed to be aligned culturally, politically or morally.

With national or international donations, there is what I call the Irish passport problem.

The Irish passport problem is that if you let everyone with an Irish passport vote then you have a substantial portion of those voting who don’t live in Ireland (because Ireland had so many emigrants). The law – which I strongly agree with – is therefore that you can only vote in Irish elections if you are present in Ireland (with some exceptions). I wouldn’t want people outside of Ireland influencing what happens in Ireland. I see the same issue with donating further from one’s locality. Donations are political or moral or social judgements so one should be careful as one’s reach goes further afield. 

A Case Study on RESPOND Hand Sanitizer Donations

The approach we took with donations from this business was a hybrid of the above:

  1. Since there were a few core team members spread across the US, we decided that each member should be free to choose a charity, provided it supported low income families in the US through COVID. (Causes I Care About Approach).
  2. We focused on charities where the amount of our donation would be around 10-20% of their annual donations received. This was for two reasons: i) so the donation would be more than a drop in the bucket for them, and ii) the donation would not be so big that they would see a big slump in donations the following year. (Aspects of the Effective Altruism approach)
  3. Give unrestricted (Paul Graham essay here). The donations did not have any restrictions on how the organisations could use the funds.

On a technical note, we checked the charities were registered, had filed their last three tax returns, and were responsive when we reached out by email or phone.

Footnote: Being a mechanical engineer, I thought I would start this blog by coining a new word – philentropy – a hybrid of philanthropy and entropy. I decided not to because it made the blog look disorderly.


An awkward story about waiters

I’m think I’m getting old because I went to see a musical at the cinema last night and I enjoyed it. It was called “In The Heights” – written and directed by the same guy who did the musical called Hamilton.

In “In The Heights” there is a story of a young hispanic woman who is a student at Stanford. She is invited to a donor event and arrives in a black cocktail dress.

While at the event, one of the donors hands her an empty plate – assuming the young lady is a waitress rather than an attendee.

This situation is – at best – awkward. In societies where the job of waiter is associated with people of a certain appearance or accent, the additional aspect of race or class is introduced.

Taking this a level further, it is interesting to me why society finds this situation awkward or insulting. I think the answer is that society sees waiters as lowly, whereas society sees event attendees as more respectable.

So, it is insulting to give an empty plate to someone who is not a waiter, but, the fact society finds this insulting is a reflection on what we think of waiters. Ouch. Seems awkward.

My recommendation on all of this is simply to be respectful and friendly to waiters whoever they are (without being over the top). I have not worked as a waiter, but there must be a lot of annoying people to deal with.

Less clear is my recommendation on how – at an event – to tell if someone is a waiter. One approach would be to ask a random person if they know where to put empty plates. This is more innocuous than forcing a plate into someone’s hands, but may still risk causing insult.

Another approach is to try and bring the empty plate back oneself. My implementation of this would be to wander around searching for a table and looking like such a tool/fool that eventually I’m spotted by a waiter who politely asks if they can help.


Reading the Science of Human-Caused Climate Change: Part 1, Data


  • I realised my understanding of human effects on climate change was stuck at what I learned in school at age 12.
  • I reviewed climate data on tornados, droughts, temperature, CO2 levels and sea levels.

My 12 year-old understanding of climate change

Back in geography class with Mr de Paor, I learned that carbon dioxide is a greenhouse gas, and greenhouse gases cause warming of the earth. At age 32, my understanding had not evolved further than that. This is despite the fact that reducing carbon dioxide emissions is a key motivation for the work I did at Sandymount (now part of Alfa Laval) to reduce the cost, energy and emissions of beer transport with concentrate-based supply chains.

I’ve done quite a bit of statistics and modelling of systems at UCD, MIT, CERN and Sandymount. So, while not an expert, I have some ability to assess data such as that of the world’s climate. I just never took the time to do it.

Governments and companies are directing a lot of resources towards CO2 emission reduction, so I thought it was time to explore the data first hand for myself. Here’s what I learned in two parts, the first focused on measurement data, and the second focused on explanations (modelling).

Are tornados and droughts going up over time?

While we regularly read or watch news about extreme droughts or hurricanes, single extreme events – or even multiple extreme events – don’t prove that climate change is happening (much less whether that climate change is human caused). To establish a pattern of anomalous climate change, one needs compare the prevalence of extreme weather events over an extended period of time.

One key challenge in measuring extreme events over time is record keeping. Today, in 2021, record keeping is very good, but that wasn’t always the case. This means that it is common to see an increase in reported extreme weather events purely because our record keeping has gotten better over time.

One way to address this issue is to focus on the large events – which are more likely to have been recorded in earlier days (because they are big and obvious). The following two graphs show larger tornados (category EF1 and larger), as well as really large tornados (category EF3 up to EF5).

Tornado count by year of large (top) and very large (bottom) tornados in the contiguous 48 states. Taken from NOAA data available at http://www.ncdc.noaa.gov/stormevents and presented in Unsettled by Steve Koonin.

The occurrence of large tornados in the US has remained largely similar, if not decreasing slightly, over the past seventy years.

Moving on to drought – typically measuring using the Palmer index which combines rainfall and temperature data – here is a plot of that index over time:

Taken from Unsettled by Steve Koonin. Data origin is https://www.epa.gov/climate-indicators/climate-change-indicators-drought

There isn’t a clear pattern indicating an increased prevalence of droughts over the last fifty years.

This data above is for the 48 contiguous states. I haven’t looked at global data, but I’ll note that the IPCC 5 WG1 report indicates low confidence in there being any increase in extreme cyclones or in drought on a global basis since 1950.

Is the rise in sea level since 1950 unusual?

To understand the context for sea levels, we can look at data on historical levels over a very long period of time. You have to read the following chart from left to right, starting with today on the left and moving back hundreds of thousands of years on the right. (“Kyrs BP” means thousands of years before present):

Figure above copied from Bianchi et al. 2011, http://dueproject.org/en/wp-content/uploads/2019/01/8.pdf OR https://www.researchgate.net/publication/242397422_Mediterranean_Sea_biodiversity_between_the_legacy_from_the_past_and_a_future_of_change

In the grand scheme of things, the world in 2021 is towards the top of a sea-level cycle (we’re also towards the bottom of an ice cycle). Historically, the level of the sea has varied by about 120,000 millimeters (120 metres) over the last half million years. Very roughly, the chart also shows periods of increase and decrease in level by roughly 10 millimeters per year (100,000 mm over 10,000 years).

Now, let’s turn to more recent trends in sea level. In the fifth IPCC (Intergovernmental Panel on Climate Change) report, the first working group has a figure showing the rate of change in sea level over the last century:

From IPCC Working Group 1 Fifth Annual Report. https://www.ipcc.ch/site/assets/uploads/2018/02/WG1AR5_all_final.pdf

The above figure shows the rate of change in sea level in millimeters per year since 1900. It’s clear that the sea level was already rising before heavy industrialisation caused carbon dioxide emissions to start in earnest (you can find that graph below). Whether there is a recent (i.e. since 1950s) acceleration in sea level rise is not very clear. Sea levels are definitely rising, but they have been for a long time, including in the early 1900s.

How much have carbon dioxide levels increased since 1950?

Here is a plot of historic C02 concentration in the atmosphere. Data from 1959 onwards is directly measured from the atmosphere whereas earlier data comes from ice cores. Sorry for the small font on the graph. If you find the graph hard to read, it shows CO2 steady around about 280 ppm from 1000 AD until 1850 AD, and then starting to rise around 1850. There is an increase of about 10 ppm by 1900, then an increase of about 40 ppm by 1950 and then an increase of 80 ppm by 2020.

CO2 data from https://www.co2levels.org/#sources . Note that the y axis starts at 260 so the increase is more dramatic than it looks (although the increase is still dramatic).

The level of confidence is high that the increased levels of CO2 are human-caused because i) the high levels of CO2 in the last 50 years are anomalous – with high confidence – comparing them with the previous decades in recent history (what the above graph shows), and ii) the rough mass balance between fossil fuels burned and CO2 increase in the atmosphere seems to roughly check out (I haven’t checked this myself but would like to).

How much has the global average temperature increased since 1950?

Recent information on temperatures – of both the ocean and atmosphere – comes from direct thermometer measurements (although other measurements like satellite data are now becoming more common).

In the top subgraph below you can see the temperature anomaly of land air (LAT) in red, seawater surface (SST) in blue and HadCRUT (a combination of air and seawater data taken from a separate dataset of measurements) in black. The “temperature anomaly” (on the y axis) is the difference between a temperature measurement and the average temperature over some reference period (typically a short pre-industrial period).

[Side note: The bottom subgraph just shows the difference between the red and black and the blue and black lines in the top subgraph (because the paper is comparing different models and interested in the degree to which the models agree). Don’t worry too much about the bottom subgraph – it is there because the paper is discussing how the red and blue models compare to HadCRUT, which is a commonly used model that – as I understand – combines air and water temperatures.]

Temperature anomalies take from Rohde et al, 2020 (https://essd.copernicus.org/articles/12/3469/2020/)

Looking at the top subgraph above, it is clear that temperatures have increased since 1850. There is some increase from 1920 – 1940 (a period when carbon dioxide levels had not yet risen much), some flatness from 1940 – 1960 (when carbon emissions became significant relative to natural levels) and more of an increase, particularly in air temperature anomalies, after 1960 (when carbon dioxide levels increased significantly).

Satellite temperature data is also now available for more recent decades. One benefit of satellite data is that it does not suffer as much as thermometer based measurements where there can be local warming effects associated with cities, known as urban heat island effects. I need to do more cross referencing before I put up data here, but I think that cross checking earth based measurements with satellite measurements is a sensible thing to do. If you have a peer reviewed reference cross validating thermometer and satellite data I would be interested.

Are recent global temperature increases anomalous compared to historical data?

On page 491 of the fifth IPCC report: The Physical Science Basis, you can see various temperatures over the last two thousand years. If you read the small text in the caption of the diagram you can see what temperature each line corresponds to (broadly speaking, the coloured lines cover different lattitude and land/sea datasets).

Taken from IPCC Fifth Annual Report: The Physical Science Basis

Here is the comment from the IPCC in the report on the above chart:

“Considering these caveats [fewer and less reliable proxy records for earlier centuries], there is medium confidence that the last 30 years were likely the warmest 30-year period of the last 1400 years.”

I can see how drawing conclusions about a trend in the last 30 years of temperatures is difficult because a) historical temperatures show a lot of scatter and b) the graph (see the caption above) uses 50 year smoothing of the data.

What does it mean for global average temperature to increase?

Global average temperature is an abstract concept and doesn’t reflect the temperature at any single place at any single time. Rather – as the term implies – it aggregates temperatures across the globe and across time (typically one year).

When you have an increasing average temperature, that can mean a lot of things. For example:

  1. there may be one or two places that are dramatically warmer than before while everywhere else stayed the same, or
  2. there might be many places that have become slightly less cold while everywhere else has remained the same, or
  3. Some places became much hotter, while fewer places became much colder, so the net effect is a warming.

It is therefore worth breaking down a bit more the temperature patterns that result in an increasing average. I’ll do this with the example of North America.

The next figure shows the trend in reported maximum high and minimum low temperatures over time in the US:

Taken from Unsettled. by Steve Koonan, 2021 – with an original data source of https://www.ncdc.noaa.gov/data-access/land-based-station-data/land-based-datasets/us-historical-climatology-network-ushcn

The number of record high temperatures in the above charge is fairly steady, but the number of record low temperatures is falling. In short, the really cold days got milder.

A non-summary

I’m not going to summarise because I’d like to encourage you to look at the graphs for yourself and go and find other graphs and data. I think that more people looking at and working to understand the data is a good thing, especially when it’s an important topic.

I plan to write a follow on blog (Part 2) where I’ll talk about different explanatory models and hypotheses for what we are seeing in climate data.

In the meantime, please do write to me if there are further graphs or data that you find interesting and/or important or see errors in what I have posted so far. I’ll plan to make some updates to this blog as I get feedback.


Simple Weight Lifting to Build Strength

My inspirations for starting to lift weights

I played hurling and Gaelic football competitively until about 18, but I never did any serious weight training. I started thinking about weights for two reasons: 1) my girlfriend at the time was able to lift a savage amount of weight – very good technique and very high weight lifted to body weight, and 2) I had been reading books by Nassim Taleb, such as Antifragile, where he talks about how weightlifting – specifically, doing small amounts of heavy lifts – is particularly good for strength while avoiding the wear and tear of repetitive exercises like running.

Not the original inspiration, but I have been reading Balajis Srinivasan lately (Balajis.com) and his approach to life is health, truth (as in understanding of the world), and then wealth – in that order. I think that’s one reasonable approach and v

A simple approach I use for weightlifting

I have a few rules I followed for weightlifting:

  1. Three lifts. I mostly just do the three main lifts:
    • Benchpress (lying on my back using my arms to push up a barbell with weights),
    • Squat (barbell across the back of your shoulders and then squating down), and
    • Deadlift (lifting a barbell with weights from the ground up to a standing position with arms down over the thighs).
    • The above three cover a lot of muscles. I do mix in some rows, overhead presses and pull-ups to hit some antagonistic/opposite direction muscles.
  2. Empty bar first. I always do a first set of lifts with no weight on the bar. This is to check I have the safety bars on the weight rack in the right position to catch the barbell if I fail on a lift.
  3. Three sets of three repetitions. I do each lift with full weight (the most I can lift) three times, take a break for a minute or two, and then repeat that process two more times to give three sets of three lifts. This is considered high weight, low reps and is considered good for strength but not the optimal approach for looking ripped like Arnold Schwarzenegger.
  4. I don’t lift to failure. A lot of weight lifters will increase weight or number of lifts until they fail with a lift and the weight is caught by the safety bars. This might be optimal for getting strong fastest, but I think it comes with some risk of injury. Instead, I lift a weight close to my limit, but if I find a lift very hard, I’ll stop and either reduce the weight or move on to the next lift.

I’m not qualified or anything in weight lifting and the weights are heavy and I can see from doing it that – used the wrong way – weights can cause a bad injury. When I joined a gym there was a free lesson, so I used that to go through the lifting technical I do with someone trained.

Progress over eighteen months

In the graph below you can see my progress on weight lifting over 18 months on the three main lifts. On the y-axis you can see how much I’m lifting relative to body weight – which I think is not a bad way to think about things. My goal is 1X on bench, 1.5X on squat and 2X on deadlift. These are somewhat respectable targets for someone wanting to be strong, but certainly far from competition level.

In terms of the graph below, the x-axis isn’t evenly distributed in time, so gaps of time where I took breaks (due to COVID) aren’t obvious unless you read the dates. For me, the main notes are:

  1. My benchpress was already decent and I started near to my 1X bodyweight goal.
  2. My squat was the weakest and took some work to bring from just above 1X up to 1.5X.
  3. My deadlift also wasn’t epic and took about seven months to get from about 1.4X bodyweight up to 2X bodyweight.

A few reasons I still like doing weight lifting like this

  1. It’s quick and less boring than running on a machine – doing high weight and low repetitions means a good session in less than half an hour.
  2. My back feels stronger – hard to tell for sure, but I think my posture is better and I don’t remember having any back trouble over the last year and a half. I’m short so that probably helps, but I think I’ve fewer issues now since lifting weights.
  3. Concentration – it takes a lot of mental focus to lift heavy weights. That takes my mind away from other stuff that I would be thinking about if I was just jogging. I think that focus is probably good.


Side Notes:

Sidenote 1: I once met the Kildare footballer Dermot Earley (junior) at the Stand House hotel gym in Kildare. It was probably around 2007 and he was towards the end of his career (The Stand House has closed since.). I wasn’t sure what to say to him but we had a quick chat and I remember him saying the following about weights: “Remember to do some work on the legs”. I don’t think it was specifically a comment on my legs.

Sidenote 2: I’m not aware of data supporting or contradicting the claim of wear and tear with with highly repetitive activities. One injury that does come to mind is elbow injuries in baseball – often fixed with Tommy John surgery. Generally, my rough sense would that nearly any form of exercise is probably good.


Centralisation through the lens of Mobile Phones, COVID, Cyberhacking and Ethereum


  • Information systems today are largely centralised – with Amazon, Facebook and nation state governments as examples.
  • Centralised systems are more efficient than decentralised systems, although they are less robust and more vulnerable to security attacks.
  • Mobile phones are an example of how technology has decentralised computing power.
  • Ethereum is a technology that is attempting to decentralise information.
  • Ethereum is highly inefficient today. The question is whether technology can take it to a point where efficiency is high enough to be irrelevant.

Mobile Phones

When computers first emerged, it wasn’t clear we would all, someday, have one in our pocket (and soon attached to our brain?).

The reason we all have mobile phones is not because mobile phones have become just as powerful as larger computers – it is still more cost effective, on a per calculation per second basis, to build a large computer. The reason is simply that mobile phone computers are fast enough that computing power is now just one among many attributes to consider in choosing a phone.

Restated another way: In the early days, only big computers made sense. Over time, technology improved so much that it is worth us each having a personal computer, even if operating a larger computer would be more efficient. Technology has allowed us to decentralise computing power

Technology may allow us to (re-) decentralise production of food, clothing and medicine

Extending this framework, consider now the farming of vegetables, the raising of livestock, the production of vaccines and the manufacturing of clothes. Today, all of these seem best outsourced to large centralised companies. However, we should ask the question of whether technology can get to a point where these become processes that are done in our homes. To be clear, I don’t mean a reversion to classical small-scale farming), but rather to high-tech versions such as home chemistry/pharma kits and home cultivated “lab” meat.

COVID driving robustness in systems instead of efficiency

With COVID we realised that many of our supply chains, particularly around urgent care, are concentrated in certain parts of the world. Whereas this has allowed efficiency, it has not allowed for robustness when travel stops. COVID is therefore a driver of decentralising processes, e.g. building vaccine production in each country (and maybe some day, in each home).

2019 was a narrative of efficiency through specialisation through centralisation. 2020 brought more of a narrative of generalisation through decentralisation.

Cyber Hacking as a Driver of Decentralisation

We have lately heard of cyber attacks on pipelines (US) and healthcare systems (Ireland). Systems containing sensitive information are always vulnerable. If information is valuable but not sensitive, then you can make backup copies. If information is valuable and sensitive, the best you can do is to defend against attacks. The only alternative is not to collect and store that information centrally in the first place, but this hurts efficiency!

For a company operating in healthcare, it is currently highly inefficient not to have the patient’s information on hand. The same goes for Facebook and Google – neither company would be valuable if they were to give up on their centralised storage of information. Today, it is simply inefficient for a government or company not to store lots of information centrally.

But what if it were possible to efficiently store and access information in a decentralised way?

We think decentralisation. is too inefficient to be possible

Today, we don’t believe that information can be decentralised because we think this kind of system is too inefficient. We believe monetary and payment and government systems will always have an advantage to systems that are decentralised.

These statements are probably correct, centralised systems will always be more efficient. However, what is missing here is that decentralised systems may get to a point where their relative lack of efficiency becomes unimportant, allowing other attributes like robustness and security to prevail.

Large computers are still better than small computers on the basis of calculation efficiency. However, small computers are now good enough that this doesn’t matter. Likewise, decentralised blockchains right now are inferior to centralised systems. The question is, with technology, will they improve to a point where their disadvantage no longer matters.

Ethereum as a Test Bed for Decentralisation

Far beneath the hype and price swings in crypto there is an effort to answer the deep question of whether technology can make possible the decentralisation of information. Ethereum is one embodiment of this effort.

Roughly speaking, there two challenges that Ethereum faces with respect to decentralisation – 1. the energy problem, and 2. the speed problem.

  1. The transaction validation problem (aka the energy problem)

Contrary to popular news, the problem of energy consumption in Ethereum mining (and Bitcoin) is the less interesting one. At its core, energy consumption is an answer to the question of who gets to say which transactions on Ethereum are valid and which are not (since there is no central bank or private company to make that call). In very rough language, Bitcoin first solved the transaction validation question by using a maths problem to dictate who deems a batch of transactions to be valid. This is called Proof of Work and is the same mechanism that is used on the Ethereum blockchain as of May 2021. Solving random maths problems takes computing energy, and this is why cryptos today are energy intensive.

Since then, another approach to transaction validation has emerged called Proof of Stake. With Proof of Stake, the validation of each batch of transactions is randomly assigned someone holding the Ethereum currency (called Ether). This approach – and perhaps there will be even better approaches still – does not require solving random maths problems, and so the energy requirements are a tiny fraction of the main blockchains today.

Though not yet implemented on Ethereum (it has been implemented on other blockchains like Celo.org), there are known solutions to the energy problem.

2. The decentralisation problem with Ethereum (aka the speed problem)

The more interesting problem in blockchain is how to keep the system both decentralised and quick to approve transactions.

A simple way to think about whether the system is decentralised is whether it is practical for owners of standard computers to participate in validating batches of transactions. Bitcoin took a solid step in this direction, but you still need to have specialised computers in order to help validate Bitcoin transactions. Ethereum 2.0 – coming soon – will allow normal computers to be used to support transaction validation. Indeed there are over 150,000 computers already set up and ready to host the new version of Ethereum when it emerges over the next months.

So, the core challenge for Ethereum is that the software has to be able to run on a standard computer. This means that the whole network is somewhat limited by what a personal computer can do. If you think this sounds a bit stupid then you are right – it is like trying to do mobile phones in 1970. Slow.

Right now, as of May 2021, so many people are using the Ethereum network, that the cost of doing transactions can be over $100, which is insane. Ethereum simply cannot keep up.

This brings us to two broad solutions to the speed problem:

i. Use bigger, less centralised, computers

If you use fancier computers (like big Amazon datacenters), you can get great speed. By just having a few computers do validation – rather than hundreds of thousands – information can propagate much faster. However, this is at the cost of decentralisation (people can no longer support the network with their personal laptops).

Networks like Binance (with heavy centralisation) and Celo.org (with more centralisation than Ethereum but much less than Binance), have taken this approach to solving the speed problem. They require more specialised hardware to support transaction validation, so the networks are less centralised but they are faster.

ii. Use better technology (software)

Ethereum is taking a different course and sticking to the goal of being supported by personal computers. To do this, Ethereum is relying on mathematical tricks allowing more information to be packed into the same amount of computer storage. There are various solutions here and you will find words such as “zksnarks”, “optimism” and “sharding”, all of which are ways to pack more information into less storage via mathematical tricks. As I see it, these mathematical tricks are real knowledge progress, in the same way that Einsteins General Relativity is real knowledge progress – and I say that with only some exaggeration.

Decentralisation beyond Efficiency

I’m reluctant to say that technology necessarily reduces centralisation. True, we now all have mobile phones – so computing power is more decentralised – but much of the information is stored centrally with Amazon and Google. More accurately, technology changes what is centralised. At first, technology centralises things (like garment making on factory looms) or the use of supercomputers in the 1970s, but maybe later technology improves so much that “everyone can have one” type decentralisation occurs.

In 2021, the centralisation of information is driven by a need for efficiency in information exchange. Decentralised systems for information are currently inefficient, just as small computers were in the 1970s. That may change. Just as computers went from factories to phones, technology may take us to a point where factors other than efficiency – such as security and robustness – result in the decentralisation of information. Ethereum is an early test case in this.

Footnotes: A Summary of What Drives Centralisation

Summarising, in no particular order, we have desires for:

  • Efficiency – driving centralisation. (e.g. storing customer information on one database)
  • Robustness – driving decentralisation. (e.g. every country having vaccine production facilities)
  • Security – driving decentralisation (e.g. customer information being held only on their computers, not a central company or government network)
  • Technology – first driving centralisation (via economies of scale or network effects) but in some cases (such as computation) reaching an inflection point driving decentralisation again (e.g. mobile phones).


  1. A great piece from Vitalik Buterin on the the tradeoffs between efficiency and decentralisation on blockchains.

Five Takeaways From The Berkshire Annual General Meeting 2021


  • Buybacks accelerate as Buffett continues to view the stock as cheap.
  • Shareholder votes will test Berkshire’s operational model after Buffett.
  • Berkshire’s executive compensation is unique and best for lower growth companies.
  • Berkshire is being eaten by index funds.
  • I’m still sleeping well at night owning Berkshire.

Stock buybacks accelerate as Buffett continues to view Berkshire stock as cheap

In 2020, Berkshire repurchased almost $25B worth of its own shares. The report on the first quarter of 2021 indicates two trends:

1. Buybacks continue to accelerate, with about $6.5B worth of stock repurchased in the quarter.

2. The average repurchase price is increasing. Shares were purchased at an average price of $251 per B share in March 2021, up from an average price of $225 in December 2020.

This indicates that Buffett and Munger view their stock as undervalued – even at a price of $251 per B share.

Shareholder votes as a harbinger of market pressures after Buffett

Two shareholders proposals were brought to a vote this year: one requesting Berkshire to disclose climate change risks to its business, the other centered on Berkshire disclosing its approach and metrics on diversity and inclusion.

Such proposals are not unique in the history of Berkshire. For example, in 2011 there was a shareholder proposal for Berkshire to report on how it would meet new EPA guidelines around greenhouse gas emissions. In 2021, and in years past, Buffett and the Berkshire board recommended against supporting these proposals and shareholders followed their recommendations. All proposals were voted down.

Buffett’s position on such proposals is that Berkshire is a collection of companies that operate independently in different countries and different businesses with different social norms. For Berkshire’s central management – a team of a few dozen employees in Omaha – to impose a one-size-fits-all approach for these independently operated businesses with over 300,000 employees would would reduce rather than improve the ability of the business to address questions of diversity and inclusion at a local level.

Berkshire’s approach to allowing its businesses operate independently is unique among corporations. Indeed, it is a major part of Buffett’s pitch in buying privately owned businesses when he can credibly say that – unlike many private equity firms – he won’t be radically cutting staffing, changing management or saddling the company with debt.

There are other philosophical differences on these proxy votes that warrant a brief mention:

On climate change

Buffett’s position is that climate change is a threat to the environment and to Berkshire’s business, hence the shift of Berkshire Hathaway Energy out of fossil fuels and towards renewables. However, Buffett believes that fossil fuels will be important to the economy for the foreseeable future, and that companies like Chevron – who he recently invested in – serve a positive social purpose.

On diversity and inclusion

One view is that increasing the measurement and reporting of inclusion and diversity metrics is an effective means for increasing inclusion/diversity. Reading between the lines, Berkshire’s view is that, where you introduce global metrics, you introduce incentives that can lead to problems as well as solutions – particularly in dealing with issues of inclusion that can be difficult to quantify and address. Berkshire’s preferred approach is that metrics and approaches are therefore best developed by individual businesses at a more local level.

Berkshire stands alone in how it pays executives

There has been criticism from pension funds and index funds of Berkshire’s abnormal executive remuneration policy. As with the other shareholder proposals, I see Berkshire’s unique approach as a good rather than a bad thing. Here is the exhibit of executive pay. Pay particular attention to the pay of Greg Abel and Ajit Jain who, for all intensive purposes, are the operating CEOs of Berkshire:

Notice that, of the $19M in annual compensation each receive, the vast majority is in the form of a guaranteed salary. $3M is provided as a bonus and there are no stock options. Institutional Shareholder Services criticises this as a lack of incentive based compensation.

However, this lack of incentive based compensation can be a feature rather than a bug. Contrast Berkshire with a typical public company where CEOs are remunerated with large blocks of stock options that they cash in on when the market rises. This can make sense for high growth companies where the CEO has a big influence on company growth and stock price. However, for lower growth companies, where stock price depends much more on the overall market, stock option compensation makes a lot less sense (even though it sounds good when you call it “incentive based”).

Further, paying a high base salary is more transparent for investors examining financial statements. While a salary appears clearly as an expense in the profit and loss statement, the treatment of stock options is a lot harder to understand because the cost to the company is one of dilution of shareholders rather than a clear cash expense. This can allow management to hide the true costs of compensation in stock option arrangements.

For a relatively low growth company like Berkshire, I’m happier for Berkshire to pay a high salary than paying incentive stock options.

Index Funds are eating Berkshire Hathaway

Over the past decade, Warren Buffett has espoused the virtues of investing in index funds like the S&P500 – even to the point of recommending the S&P500 over buying Berkshire Hathaway stock. Buffett is one of Jack Bogle’s biggest fans (the founder of Vanguard Index funds).

Buffett has recommended that, upon his death, the ~0.3% of his wealth that will accrue to his wife (the remainder going to charity) be converted by her from Berkshire stock into shares in an S&P500 index fund. This is not exactly a vote of confidence in Berkshire stock, although I would temper this by adding that Charlie Munger – Buffett’s side kick – is clear in saying that he believes Berkshire will outperform the overall market over the long term.

I’m unsure if this is ironic or not, but when Buffett passes away, Berkshire will increasingly come under control of the index funds he espouses – as index funds take up an increasing percentage of the overall market. You can see here lists of the top shareholders in Berkshire today – taken from marketscreener.com :

As I have previously written, I see the rise of index funds (e.g. Vanguard, Blackrock and Statestreet (SSgA) together owning almost 20% of Berkshire) as a net negative because it encourages retail investors to own a little bit of everything they know nothing about rather than own a few stocks they know something about. By buying into index funds, we have dedicated responsibility for governing our corporations to yet another layer of corporations like Blackrock, State Street and Vanguard – rather than directly owning and voting our shares ourselves.

Final Thoughts on Berkshire’s 2021 AGM

I continue to believe that Berkshire is a wisely and conservatively managed collection of high quality businesses. Further, Buffett points out that he is seeing price inflation in the economy, and I see ownership of Berkshire as a way to hedge out that inflation risk to some degree because they own contracts (via their energy and railroad and insurance businesses) where prices can adjust to inflation.

That said, I am concerned about Berkshire post-Buffett. Greg Abel as future CEO stands a chance of preserving the Berkshire culture of operating its businesses independently. However, it will become increasingly difficult without Buffet’s large vote, and, with the rise of index investing to resist short term pressures index investment groups. For now, I will continue to hold 30% of my liquid net worth in Berkshire and sleep well.


Why I’ve Bought in to Ethereum

  • I have now reduced my cash holdings from 30% to 25% in order to hold 5% Ether.
  • I’m investing in Ethereum because I like what I read from Vitalik Buterin.
  • It is possible I will lose the full investment because crypto has an uphill battle against traditional currencies that work quite well.
  • I see that knowledge about digital security and transactions is growing – my bet is that this will prove valuable and that value will accrue to the Ether token.


Until now, I have held my investments as follows (link here to my explanation why):

  • 30% Berkshire Hathaway
  • 30% Store Capital (a real estate investment trust)
  • 30% cash
  • 5% gold index fund
  • 5% Bitcoin

I have now reduced my cash holdings to 25% in order to hold 5% Ether. This brings me to 10% holdings in crypto across Bitcoin and Ethereum (plus a much smaller amount of some altcoins).

Note: To avoid spamming everyone on this channel with crypto, I have moved my crypto writing over to a pseudonym at Pinotio.com .

What is Ethereum and what is Ether?

Ether is a cryptocurrency, much like Bitcoin. Ethereum is the software platform on which it Ether is transacted.

The purpose of Ethereum is a lot broader than the Bitcoin network. In addition to allowing transactions involving Ether, Ethereum (the software platform) allows you to write lines of code that are executed and recorded publicly.

Many of the altcoins (the lesser known cryptocurrencies) are built and hosted on the Ethereum platform. For example, there is an automated cryptocurrency exchange called Uniswap.org that is built and runs on the Ethereum platform. Uniswap allows you to exchange one cryptocurrency for another at the prevailing market rate – without the need of a bank or a trusted third party. Everything is automated via lines of code.

To execute lines of code on Ethereum, you must pay the network. Those payments must be made in Ether, which is the native currency of Ethereum.

Why invest in Ether?

Ethereum was invented by Vitalik Buterin a young Russian-Canadian polymath. Reading his website – Vitalik.ca – it is clear that Vitalik is highly knowledgable and thoughtful.

While there are owners of Ether, there are no owners of the Ethereum platform, which is open source. The influence that Vitalik bears on the Ethereum network is not a formal one, but achieved through the what he shares through his writing online, his contributions to code that, and The Ethereum Foundation – a non-profit to promote the growth of the Ethereum network.

I’m investing now in Ethereum because:

  1. I like Vitalik Buterin and Ethereum is an imperfect embodiment of Vitalik’s ideas. Since Vitalik is young, I believe he has many decades left during which he will continue to contribute in ways that improve upon the Ethereum platform.
  2. Ether is the second largest cryptocurrency by market capitalization, after Bitcoin.
  3. I have read about other technology-forward thinkers holding a combination of Bitcoin and Ether – notably Balaji Srinivasan and Mark Cuban.

Pitfalls with Ethereum

Fiat Currencies Remain Strong. As with any crypto, I expect it is possible that I will lose all of my investment. As Tyler Cowen recent pointed out on Lex Fridman, crypto has a lot of work to do in order to displace fiat currencies (US dollar, Euro etc.). The currencies we have today work very well and getting consumers to adopt something new is an uphill battle. My view is that the use cases for crypto remain weak. At best, it seems that we may be able to reduce remittance and currency exchange fees. However, I believe the technical progress being made – by people like Vitalik – on improved encryption and data storage methods is impressive. You can read Vitalik’s writing on zk-snarks or you can read the Bitcoin Whitepaper for examples of this progress. Einstein discovered beautiful equations, but their full utility only became apparent with time. That is the hope – not the guarantee – of crypto.

The Ethereum Network is Overloaded. Right now, the ethereum network is not able to cope with the number of lines of code being executed. The cost of doing a transaction can fluctuate to above $100. There are technical solutions to this but they are taking time to roll out. For the technically interested, a big improvement will come as the network moves from proof of work (like Bitcoin) to proof of stake – which is a much lower energy (and hopefully cost) way of implementing transactions. Alternatives to Ethereum are emerging that provide improved speed. However, Ethereum remains far more widely used and, like Facebook does, Ethereum is likely to learn from the best ideas in other software platforms (like Facebook adopted Snapchat’s Stories feature).

In conclusion

It is clear to me that knowledge is growing rapidly in the area of decentralised transactions and information exchange. I see investments in Bitcoin and Ethereum as ways to gain exposure to this trend, although I may be completely wrong that this knowledge proves valuable over time, and/or wrong that the value accrues to the Bitcoin and Ether tokens.


S&P 500 returns since 1971 rely on dividends to beat gold


  • The US stock market index has not progressed since 1971 if you denominate the index in ounces of gold rather than dollars.
  • This is true but neglects the role of dividends in driving stock market returns.
  • Including dividends, the total stock market return is close to 4X that of gold.

S&P500 Index Denominated in Gold:

Anthony Pompliano makes the claim on Lex Fridman’s recent podcast (53 mins in) that the US stock market is down since 1971 if you denominate it in gold rather than in US dollars. This is true if you look at the value of the S&P 500 index, here denominated in ounces of gold:

Figure 1

Notice how – when denominated in ounces of gold – the S&P index is down significantly compared to the dot com bubble in 2000 and about equal to its level back in 1971.

Anthony Pompliano’s main point is that it’s not so much that stock prices have increased but that the US dollar has been devalued. This is true, but omits the important role of dividends in driving stock returns.

S&P500 Index Denominated in Gold – including dividends:

Here is the chart when dividends are included to give an S&P500 total return index, still denominated in ounces of gold:

Figure 2

This chart of total return includes continuous reinvestment of dividends earned back into the S&P index. Including dividends, the S&P500 now shows a return of 4X since 1971 when denominated in gold (compared to about 1X without dividends). However, the index – even including dividends – is down compared to 2000 when denominated in terms of gold.

In Conclusion:

Much of the appreciation in the stock market reflects a loss in value of the US dollar. Anthony Pompliano makes a good point this regard. However, when dividends are included, stock markets have outperformed gold by 4X since 1971.

A further technical note:

In practise there are taxes on dividends unless the stock index is held in a tax free retirement account. This means that – outside of retirement accounts – the return on the stock market since 1971 is in fact between 1X and 4X.

Data Sources:

Monthly data for the S&P index and dividends is taken from Robert Shiller.

Monthly data for gold is taken from Macrotrends.net



The Beginning of Infinity: My First Reading


  • Justified Belief vs Fallibilism
  • The scientific method is not just about being testable
  • Edison’s 99% perspiration was not mindless
  • Creationism will become an unremarkable attribute of a god
  • Genes and Memes – but not what you’re thinking about
  • The drawbacks of proportional representation
  • That Easter Islanders survived for so long is a tragedy

Why did I read this book?

Naval Ravikant – a technologist and preacher of wealth (preachings here!) – recommends this book as one of the most important to read. Naval’s approach towards reading is that he would prefer to read one hundred books many times each than read thousands of books just one time each. I think this logic makes sense. The best books/films bring about new learning/entertainment each time you read them.

What was it like for me reading the book?

Many of the chapters – such as one covering the nature of infinity, and another covering quantum entanglement – were impossible for me to understand.

David Deutsch’s primary premise for the book is that everything in the world, universe and beyond has the potential to be understood. It is a claim that can not be proven but, ultimately, no explanation can be entirely proven – it can only be checked for consistency against other explanations.

Confused by my last sentence? If so, that is exactly how I felt while reading 75% of this book. Still, there was at least 25% of the book that I understood, and Deutsch has quite a few takes (many inspired by Karl Popper, the philosopher from the late 1900s).

Here is a breakdown of what I think I have understood so far. I don’t have the intellectual background to be able to refute many of his arguments, so I present them largely as he does, providing only what minor challenges to them I can.

Justified belief vs fallibilism

When it comes to knowledge, we often refer to something being true when it is deemed true by some higher authority. This may have been a God (or high priest) proclaiming a truth of how the world was created. In more recent times it may be the World Health Organisation on how Covid-19 is spread, or the International Monetary Fund on how third world economies should run their economies. [Both are my examples, not Deutsch’s.] Historically, we had “trust the Gods”, now we sometimes have “trust the scientists” or “trust the experts”. In cases like this where truths are defined by appealing to a higher power, the approach is called justificationism.

The alternative to justificationism is fallibilism, which is the idea that there is no way to fully prove any explanation, and all explanations are error prone.

Importantly, fallibilism does not mean that we cannot explain anything, it means that we have to always expect we could be wrong. Fallibilism also does not mean that all explanations are equal. There is also a hierarchy in the quality of explanations, and that hierarchy is in how universal (or difficult to vary) an explanation is. For example, Newton’s classical theories of gravity were good, but Einstein’s are more general. I’ll touch more on this hierarchy of explanations in the next section.

The scientific method is not just about being testable

Deutsch points out that we often think about the scientific method as putting emphasis on testing our predictions. However, testability is not sufficient.

For Deutsch (and I think Karl Popper), the scientific method is about more than just being able to make predictions. For example, you might well be able to predict that a bird will reappear at the end of a magician’s trick. You may observe this repeatedly and this prediction may be borne out empirically, and others are able to test this prediction. However, this is not the scientific method, because you still may not know how the trick works. The belief that science can only provide predictions, but not explanations, is called instrumentalism and it is inherent any time we draw a graph of two variables without providing an underlying explanation. The scientific method is about explanations for causation, not just demonstrations of correlation, which is instrumentalism.

As a brief aside, the statement “this is my truth” can be interpreted as a form of instrumentalism. It takes an approach that there are no underlying explanations that can objectively be considered, there are just predictions and occurrences, and all explanations are entirely subjective and depend on your personal point of view. The scientific method, by contrast, at least according to Deutsch, requires explanation to be possible.

Zooming back out, for Deutsch, the scientific method involves explanations that are testable and lastly these explanations have a hierarchy, so not all explanations are equal. That hierarchy is determined by how difficult the explanation is to vary.

Let me start with an explanation that is easy to vary. Consider an explanation whereby global warming is explained by a God Athena, who becomes angry when humans emit carbon dioxide into the atmosphere. When we emit CO2, she causes heat to be retained close to the earth’s surface.

You might argue that the Athena theory is not testable, not least because there is no precise description of how she retains heat close to the earth’s surface. But, it is testable. You could monitor CO2 emissions and heat flows and determine that’s Athen’s prediction of a warming surface is correct. Conversely, there are aspects of what we consider good explanations that are not testable. For example, we can test the absorbing and emitting properties of carbon dioxide, but, we cannot conduct an experiment that independently varies the carbon dioxide content of the atmosphere while holding all else constant. So, what makes our greenhouse effect explanation better than the Athena explanation? The answer is that in the greenhouse theory, the details play a precise role in the explanation and are hard to vary. This is unlike Athena’s anger, which is imprecise and easy to vary as an explanation – making it inferior.

Edison’s 99% perspiration was not mindless

“Invention is 99% perspiration and 1% inspiration” is a quote commonly attributable to Thomas Edison of lightbulb and GE fame.

Deutsch argues that the 99% perspiration is in fact creative work. It typically involves human thought, not mindless work that a computer could do.

If a portion of that 99% perspiration can be done by computers it is only because something was learned by humans when they did that perspiration and programmed the computer! By definition, if something has been automated, the fact that humans can automate it means that there was learning/creativity/innovation in the work – so it wasn’t really perspiration.

In short, if a portion of perspiration can be automated, then there was learning in that perspiration!

Creationism will become an unremarkable attribute of a god

Virtual reality will become so good that we will be able to recreate a high fidelity version of life on earth. At that point, it will be unremarkable for a god to have created the earth.

Genes and Memes – no, not the kind of meme you’re thinking of

Deutsch makes the case that there are two pieces to human progress. First – genetic evolution – a slow phenomenon, and second – ideas (what he calls memes) – the more rapid phenomenon that has driven human progress over the last few thousand and, particularly, few hundred years.

Human genetics provided a platform allowing humans to develop ideas and pass those ideas down through generations as knowledge. Over the last thousand of years our genes haven’t changed so much as our memes.

Why then have humans seen such strong technological progress in the last hundred years, as compared to the last thousand. Deutsch’s argument is that we have long had the capacity to build an infinite databank of ideas, but we did not always have the culture of criticism that permitted (much less encouraged) such growth of ideas.

While critical cultures have existed for periods before (for example, during the Medici’s in Italy in the 15th century), they did not last or develop technologically as the West has done since the Englightenment.

What could some alternate theories be to Deutsch’s (and I am no expert here)?

Genetic improvement: One theory might be that humans adapted through evolution to be superior from the 17th century onwards. This seems unlikely given genetic evolution has a much longer time scale than centuries.

A triggering technological event: Another hypothesis is that we discovered the printing press, and that allowed the proliferation of ideas. A counterargument is that – while books and information were pivotal to growth – much technical progress did not occur for some centuries after Gutenberg’s press in the 15th century. There are also many other discoveries (steam engine, algebra) that we could point to as important, but it is difficult place one single discovery above all others. If we can’t pick out one single discovery, we revert to the question of why there seems to have been a collection of discoveries focused on the most recent centuries.

Geography: In the book “Guns, Germs and Steel”, Jared Diamond makes the argument that progress depended on geographical factors, such as animals being available for domestication in Europe but not so much in Africa or South America. As I see it, this seems a little bit like the “triggering technological event”, except the theory is of triggering geographical factors (presence of animals, good weather, good terrain etc.). However, these geographical factors have existed for quite some time longer than the technological progress of the last centuries, so this theory does not fully explain the question of why now? Why the progress this last century and not in the first or the fifth or the tenth century?

Back again to Deutsch’s argument that ideas drove progress and that the growth of ideas emerged from societies allowing criticism. Back in 0 BC, or even before, humans had the genetics to create knowledge as are doing now, but they didn’t. Why not? Because their societies didn’t encourage critical thinking – says Deutsch. Why then didn’t those societies allow critical thinking but later ones did? It’s not clear, and – to be fair – Deutsch is clear that this is a difficult and unanswered question.

The low hanging fruit in science and engineering is not gone

Going through a PhD in mechanical engineering and making use of equations that Sadi Carnot invented in the 18th century, I could only think that all of the low hanging fruit in science and engineering was gone.

Richard Feynman seems to have subscribed to a similar view, and felt there could only be a finite amount of theories that would soon all be discovered.

Deutsch’s philosophy is the exact opposite. Deutsch believes that there are infinite theories that remain to be discovered and so we not even close to the end. In fact, we will always be close to the beginning.

What is Deutsch’s basis for this? First of all, he points to all of those who have proclaimed the end of science before – only to be proven wrong. A good example is scientist Albert Michelson (of Michelson-Morley experiment fame) who – in 1894 – proclaimed that all physics that remained to be discovered was about the sixth decimal place. In other words, the low hanging fruit was gone and all that remained were minor tweaks to explanations and discoveries.

Second – and a rather technical point – we still do not have a reconciliation of quantum theory and general relativity. This is a clear and known gap in our knowledge of one of our most universal theories to date.

Third, it has been common in our past for theories that were widely accepted to be replaced by completely new frameworks. For example, Newton’s theory of gravity has been replaced by Einstein’s relativity. While much of Newton’s theory is a good approximation for behaviours on earth, Einstein’s theory is more universal and very different mathematically.

So, I left that chapter with more optimism than when I finished my PhD.

The Drawbacks of Proportional Representation

Today, many countries are migrating towards proportional representation – as we have in Ireland. The touted virtue of proportional representation is that it does a better job of representing minorities, which is true. Deutsch makes a number of counterpoints:

Deutsch – taking from Karl Popper – believes government should be designed such that rulers can be removed without violence if they are doing a bad job. Proportional representation means that governments are often composed of a coalition, making it difficult to remove the entire government from power (inevitably, some of the parties in power often stay on). Furthermore, proportional representation gives disproportionate power to the kingmaker parties – those that are often third or fourth in the polls and determine which larger party is in power. This means that the top polling parties are less incentivised to adapt their policies to the public, as their goals shift towards accommodating minority partners. Seeing the Irish system (and I believe the Israeli and Dutch systems are similar), I can appreciate Deutsch’s points.

What does this aside on political systems have to do with the growth of knowledge? For Deutsch, political systems should not solely be about who should be in charge, they should allow for the testing of different options that build upon knowledge, and allow for the bad options to be ditched. The question for Deutsch is not: “Is a democracy or a dictator or a monarchy better?”. The question is: “Which system allows for bad systems or policies to peacefully be cycled out?”

I won’t get into the discussion here, but Deutsch additionally makes the point that any system of representation will have logical inconsistencies, meaning that achieving “fair representation” cannot be the core goal of a political system. Improvement of knowledge can be the goal.

It was a tragedy that Easter Islanders survived for so long

I’ll finish on this example from one of Deutsch’s final book chapters.

Deutsch tells the story of two documentaries about Easter Island – an island where the civilisation went extinct in the midst of building large stone monuments.

The first is by Jacob Bronowski (commissioned by David Attenborough) and called The Ascent of Man, and it portrays a civilisation that failed in making sufficient technological progress to survive.

The second is by David Attenborough and called The State of the Planet. It draws a parallel between how the Easter Island civilisation died as it did not look after their environment, much as we today face extinction if we do not look after planet earth.

For Attenborough’s recent documentary, the extinction of the Easter Island civilisation was a tragedy. For Bronowski, the fact that the Easter Island civilisation survived for as long as it did was a traged. Whereas Attenborough sees a failure of responsibility to the planet, Bronowski sees a failure of responsibility to themselves as their population suffered needlessly building pointless stone monolith’s rather than thinking critically about how to advance their civilisation. You might think both are arguing for the same thing, but Deutsch thinks not. Attenborough is saying to go backwards to a life that is more sustainable. Bronowski is saying to go forwards and progress to the next level of technology.

In doing so, Deutsch raises the question: “What is the basis for sustainability?” Should sustainability be based upon living standards of the 1800s or the 1900s or the 2000s? Or is sustainability the wrong approach? For Deutsch, the answer is not sustainability, it is infinite progress.


What is a Great Debate?

What is a great debate?
Is it one where the opponent is crushed?
Or where both are evenly matched?

What is a bad debate?
Is it Trump’s second debate with Biden?
Was it bad for a lack of content or an excess of drama?

Is a debate great in the same way as a sporting match?
Is a debate great in the same way as a physical battle?
Or, is knowledge what makes a debate great?

What is a great sporting match anyway?
Is it France vs Brazil in World Cup ’98?
Was it great because you are French?

Are the greatest debates ones where sides never meet?
Debates between Newton and Einstein?
A debate between Karl Popper and Rene Girard?

Maybe the greatest debates only have one side?
Why toot your own horn,
When two can toot two?

Are the greatest debates between equals?
Or, between unequals,
Where David triumphs over Goliath?

Whether for victory,
Or for knowledge,
What, for you, is the greatest debate?


Liverpool Jerseys and Ferrari Cars: A Cultural and Business Time Capsule


*Football jerseys and formula 1 cars provide a time capsule of advertising trends.

*The trends tell us about cultural norms (what ads are accepted) and also about what industries had money to spend on ads.

*Very roughly, football moved through periods of electronics advertisers, through to alcohol, through to financial services from the 1980s to the present time.


It’s interesting to look at the types of business sponsoring sports teams over time – both from a cultural and a business standpoint. From a cultural standpoint, advertising tells us something about what’s allowed or banned at a given time, e.g. cigarettes or alcohol. From a business standpoint, advertising tells something about what companies get the highest return from reaching a large audience.

The pre-sponsorship era:

The last jersey I see on footballkitarchive.com without sponsorship (except for the shirt maker, Umbro) is the 1976/77 season:

Liverpool Jersey 1976/77

The last Ferrari F1 car I can see without sponsorship on motorsport.com is the 1967/68 car:

1967-1968: Ferrari 312/67
Ferarri 1967/68

Into the 1980s

Here’s the 1988 Liverpool jersey, sponsored now by home appliance maker Candy:

1988 Liverpool jersey

Home appliance and electronics sponsors were hot in the late 1980s. Here is the 1988 Arsenal jersey:

1988 Arsenal Jersey

By the 1980s, the Ferrari is plastered in advertisements, notably from Goodyear (a tire supplier), but also from Longines (luxury watches) – reflecting the demographic of at least some F1 fans.

1985: Ferrari 156/85
1985 Ferarri

1990s and early 2000s

Companies went heavy on the alcohol and cigarettes in the 90s. Here’s an iconic Liverpool jersey from 1995/96:

1995/6 Liverpool jersey

Leeds United in 1996 were still on the electronics train with Packard Bell:

1996 Leeds Jersey

But Leeds had caught up to the alcohol trend by 2000, with this iconic Strongbow cider sponsorship:

2000/1 Leeds Jersey

And here’s one of the iconic Marlboro cigarette sponsorships of Ferari in 1995:

1995: Ferrari 412T2
1995 Ferrari

And here’s the 2005 Ferrari – 10 years later and not all that different at first glance:

2004: Ferrari F2004
2005 Ferrari

Late 2000s and 2010s

Even up as far as 2009/10, Liverpool are going strong with the Carlsberg sponsorship with this goalkeeper’s away jersey:

2009/10 Goalkeeper Liverpool Away Jersey

The move to financial services sponsorships is on the way, and here’s the 2015 Ferrari:

2015: Ferrari SF-15T
2015 Ferrari with Santander Bank

Liverpool also go to financial services with Standard Chartered:

Liverpool 2015/16 Jersey (a nice tailored fit!)

Today – 2021

And here we are with the latest cars and jerseys of today:

Liverpool 2021 Jersey
2020: Ferrari SF1000
2020 Ferrari (except a redesign in 2021)


Is Joking Intelligence?


  • Jokes are hard to explain and highlight how we cannot explain creativity.
  • Many engineers and scientists have tried to achieve creativity using formulas.
  • Creativity might be something different than formulas.


David Deutsch thinks that achieving human level intelligence with computers is a matter of understanding the essence of creativity (The Beginning of Infinity). A joke is an interesting example of creativity because humour is difficult to explain.

Is joking “Intelligence”?

One type of joke involves taking something from one context and replacing it with something from another context that is only related in a limited – and often absurd – way.

Girl (holding up an an apple): Mammy, would you like to try this vaccine?

Mother: That’s not a vaccine young lady. It’s an apple.

Girl: Sorry, I thought I’d give it a shot.

Sample joke.

I don’t think I can perfectly explain why this joke is funny, but let me try anyways:

Attempt #1: The joke is funny because the word “shot” is a pun. This is a bad explanation because it now kicks the can to explaining the word “pun”.

Attempt #2: A joke is funny because it explains something non-obvious. People appreciate explanations, particularly non-obvious explanations, and react by laughing. In this case, it is non-obvious why the girl would call an apple a vaccine. The explanation is that calling an apple a vaccine does make sense if it is a reminder of the broader context of the COVID outbreak the girl and her mother are living through. This metaphor is further engrained by the pun, “shot”, which ties together the attempt at a joke and the vaccine. I think this is a bit better as an explanation but it is also a bit circular because I’m saying that the joke is partly funny because it is pointing out that the setup line is a joke.

Ok, I’ve run out of more ideas, but comment below if you have some.

The temptation with creativity is to find a formula

Computers are good at implementing formulas. However, formulas/models seem to be human. Even if the computer is “machine learning” like Alpha Go for Chess, there were people who developed the model.

If we want computers to come up with their own formulas, how would we do that? The temptation again is to come up with a formula that allows computers to come up with formulas.

This leads to the question of whether artificial intelligence could be achieved simply by having sufficient layers of formulas. This is what it seems humans have been trying to do roughly since Alan Turing defined a test of whether an AI is real back around the second world war – use more formulas with more data and more computing power.

Another possibility is that creativity is a separate thing to formulas. The End.


Now is a good time to prepare for your phone and computer getting hacked


  • I think that the risk of a major cybersecurity threat – of similar economic magnitude to COVID – is under-appreciated.
  • My friend recently got hacked and this highlighted how easily we can lose access to our email, documents and bank account if our computers and phones.
  • It’s good to take steps to avoid being hacked, but, it’s equally – if not more – important to be prepared to recover from a hack.


Last week, one of my friends got seriously hacked and lost control over their phone, computer and internet router. They have made good progress in resolving the issue, but it was a reminder for me to revisit my own preparedness for a computer and phone hack. In particular, it made me realise how much security is now dependent on our phone.

I think that a major global cyber hack is likely over the next decades and will arrive unexpected, much like COVID-19. I’ve taken some time to review the security of my information and am presenting my approach here. I’m not a cyber or software expert, and this approach is imperfect, but I hope that it serves as a reminder to you to check if you could survive a hack.

Plan to survive, not just avoid, a hack

Yes, it’s good to take measures to avoid getting hacked (i.e. someone getting control of your computers and digital information) – and I’ll cover some of those. However, I don’t believe any system can be 100% protected from hacks, so it is best to assume you will get hacked and think about how you can cope in that scenario. (I highly recommend the movie Zero Days as an eye-opening insight into how easy computers are to hack).

0. Simulate a Hack

Even if you don’t read any of the rest of this article, I recommend doing the following:

a. Open up a browser in private mode.

b. Navigate to the website of an important service (e.g. your bank or email provider).

c. Assume that you no longer have i) your password/or it has been changed, and ii) your phone.

d. Try to log in to your account.

If you can’t find a way to log in, then you probably are vulnerable if your phone and computer get fully hacked.

Oftentimes, the only ways to avoid this are i) having back-up codes on paper – if the service allows that, ii) having a backup e-mail to recover your password, iii) calling the company (often a slow and painful option).

1. Two Factor Authentication

For important accounts, I always turn on two factor authentication (2FA). I use an authentication app like Google Authenticator where possible. Sometimes SMS/text messages to my phone are the only option for 2FA. SMS is better than nothing, but text messages can be intercepted, so two factor using an authentication app is preferable.

2. I use a VPN (virtual private network)

I always use a VPN on my laptops. A VPN helps to hide your IP address (the address of your computer when accessing the internet). This helps to stop your computer from being targeted – especially if on a public wifi network.

For a computer, there are many VPN services available like Proton VPN or Nord VPN or Express VPN that you can buy for a few euro per month.

There are also services that provide VPN for mobile phones. These are less common but becoming more common. Brave VPN is one new option for iOS that I use.

3. Consider putting the most important passwords on paper

Yes, there is a risk that somebody could find/steal these passwords and that risk is worth considering. However, it’s always possible that i) your password gets stolen by hackers if you’ve stored it on your computer or phone, ii) you forget your password, or iii) you just die. Writing your password on a piece of paper opens you up to the risk that someone finds the piece of paper, but it does a good job addressing risks i-iii. My view is that there are many more hackers that can steal stuff from my computer than there are thieves that can find a piece of paper.

The risk of someone finding the piece of paper can also be partially mitigated by setting up the app/service so that you are notified – to your e-mail and/or phone – if your password is updated or if there is a sign-in from a new device. If someone finds your piece of paper, you can – in theory – take action right away once you are notified that they are trying to log in.

4. Save passwords to a password manager

There are many options for password managers – from web browsers to Keepass to OnePass. These are not perfect but are often better than an unprotected Excel or Word document.

The benefit is that you can use different passwords for every website. This is important because it is highly likely that some of the services you use will get hacked and passwords will get exposed to the public. Of the 200+ services that I have passwords for, I know of at least three that have been compromised. If I had used the same password for all services, then the password for all of my services would be compromised!

Of course, holding all of your passwords in one password protected database isn’t perfect. There is the risk that your password database gets hacked – especially if you hold the database on a device that is online. This is why two factor authentication is important to use for critical accounts.

There is also the risk that your password database gets damaged and the passwords can’t be recovered. One way to provide some protection for this is to regularly back up your database to a hard drive that you store disconnected from the internet.

5. I do regular offline back ups of my password database, email account, website/blog and documents

One outcome of a hack is that the hacker uses your personal information to steal money or make use of your identity. Another possibility is that you are drawn into a large scale hack where an attacker gains control of your accounts but their ultimate target is someone else in the larger group being hacked. They may not care about your information, but you may still lose it or be denied access to your accounts. In other words, it’s possible you will simply be collateral damage.

If you do lose access to your accounts and/or information, then you need a backup copy – ideally one or two offline hard drives. I try not to be in a position where my most important information is only stored online.

6. Create a Backup E-mail with a different service than your main e-mail

For example, if your main email is iCloud, use Gmail as a secondary email; if your main email is Protonmail, maybe use iCloud as your backup. The key point is that I don’t want to have your backup e-mail being managed by the same service as my primary e-mail – that’s a recipe for disaster if my main email service is hacked.

7. Prioritise and try to keep things simple

We all have so much information that security can seem overwhelming. Try to prioritise what is most important to you. Maybe that’s your e-mail. Maybe it’s your bank account. Maybe it’s your cryptocurrency.

Work through the exercise of trying to recover your accounts and critical documents if you lose your password, your computer and your phone.

8. Wildcard measures

Take some additional precautions that aren’t disclosed on a public blog 🙂

Invitation for Comments

What further tips do you have on avoiding and surviving hacks? Do you have any concerns with any of the approaches above? Please do comment below.


Berkshire’s largest investment in 2020 was itself


  • Berkshire Hathaway’s 2020 Annual Report was released on March 1st 2021.
  • Buffett did not make any big sales or acquisitions in 2020, so the business is little changed from 2019.
  • Operating earnings, which excludes stock gains/losses, were down close to 10% in 2020 compared to 2019.
  • Berkshire continues to accelerate the rate at which it is buying back its own stock, spending $24.7 billion on buybacks in 2020, compared to $5 billion in 2019.
  • Buffett notes that Berkshire’s insurance businesses are at a particular advantage over competitors because Berkshire can invest the float in equities instead of low yielding bonds.
  • Buffett breaks down the ownership in Berkshire into five groups: Himself, index funds, money managers, individuals who dip in and out of Berkshire shares, and individuals who invest in Berkshire for the long haul.
  • Buffett presents, for the first time, the business as a combination of four jewels: A 5.4% stake in Apple, Berkshire Hathaway Energy, BNSF Railway, and GEICO.

Why I own Berkshire and why you might not want to:

Being known as the most famous investor in the world sets a high bar for what to write in your annual report. The 2020 report is not Buffett’s finest, but he does provide new perspectives on where value lies in the business, and how he breaks down Berkshire shareholders into different groups.

If you’ve read about how I invest my money, you’ll know I hold 30% of my liquid net worth in Berkshire stock, primarily because I believe in Berkshire’s investing principles – documented in Buffett’s long history of annual reports.

That said, tread carefully before making an investment in Berkshire because i) it is likely that Buffett’s successors will not perform as well as he did, ii) Berkshire is now very large, which makes it harder to find good investments, iii) if you do what I’m doing, you’re putting a lot of eggs in one basket – it’s always possible for a company to be hit by fraud or a scandal that takes it down, iv) Berkshire doesn’t do a lot of “tech”, so maybe the business will fall behind the rest of the economy, and v) cybersecurity risks could impact the business – it’s an under-appreciated risk but listed as the first risk in Berkshire’s annual report!

Berkshire in 2020 in Brief:

Excluding any gains and losses from investments, Berkshire’s operating earnings dropped from roughly $24 billion in 2019 down to $22 billion in 2020. This is nothing remarkable – good or bad – given the COVID background.

Berkshire bought and sold some equities in 2020, but nothing accounting for more than a few percent of the overall business’s value*. The largest line item to report was the repurchase of its own shares, with Berkshire buying almost $25 billion worth of its own stock in 2020 (getting close to 5% of outstanding shares).

Compare this to 2019, when Berkshire repurchased only $5 billion of its own stock, and you can see that Berkshire is very likely to become a large net purchaser of its own stock over time. This is exactly what Charlie Munger – Buffett’s sidekick – has said in annual meetings.

I wouldn’t be surprised to hear that Berkshire buys back even $50 billion in 2021 if the stock price remains in the $200-$250 range (B shares).

A Note on a Moat

Buffett loves to talk up the businesses that Berkshire owns, and that can become fatiguing. One good point he makes this year, given interest rates have reached near zero, is that Berkshire’s insurance business – which includes GEICO – is at a particular advantage over competitors. Pure-play insurance businesses can be restricted – for regulatory reasons – to investing the premiums they receive primarily in bonds (currently earning close to zero income). Berkshire – as a diversified company in many businesses – is able to weight the investment of the insurance premiums it holds much more towards equities.

Note: Buffett isn’t earning a whole lot on the ~$130 billion in cash on the company’s balance sheet. Ironically, the amount of cash he has is very close to the size of the insurance float ($138 billion) where he is emphasising his competitive advantage 😂.

Investor Groups

Buffett didn’t get deep into politics in this year’s report (no surprise there), but he did split up Berkshire shareholders into different groups for the purpose of emphasising his admiration for group 5:

  1. Himself – a shareholder group with a limited lifespan.
  2. Index funds – interesting that he would break out index funds as a category. Buffett has written favourably of the index investing approach in the past. However, the language in this report seems a bit more mixed: “Index funds, it should be emphasized, own Berkshire shares simply because they are required to do so”. Maybe I’m reading too much into it this as a slight on index funds – or maybe I want to read too much into it!
  3. Money managers. An “honourable but difficult occupation”, Buffett says in the report. Maybe because it’s his own occupation too 😂? (A bit harsh from me there. In fairness the whole thing about Buffett is that he has all his net worth behind what he invests in – that’s very different from most money managers.)
  4. Individuals who dip in and out of Berkshire. Enough said.
  5. Long term individual shareholders in Berkshire. Clearly, this group is important to Buffett. He didn’t mention it in this annual report, but he has previously emphasised how many individuals invest their life savings in Berkshire, and this greatly influences Buffett in how he runs the business.

The most notable thing about this breakdown for me is that #1 won’t be around for ever, and, in a few decades time, we’ll look back at this as evidence of how index funds became a class of shareholder.

The Crown Jewels

Buffett highlights that the Berkshire business rests on four crown jewels:

  • Berkshire’s stake in Apple (#1) – valued at $120 billion at 2020 year end.
  • BNSF Railway (#2) and Berkshire Hathaway Energy (#3): Generating a combined $8.3 billion in earnings in 2020. [Conservatively, applying an earnings multiple of 15, that would value #2 and #3 at about $125 billion.]
  • Insurance businesses including GEICO (#4): Worth $138 billion, if you value it purely off of float. [actually, Berkshire often runs a profit on insurance so it’s likely worth more, although you could argue the float is worth less than cash because it cannot be invested quite as freely].

So, roughly speaking, the four crown jewels are worth around $360 billion in today’s market.

Add to that about $130 billion in cash, $160 billion in stock holdings excluding Apple, and another $6 billion in operating earnings (worth $90 billion at a 15X multiple) excluding BHE, BNSF and investment income. This brings Berkshire towards a $740 billion market cap, which is equivalent to a price per B share of roughly $330.

At the time of writing, Berkshire is at a price of $250 per B share and a market cap of around $560 billion, so you can see why Buffett and Munger have been buying back the stock from 2020 lows of around $162 all the way up to the present time.

To be clear, I’m not saying it’s obvious that Berkshire’s stock price will go up soon. You could pretty much have done the same calculation as above for the past five years and gotten the same favourable result – but Berkshire stock prices have not seen strong growth over the last decade.

*Rough calculation: $6 billion = $22 billion in operating income less $8 billion from BHE+BNSF less $8 billion in investment income (e.g. dividends from equity holdings).

A note on the annual shareholder meeting:

I’m not really sure why, but the Berkshire annual meeting on May 1st will be streamed from Los Angeles rather than Omaha this year. The good news is that Charlie Munger will be back on stage to share a few zingers. Ajit Jain will also be on stage, which I’m looking forward to because I haven’t really heard him speak much before. He heads up insurance operations for Berkshire.

*Quite awkwardly, Berkshire sold it’s positions in JPMorgan (Todd Combs, who works for Buffett, is on the JPMorgan board) and also in CostCo (Charlie Munger is on the board). Neither positions were large, but the board connection struck me as awkward – certainly for the CEOs of JPMorgan and Costco to have board members who are selling all of their shares. Ouch! 🤯


How to Legally Offer Cheap Antigen Testing within your Company or Organisation (non-US)


  • Antigen tests give results in 15 mins and are 10X cheaper than PCR tests.
  • There are antigen tests available that are EU approved for clinical use.
  • As an employer or organisation, you can legally do antigen testing if you contract a clinical work to oversee the tests.


Employers/organisations currently have three options to avoid the spread of COVID-19:

  1. Use social distancing and improve workplace ventilation = good idea, not perfect but should be continued.
  2. Monitor symptoms, i.e. give employees questionnaires for symptoms (fever, cough etc.) = inaccurate because i) infected people can spread before showing symptoms, ii) not everyone shows symptoms.
  3. Do PCR testing = $100+ per test and slow to get results.

In the European Union, there is now a fourth option that is legal, which is antigen testing with clinical supervision. This blog provides some ideas on how employers/organisations could implement this approach to make their workplace safer.

The benefits of antigen testing

  1. You can test cheaply (<10 euro per test) and therefore frequently.
  2. Test results are immediate (<15 mins).
  3. Tests are quite accurate at detecting contagiousness (>95% sensitivity).

Here is my demo of taking an antigen test.

Here is a great resource for antigen tests in general – www.rapidtests.org – built by a Harvard professor.

The one caveat is that antigen tests are currently only approved in the EU for clinical use for symptomatic patients. However, you can comply with this requirement as an organisation or employer by contracting a nurse/doctor to oversee testing. You could maybe even get creative and do this by video.

How to implement this in your workplace or organisation:

  1. Get some antigen tests:

The test with the highest sensitivity I can find is the Flowflex Antigen test (EU approved but not approved in the US yet). These can be purchased in bulk for less than 10 euro per test. Comment below or contact me and I can connect you to a bulk supplier (minimum order is 800 tests) if that would be of help.

2. Hire a nurse or doctor to oversee testing:

I’m a bit green on this front but I think it should be possible – potentially via locum nurse/doctor hiring services. There may also be insurance matters to consider.

Please contact me if you know how to hire nurses/doctors on contract. I think the ideal would be to contract with a locum agency that can provide a nurse’s or doctor’s services for a certain number of hours per week.

3. Get the nurse/doctor in once or twice per week

Either send antigen tests to people’s homes and then have the nurse/doctor instruct them by video call OR have people tested as they arrive at work.

Some caveats:

  1. Testing is not a replacement for social distancing, masks and good ventilation. Good practises should continue.
  2. If there is a positive test, the person should go for a confirmatory PCR test, if possible.

What is the benefit for you – an employer or organisation?

  1. Simple – you’re going to catch more people early on that are spreading the virus but don’t show symptoms.

Comment below with other thoughts OR if you would be interested in this kind of service as an employer or organisation. The service could be: a) remote video antigen testing for your employees – probably could work for even small organisations, b) on site testing, probably only makes sense for sites with 50+ employees/people.

Lastly, my view remains that the Irish government should provide emergency approval for rapid antigen tests (that are already allowed for clinical use) be allowed for self testing. This would be much better than people relying on symptoms for diagnosis. Here is my analysis of why.


Two things to Try for Twenty Twenty One


Journaling and Planning

For four to five years – when I was about 17 to 21 – I kept a daily journal. It was too much effort and I wasn’t enjoying it, so I stopped for almost ten years.

Over the last year, I just keep a few notes roughly once a week about things that happened and intentions I have for the next day/month or year. A few of my short learnings on making journaling easier:

  1. Pick out a dense book and read a page when you do a diary entry. If there’s a quote I like, I write it down in my journal. So far, I’ve used Letters from a Stoic (Seneca) and Meditations (Aurelius) for this.
  2. Buy a nice little diary. I’ve spent a more than just buying a cheap book. It put more love/soul into it. Splash out and get a wooden cover or something.
  3. I write down a few things that happened and how I feel about them.
  4. I write down a few intentions. I try to avoid setting myself up for failure with things outside of my control (e.g. do a fundraise for $X) in favour of things I can control (e.g. remain calm in handling upcoming difficult conversation y).

Even all of this stuff can be boring and/or not enjoyable. An alternative option is to just write once a year or once every few years.

You can do that on futureme.org for free and have the letter e-mailed to you in 1, 3 or 5 years (or whatever). It can be nice to do it with a friend or a partner/spouse. Hat tip to my brother for this idea.

Improve the News

A lot of people are always harping on about how X is so divided or polarized (and it’s getting worse). X can be people, the country, the world… insert your word of choice.

My favorite question to ask back (actually I usually just stay quiet) is: Well, what are you doing to make things less divided, if you think that’s a goal?

There’s a dude at MIT (Max Tegmark) who actually is doing something on this front with www.improvethenews.org – it’s a website that allows you to move sliders between left and right to see the news change. Try it for yourself.

Sliders set to “Left” and “Pro Establishment”
Sliders set to Right and Anti-Establishment

Lastly, my favourite tab settings:

Sliders to the middle

Might save me a bit of time compared to alt-tabbing between CNN and Fox…

Lastly – my January newsletter will be coming out later this week for subscribers. Sign up to get ahead on new business ideas:

Success! You're on the list.

DeFi Mini Course: Part 4, Decentralized Hedge Funds

Before You Get Started on Part 4.

In this part, you’ll learn how to buy tokens in a decentralized hedge fund and then stake those tokens so you can vote as well as earn rewards. If you’ve missed Parts 1 – 3 of this course, here are the links – all prerequisites for this Part 4:

  1. Part 1, Wallet Setup for DeFi.
  2. Part 2, Uniswap, a Decentralized Currency Exchange.
  3. Part 3, Lending Platforms.

At a minimum, you’ll need a MetaMask wallet with about $50 worth of Ether to do this part of the course. Skip back to Part 1 for guidance on how to add to your wallet.

This course is priced at $9.99. If you find the course helpful – but don’t have the money to spend right now – I’d appreciate if you could instead share the course on social media or with two friends.

Disclaimer: Crypto is still in its infancy and very high risk. I recommend setting the expectation that you will lose all of the money you invest as you’re learning and treat it as an education. Throughout the mini-course, I will indicate the cryptos I hold myself, so at least – if you do copy me – I’ll lose money if you do! This course is not investing advice, it is intended to teach you how DeFi works! You are reading these materials and taking any guided steps at your own risk. Crypto is technical and funds are easily lost if you are not careful with passwords as well as sending and receiving money to/from the right wallet addresses – not to mind bugs in nascent software platforms or bad actors. Learn with small amounts to minimise your risk.

Part 4, Decentralised Hedge Funds

The problem of fees when earning crypto income

You’ll recall from Parts 2 and 3 that transaction costs for moving, exchanging or lending cryptos are high. As of Jan 2021, it can cost $20 per transaction just to move crypto from one wallet to another. Here are two the two basic transaction costs in DeFi – through the lens of a lending example:

  1. Depositing crypto into smart contracts. You’ll recall, from Part 3, lending out some USDC on Compound or Aave. Depositing the crypto will have cost you Ether to pay for gas.
  2. Withdrawing crypto from smart contracts. Once done lending, you’ll have to withdraw your crypto from the lending platform, again costing you gas.

Unfortunately, it’s worse than that because currently – in the early stages of DeFi – significant returns are earned not only through interest but also by receiving platform tokens as rewards. As a specific example, lenders and borrowers on the Compound platform earn Compound tokens for their participation, in proportion to interest earned/paid. This leads to a third transaction cost:

3. Rewards token withdrawals/conversions. You earn Compound tokens as you lend on the platform. At some point, you’ll want to convert these compound tokens into another crypto – likely back into USDC if you are engaged in a USDC strategy.

So you have this cycle where you earn Compound and then have to pay to convert it into USDC and then pay again to lend that USDC out.

Decentralized hedge funds as aggregators

Given the high cost of gas (transaction fees – which are largely flat fees per transaction), it makes sense that individuals might pool their funds together in order to reduce gas costs as a percentage of their funds. This is where platforms like harvest.finance and yearn.finance emerged.

A group of yield “farmers” pull together and write some smart contracts that allow them to invest in strategies (such as lending on Compound) together. Here is how one such strategy might work:

  1. A USDC lending strategy is set up that invests USDC on Compound.
  2. Compound tokens that are earned are automatically converted – using Uniswap or similar – for more USDC tokens, and this USDC is then reinvested in the USDC strategy.

This is called auto-compounding. The Harvest smart contracts automatically do the work of reinvesting profits. If you want to withdraw your funds from the pool, you get back your initial asset plus your proportional share of compounded profits.

The fees charged by Harvest

In return for this smart contract service, Harvest takes 30% of any profits made by investments (liquidity pools) on the platform.

That might seem like a lot! It is! However, gas fees are so high that investors still find it cheaper and easier to invest via Harvest and pay that 30% fee. As of Jan 2021 there is over $500M invested on the harvest.finance platform.

Where do the Harvest fees go?

They go to… yet another governance token – called FARM!

So the owners of FARM earn 30% of all profits earned by the liquidity pools on the Harvest platform.

Now, the harvest.finance platform offers all sorts of earning opportunities (lending, exchange pools etc.), so by owning FARM you own 30% of the profits across a wide variety of diversified yield farming strategies. Tasty! (also Risky!)

Now, before diving in with a DIY step, some risks with diversified hedge funds:

Risks of diversified hedge funds like Harvest.finance

As with any DeFi initiative, there is smart contract risk involved in diversified hedge funds, only this time you’re taking on multiple layers of smart contract risk. There are the smart contracts for the earning platforms (e.g. Compound or Aave), but now you’re taking on a further layer of risk by investing via the harvest.finance layer of smart contract. There’s actually a third layer of smart contracts as well, if you’re to earn FARM rewards, which we’ll get into in the DIY section below.

Things can go very wrong indeed on DeFi platforms. Take a look at when the Harvest platform itself was subjected to an attack in October 2020 with over $30M made by the attacker at the expense of the platform. In understanding the attack, it may be of help for you to read on further and then come back to this section where I’ll give a high level overview.

In October 2020, the USDC and USDT pools on Harvest were attacked. Each pool was investing assets on the Curve.fi platform. Essentially, the attacker used a large amount of money (~17M) to move the price of USDC vs USDT on Curve, and then swooped in and out of the USDC pool on Harvest with ~$50M – taking advantage of the price mismatch cause on Curve.fi . The attacker made over $500k each time doing this and repeated the attack over ten times on the USDC pool in Harvest, and then on the USDT pool – sucking out over $30M in value from Harvest in the process.

Why did this happen and how could it be stopped? At a high level, this happened because the attacker was able to do a lot of transactions very quickly. Flash loans (where you simultaneously buy and sell at the same time) were allowed. Flash loans were subsequently disabled on Harvest, and the Harvest smart contracts broken down into more steps – which increases transaction costs (gas) – but improves security.

Harvest even then pulled together a reparations program for those who lost money and created a token called Grain to work towards making them whole. Still, FARM trades at a huge discount (over 10X) to Yearn – a competing hedge fund platform that has a similar level of assets invested.

When it comes to DeFi – particularly multi layered contracts – Caveat Emptor!

Investing with FARM and in FARM

There are two different ways to make money with or in Harvest.

  1. Investing in liquidity pools – which is just pooling your money with other investors into third party strategies (e.g. on Compound or Aave or Uniswap liquidity pools).
  2. Investing in FARM token itself – which is like taking ownership in a hedge fund and earning a share of the platform’s profits.
  1. *Investing in Liquidity Pools

*Navigate to app.uniswap.org, connect your Metamask wallet and swap $10 worth of Ether for $10 worth of USDC (you’ll have to pay for gas as usual using additional Ether in your wallet).

*Navigate to Harvest.Finance and hook up your metamask wallet.

*Scroll down to “Stablecoins”, click the dropdown and seek out USDC.

You can see USDC at the bottom, put 10 in the USDC box. Notice, to the right of the USDC symbol you can see the annualised return on the asset (based on recent returns – not guaranteed annual returns!). You can also see how much USDC is in the pool ($18.71M here), and you can see how much you’ve already deposited (I’ve got 20.227… deposited). You can also see some symbols for Compound, Idle, and Farm – the cryptos involved in the strategy.

So what’s happening here? In this USDC strategy, your USDC is going into a pool of $18.71 worth of USDC and invested into an Idle smart contract. Idle is a platform that moves money between lending platforms (like Compound and Aave) depending on where the best interest rate can be received. Idle has a platform token called Idle that you earn for using the platform. So Idle then takes the USDC and further invests it – in this case – using Compound’s platform, where the USDC earns interest but also Compound tokens

It’s a very multi layered strategy, the overall effect of which is that you lend USDC and receive interest (paid in USDC), Compound and Idle in return. Furthermore, the smart contracts sell off any Compound and Idle tokens earned and use the proceeds to buy more USDC. This means you are auto-compounding USDC to the tune of 28.56% on an annualised basis (measured based on recent returns, not a guaranteed return!).

*Scroll down and click “Deposit and Stake”:

“Deposit and Stake” button at the bottom right

Which leads to the question – what’s the difference between Deposit and “Deposit and Stake”?

Well, by depositing, you are simply pooling your USDC and investing in the strategy as described above. Technically, what happens here is that you lock your USDC into a smart contract, and the Harvest platform gives you back a fUSDC token in return. (if you locked in USDT instead, you could get back fUSDT). This fUSDC token just serves as your receipt for your investment.

However! You can take that receipt (the fUSDC) and further entrust that to a Harvest contract in return for earning some FARM rewards. This is called staking, and means you trust the platform in holding your receipt.

Actually, the 28.56% return mentioned above includes all forms of return – interest, Compound, Idle and FARM. So if you don’t stake your fAssets (the collective term for assets like fUSDC, fTUSD etc) then you won’t get quite as high of a return.

2. Investing in FARM itself

Alright, so we’ve seen above how you can earn auto-compounding interest on specific assets using harvest.finance. You can alternatively get exposure to the whole platform of assets on harvest by buying the FARM token – which earns 30% of any profits made on the platform. This is a different kind of risk profile than providing liquidity to pools – as described with USDC above – in particular because there are no assets underlying FARM tokens. Specifically, if you invest USDC in a liquidity pool, you are entitled to withdraw that USDC at any time (absent any smart contract issues). By buying FARM you are not committing assets to a smart contract that earns a return, but rather just buying an asset.

*Navigate to app.uniswap.org, connect with Metamask, and convert $10 worth of Ether for $10 worth of FARM (you’ll need to do some back of the envelope maths to figure this exchange rate out.

*Navigate to https://harvest.finance/earn and click on FARM Profit Sharing.

*Now enter the number of FARM that you have and click stake:

Use the bottom fields to stake farm. At the top, you can see that I have 10.6 FARM staked at the moment.

A recap on what you get from staking FARM

  1. The ability to vote on FARM governance matters
  2. Profitshare of 30% of liquidity pools (you get this indirectly because the platform uses those profits to buy FARM on the open market).
  3. Emissions! Yes, Harvest is still in the early phases and they are giving out FARM tokens every week to different groups, including those who have staked FARM (and USDC as in the last example).

Some Parting Technical Thoughts

Where do platform rewards come from – such as Compound or FARM? As mentioned above, by using certain platforms you can earn rewards (e.g. FARM tokens earned for using Harvest.Finance). When a platform is getting set up, those initiating the platform often decide to mint platform tokens and distribute them in certain amounts over a certain period of time. The total supply of tokens to be printed may be limited (such as Celo Gold, CGLD), or may be at the discretion of a majority vote of the platform tokens (also known as governance tokens). The ways in which tokens are distributed also varies widely, but often is designed to reward the platform founders and/or incentivise the use of the platform by early adopters. In short, platform tokens serve at least two purposes:

  1. Governance – with a majority vote of the tokens allowing changes to be made to the smart contracts.
  2. Rewards – to incentivise certain behaviours on the platform.

Interestingly – some of these tokens have no intrinsic value in themselves, other than the ability to govern the platform. Some examples:

The Farm token is entitled to 30% of the profits from liquidity pools on the platform.

The Uniswap token is not entitled to any profits, but there is a provision that 0.05% of the 0.3% transaction fee can be redirected to the protocol at some point in the future.

The Compound token does not earn anything at the moment, but it does govern the platform, so its value accrues from the right to divert earnings to the token at some point in the future. Yes, interesting…

Where do the 30% of profits actually flow – are they directly paid to FARM holders? Actually, no. Profits are used to buy FARM tokens on the open market, which puts an upward effect on the price of FARM – it’s kind of like a company buying back its own stock.

Almost done with Part 4 on Decentralized Hedge Funds!

So you now have a way to invest in liquidity pools and strategies while mitigating fees and automating the selling and reinvestment of any platform tokens received.

You also have a way to earn a share of profits from a diversified range of strategies.

That brings us to the Part 4 quiz and the end of this introductory DeFi course. If you enjoyed it, consider letting one or two more friends know about it, and also signing up for my monthly newsletter so you don’t miss out on any more DeFi updates from me.

Once you’ve clicked Submit on the quiz, scroll up to see your answers!


DeFi Mini Course: Part 3, Lending and Borrowing

Before You Get Started on Part 3.

In this part, you’ll learn how to lending and borrowing works on COMPOUND or AAVE, and lend out some USDC for yourself.

You’ll need a MetaMask wallet with about $30 worth of Ether and $10 of USDC (from Part 2) to get part 3. Skip back to Part 1 for guidance on how to fill up a wallet.

This course is priced at $9.99. If you’re new to crypto, you’ll need to read through Part 1 and buy some crypto before clicking the Buy button. If you find the course helpful – but don’t have the money to spend right now – I’d appreciate if you could instead share the course on social media or with two friends.

Disclaimer: Crypto is still in its infancy and very high risk. I recommend setting the expectation that you will lose all of the money you invest as you’re learning and treat it as an education. Throughout the mini-course, I will indicate the cryptos I hold myself, so at least – if you do copy me – I’ll lose money if you do! This course is not investing advice, it is intended to teach you how DeFi works! You are reading these materials and taking any guided steps at your own risk. Crypto is technical and funds are easily lost if you are not careful with passwords as well as sending and receiving money to/from the right wallet addresses – not to mind bugs in nascent software platforms or bad actors. Learn with small amounts to minimise your risk.

Part 3, Lending and Borrowing

DeFi is like a bank, but you can participate in being the bank 🙂

When you borrow from a bank, there is collateral involved. If you borrow for a house, you promise them the house to secure the loan. Same if you borrow for a car.

As far as I know, you can’t yet borrow for a house or for a car on DeFi. Right now, you can only borrow to buy a different crypto, e.g. you can borrow Ether , but you have to offer some other crypto as collateral, say Bitcoin.

Now, this might seem pointless for day to day life and… It largely is! Bear with me…

The current reasons for borrowing and lending are – primarily – for financial engineering. I’m going to cover these financial examples because I want to provide a sense of the full range of DeFi services to further my point that DeFi is like a bank, but you can be served by the bank or being part of the bank yourself… for better or worse!

Borrowing on a DeFi platform

Let’s look at a hypothetical example where you start with 10 Ether, each Ether worth 1,000 USDC:

Grandad Mick goes to a DeFi lending platform (like Aave or Compound) and uses his Ether to borrow USDC. Using 10 Ether as collateral (together worth 10,000 USDC), the platform allows Mick to borrow 8,000 USDC (80% of your collateral). They won’t allow Mick to borrow the full amount of his collateral, just as most banks won’t allow you to buy a house without a deposit – unless it’s 2008…

You take that 8,000 USDC and you buy 8 more Ether. Now you have 18 Ether in total.

Let’s say Ether doubles to a price of 2,000 USDC – so you know have 36,000 USDC worth of Ether. Grandad Mick is in good shape! He pays back his loan of 8,000 USDC (plus interest – which we’ll ignore for now) and now has 28,000 USDC (or 14 Ether) – a tidy profit of 4 Ether.

Now, let’s look at the downside scenario. Let’s say that Ether priced in USDC drops by just 10% down to 900 USDC per Ether. Mick’s collateral of 10 Ether is now worth only 9,000 USDC, and his lending limit (80% of collateral) is now only 7,200 USDC – which is now less than your loan of 8,000 USDC. Here’s what happens next – and I’ll use the AAVE example.

The Aave protocol now allows great-aunt Mary – sitting over there in the armchair – to swoop in and repay up to 50% of borrowed amount (at the new cheaper price) and get a 5% bonus from Mick! She repays 50% of the borrowed 8,000 USDC = 4,000 USDC for 4.44 Ether (4,000 / 900), plus gets a bonus of 0.22 Ether – for a tidy packet of 4.66 Ether. Fair play to you great aunt Mary!

Off of 10 Ether of initial collateral, Mick has now lost 4.66 Ether – pretty harsh for a 10% drop in the price of Ether per USDC. Ouch!

So, what’s happening here? Basically, the platform (Aave) is making it highly punitive for borrowers to go below their borrowing limit and incentivising them to have a lot of collateral behind their borrowings in case prices fall. This mechanism is designed to make it safer for lenders to lend, but it’s not entirely safe!

Sudden and large price movements

Let’s say Ether drops to 200 USDC. Mary can now repay 50% of Mick’s borrowed 8,000 USDC for 40 Ether – but there isn’t 40 Ether of collateral for her to take – there is only 10!!! There is insufficient collateral backing borrowings!

What happens next depends on the lending platform. Some platforms – like Aave – use a governance token to provide reserves, or partial reserves, against this kind of a scenario. Depending on the severity of price movements, those reserves may or may not be enough to save the day. Ultimately, if there is not enough collateral or reserves, it’s the lenders that lose out – that’s the risk they are getting paid for.

So now that you see one of the risks of lending (and remember there is always smart contract risk as well – the risk that the code makes a mistake) – let’s try out some lending.

*Lending on a DeFi Platform

*Navigate to https://app.aave.com/deposit or to https://app.compound.finance/ and connect with your MetaMask wallet. Note that you can use Aave with a Coinbase wallet – Aave’s documentation is also more clear.

*Select USDC (USD Coin) and then deposit/supply $10 of USDC.

*You’ll be prompted to select your transaction speed (duration). If gas fees are very high ($20+), you may wish to come back another time and hope they are lower. Yes, gas is very expensive right now.

*You’ll now start to earn interest on the USDC that you have supplied. (On Aave, you’ll be given aUSDC in return, on COMPOUND you’ll receive cUSDC).

One last question then…

How are interest rates set?

As a concrete example. Let’s consider that there are lenders lending 100 Ether and borrowers borrowing 50 Ether. That ratio – of borrowing to lending – is called the Utilisation Factor.

The platform (Aave or Compound) puts together an algorithm that aims to increase the Utilisation factor to a certain level (a level that is different for each crypto). They do that with an interest rate curve that pays more interest as the utilisation increases – which encourages more lenders to come onto the platform. Here is an borrow rate curve for USDC and USDT – both US dollar stablecoins:

Note that there are variable and fixed rates – we won’t get into the details for that.

One last point on interest rates – you’ll see that borrowing rates are always higher than lending rates. Check out the comparison for yourself here: https://app.aave.com/markets .

The reason for this is the Utilisation Factor being less than one, i.e. only a portion of assets loaned to the platform are borrowed out by borrowers – and the interest paid by borrowers is shared out among all lenders to the platform (regardless of whether none, part or all of their lendings to the platform are being used by borrowers).

Almost done with Part 3 on Lending Platforms!

So, with DeFi, where there is a service (e.g. currency exchange or borrowing), there is also a complementary earning opportunity (e.g. providing liquidity to exchanges or lending). These earning strategies are also known as farming or yield farming.

Taking this a level further, you can imagine a platform that splits money between different earning strategies – somewhat like what a hedge fund does in traditional finance. That’s what harvest.finance and yearn.finance do – and we’ll cover those in Part 4.

First, take a look at the quiz below before moving on to Part 4.

Once you’ve clicked Submit on the quiz, scroll up to see your answers!


DeFi Mini Course: Part 2, Currency Exchange

Before You Get Started on Part 2.

In this part, you’ll learn how to swap one crypto for another on Uniswap so you can lend it out on the COMPOUND or AAVE platform in Part 3 of this course.

You’ll need a MetaMask wallet with $100 worth of Ether. Skip back to Part 1 for guidance on this.

This course is priced at $9.99 for a four part series. If you find the course helpful, but don’t have the money to spend right now, I’d appreciate if you could instead share the course on social media or with two friends. If you’re new to crypto, you’ll need to read through Part 1 and buy some crypto before clicking the Buy button.

Disclaimer: Crypto is still in its infancy and very high risk. I recommend setting the expectation that you will lose all of the money you invest as you’re learning and treat it as an education. Throughout the mini-course, I will indicate the cryptos I hold myself, so at least – if you do copy me – I’ll lose money if you do! This course is not investing advice, it is intended to teach you how DeFi works! You are reading these materials and taking any guided steps at your own risk. Crypto is technical and funds are easily lost if you are not careful with passwords as well as sending and receiving money to/from the right wallet addresses – not to mind bugs in nascent software platforms or bad actors. Learn with small amounts to minimise your risk.

Part 2, Currency Exchange

DeFi is like a bank, but you can participate in being the bank 🙂

In the same way that banks make money by offering currency exchange and lending services, you can participate in DeFi and earn money just as a bank does.

This part of the mini-course will walk you through a currency change platform (UniSwap) where you can either exchange currency OR earn crypto by supporting the currency exchange mechanism.

Exchanging Currency on UniSwap

You will exchange some of your Ether for USDC (US Dollar Coin), so that you can later lend it out on Compound or Aave in Part 3 of this course.

US Dollar Coin (USDC) is what’s known as a stable coin. Each USDC has it’s value backed by a real US dollar that is held in reserve. Stable coins make it easier to transact with crypto because – unlike with Bitcoin – their value is fixed relative to a real/fiat currency and does not fluctuate. USDC is one of the more popular stable coins (others include Tether or Celo USD) and it is one that is supported for lending on Compound, as well as earning a good interest rate, which is why we will choose it for now.

*Swap $10 worth of Ether for USDC

*Navigate to app.uniswap.org and connect in using your Metamask wallet.

*Select Ether and USDC and adjust the amount of Ether you’ll swap until it gives you an output of about $10 of USDC. It should look a bit like this:

There are a few costs to keep in mind before you press Swap:

  1. Transaction costs (gas) – which will depend on how busy the network is. Probably will cost you a between $3-$15. Yes, expensive! We’ll discover a partial solution to that in Part 4 of the course!
  2. Liquidity Provider Fee of 0.3% – paid to those facilitating this exchange (learn how to earn that in the next section!).
  3. Price impact – the larger your trade, the bigger the difference between the exchange rate you get and the current exchange rate. For trades as small as we are doing here, this will be small.

*Click Swap and you’ll be directed to choose your transaction speed, which affects your gas price. There’ll be a link to Etherscan where you can see your transaction status.

You now have USDC that you’ll be able to lend out on Compound or Aave in Part 3 of this course.

How can you be the bank with UniSwap?

Alright, so you’ve seen how to use UniSwap to exchange crypto. You can alternately make money by facilitating this exchange. As a concrete example, let’s think about how you could facilitate (as the bank) the EtherUSDC trade that you just did – known as “being a liquidity provider”.

At the time I did the transaction above, the price of USDC was 0.0008 Ether. The way Uniswap works for liquidity providers is that you contribute both Ether and USDC to a pool of funds (known as a liquidity pool) in the same ratio as the current exchange rate between the two cryptos. For example, you could contribute 0.0008 Ether and 1 USDC to a an Ether-USDC pair liquidity pool. [We won’t actually do this because we’ll waste too much gas for these small amounts invested, but you can do so using your Metamask wall here: https://app.uniswap.org/#/pool].

The reward the liquidity pool gets for providing that liquidity is 0.3% of each transaction submitted to that pair’s liquidity pool, distributed among pool providers according to the value they provide. That’s how you make money as the bank!

A few further notes on that 0.3%. This is 0.3% of each transaction submitted to the pool – not 0.3% of the pool. To figure out what your return is on lending to the pool you have to divide the daily transaction fees received by the pool by the total amount of funds in the pool – and then annualize that ratio to get an annual rate. UniSwap does that for you, and you can see the instantaneous interest rates being earned on different pair pools here: https://info.uniswap.org/pairs .

Which leaves us with one more key question – how is the exchange rate determined and updated on Uniswap? The answer is market forces… read on.

Calculating Exchange Rates on UniSwap

During a transaction on a UniSwap liquidity pool, the ratio between the two cryptos must obey the relationship x * y = constant, where x is one crypto and y is the second.

For example, in an ETH-USDC pool, the product of the number of Ether times the number of USDC must be constant during a transaction. What does that mean?

Well, lets say there are 8 units of Ether and 10,000 units of USDC in a liquidity pool. Now, I – as someone who wants to exchange Ether for USDC come along with 11 Ether. The formula the exchange must follow is:

8 * 10,000 = (8 + 1) * (10,000 – z), where z is the amount of USDC I get for my 1 Ether.

Solving for z, I get back $1,111 USDC.

There’s a subtle but beautiful point here, the more of a currency I need to exchange, the worse of an exchange rate that I get from the pool. This is by design because it makes it expensive to manipulate/attack the price of the pool by doing sudden exchanges. I realise I’m not giving a precise explanation of an attack here, but hopefully you can grasp the high level point that the system is designed to make attacks expensive.

How exchange rates are maintained on UniSwap

The question you should now be asking yourself is how – if someone can swap one currency for another using a pool – the pool doesn’t get imbalanced and move away from the exchange rate in the broader crypto market. For example, could someone buy up all of the Ether using USDC?

In fact, the exchange rate does move! But, market forces move it back to equilibrium. For example, if the pool moves to 1,100 USDC per Ether on Uniswap but you can find an exchange rate of 1,200 USDC per Ether on another exchange platform (e.g. Sushiswap), then an arbitrage opportunity opens up and participants will buy Ether on Uniswap and then sell it on Sushiswap. Yes, the platform names are hilarious.

The bad news – and yes there is bad news for liquidity providers – is that this process of arbitrage is a cost to the liquidity pool. According to the x*y=constant formula, the larger the value of the constant (i.e. the size of the liquidity pool), the less the effect on exchange rate of a transaction of fixed size. This loss to the pool is called impermanent loss, and is a risk of providing liquidity to pools. Generally, the larger the pool and the less volatile the cryptos being exchange in the pool, ,the lower the impermanent loss. As a concrete example, exchange pools involving stable coins like USDC will generally have less impermanent loss than exchange pools involving Bitcoin or Ether (not stable coins).

My perspective on Uniswap

I don’t own any Uniswap governance token. I’m not an expert on Uniswap but my initial sense is that I like the platform and the incentives are community driven. There is the option for 0.05% out of the 0.3% liquidity provider fee to be redirected to the protocol at some point in the future, so potentially the UniSwap token might accrue value from that in the future. For now, the token seems focused on governance and I think Uniswap is providing a useful exchange service to the crypto community.

Further Technical Notes – for the Quiz Bonus Section!

Uniswap Pool Formula – Actually, it’s not x * y = constant. The constant increases slightly on each transaction because there is a transaction fee of 0.3% of the transaction amount.

GOVERNANCE – Many DeFi platforms, such as UniSwap and Compound, have what’s called a governance token – often named after the platform (e.g. UNI and COMP). This governance token often has a form of voting associated with it that allows the platform to propose, accept and reject changes to itself. Sometimes the token purely serves as governance. Sometimes the token also earns a share of transaction fees on the platform and/or is issued to participants in the platform to incentivise certain behaviours (e.g. providing liquidity to a certain pool). We won’t delve too much into this until Part 4 of this course where I’ll cover Harvest.Finance and the FARM governance token.

A note on fees – What we are doing here buying small amounts of crypto (e.g. $10) doesn’t make financial sense because any return that might be earned lending out this small amount is easily be dwarfed by the cost of gas. With improvements to the Ethereum protocol, gas prices should go down over the next years. For now – it’s necessary to invest larger sums of money in DeFi to get a return beyond gas fees. There is a partial solution to that – pooling resources with other market participants – and we’ll get to that in Part 4 of this course.

Almost done with Part 2!

You now understand how exchanges work on DeFi…and you have some USDC ready in your wallet. Take the quiz below! If you pass, you’re ready to move to Part 3 and lend out your USDC on Aave or COMPOUND!

Once you’ve clicked Submit on the quiz, scroll up to see your answers!


Wifi that stretches for miles


  • Helium provides a communication network using hotspots that people install in their homes.
  • As a hotspot owner you earn helium tokens – a cryptocurrency – for providing coverage.
  • There are now about 5,000 hotspots in the world, providing full coverage in most major US cities.
  • So far, the Helium network has been used for tagging/tracking devices and also smart water meters, among other applications.

What is the Helium Network?

The easiest way to think of Helium.com is as a communications network – just like for mobile phones, but with two key differences:

  1. There are no phone/internet masts providing coverage. Coverage is provided by regular people plugging a device (called a hotspot) into their standard internet router.
  2. Each hotspot can cover miles in distance. This is less than radio, but a lot longer than wifi.

People like you and me (yes, I did buy one) buy a hotspot, plug it in at home, and then are providing coverage to the network. At the time of writing, there are about 5,000 hotspots active in the world:

4,995 Helium Hotspots on January 11th 2021

Why would you run a hotspot?

If you have a hotspot you get paid for providing coverage and also for transactions that happen on the network. Surprise surprise, you get paid in cryptocurrency – called Helium – worth around $1.25 at the time of writing.

I myself bought a hotspot for $349 back around October 2020 and have had it running since then. Right now, I’ve earned roughly 450 Helium tokens, so I’ve made about $600 by now providing coverage. If you’re interested, you can track how many tokens each hotspot is making at https://www.sitebot.com/helium/hotspots/. Some of these hotspots are making absolute coin (over 20,000 Helium tokens from one hotspot in Cambridge, MA that got in early on the platform) – particularly in places where there enough hotspots to communicate with a few others but not so many that the rewards start to get shared too much.

These high returns are temporary and will reduce as more people buy hotspots. Right now, hotspots seem to be back ordered as far as the end of March 2021 or later.

Furthermore, much of the Helium tokens being earned are due to what is called “Emissions” – the euphemism for printing money! Like Bitcoin, there is a specific schedule during which Helium (HNT) tokens are minted over time and given to those who provide coverage. Over time this reduces, and hotspot owners become more reliant on users of the network paying them for transactions. Which brings us to the next important question:

Why would anyone use the Helium network?

The Helium network is low bandwidth, which means you can use it to send something like a text message, but not video content. This makes the network good for small devices that need to be low cost and use low amounts of battery. The best example is tracking devices (could be for lost items, or dogs) or passive devices like weather stations or water flow meters – particularly those at a low price point that need to run on very low battery.

Many of you may know Tile.com – a trackable tag you can put on anything and then find using Bluetooth. Well, Helium have developed their own device called Helium Tabs, which is basically the same thing, but not limited by bluetooth. Looks pretty good on paper (I haven’t tried them):

Features of Helium Tabs

Is the Future Bright for Helium?

The Amazon Risk

The biggest risk I see with Helium is someone like Amazon competing with them. How? Amazon already sells doorbells and Alexa devices to homes that are connected to internet. Simply by adding additional frequency bands to the hardware, Amazon could very quickly build a network across pretty much all cities in the world… Amazon is already doing this in a way via their Sidewalk program.

Why would Helium succeed?

Helium founders (e.g. Amir Haleem – an eSports guru, Sean Fanning – Napster founder, and Sean Carey) and investors (Khosla, Google Ventures) are experienced, so that’s a plus.

The other angle Helium is pushing is the decentralised or “The People’s Network” benefit over Amazon (where Amazon would control it’s hotspots, even if it’s your device). There are a few points worth noting on this:

  1. It is always a delicate balance between allowing a network to be decentralised and also being a for profit business involved in that business. Facebook had a challenge building the Libra cryptocurrency, where it tried to set up a quasi-autonomous governing body.
  2. There is a risk that companies will start to buy up networks of Helium hotspots. There are chat groups already about putting together larger and wider spanning hotspots. Ensuring incentives are aligned towards growth and sustenance of the network is challenging.

All of this said, I am glad that Helium are doing what they are. It is a very interesting example of a network, and of using a decentralised and crypto approach to running the network. I’ll certainly be hodling my helium tokens and keeping my hotspot running.

Parting Questions to Keep in Mind

  • How will Helium Tabs perform versus Tile?
  • What impact will Amazon sidewalk (or similar) have on Helium?
  • How will the price of Helium tokens evolve over time (so far, it has been reasonably steady)?
  • Will hotspot ownership continue to be distributed or will it get concentrated into the hands of fewer people/groups?

P.S. www.Fleetspace.com is also worth a read and they combine low power wide area communication with a satellite approach.

If you enjoyed reading about this project, and would like to follow more projects that you can try out for yourself, you might be interested in subscribing to the newsletter:


DeFi Mini Course: Part 1, Wallet Setup for DeFi

About the Mini-Course

This is a learn-by-doing mini-course in decentralised finance (DeFi). The idea is to let you explore how decentralised finance works and how to make (or lose!) money in decentralised finance. Each part of the course will include:

  1. A short introduction to the theme.
  2. DIY sections – marked with a (*).
  3. A short quiz to test your learning.

Course duration: 4 X 1 hour sessions (including DIY activities & quizes).

What is DeFi? Decentralized finance (commonly referred to as DeFi) is an experimental form of finance that does not rely on central financial intermediaries such as brokerages, exchanges, or banks, and instead utilizes smart contracts on blockchains, the most common being Ethereum.

Who is the course for? If you already have some Bitcoin, this course is probably about at the right level. You don’t need to know code, but this course is somewhat technical and hits an intermediate level. You’ll also need to be willing to spend $100 on this course in order to complete the transaction costs in the DIY steps.

Mini-Course Outline:

  1. Part 1, Wallet Setup for DeFi.
  2. Part 2, Uniswap, a Decentralized Currency Exchange.
  3. Part 3, Lending Platforms.
  4. Part 4, Harvest.Finance, a Decentralized Hedge Fund.

Course Pricing:

I’ve priced this course at $9.99 for a four part series. If you find the course helpful, but don’t have the money to spend right now, I’d appreciate if you could instead share the course on social media or with two friends. If you’re new to crypto, you’ll need to read through Part 1 and buy some crypto before clicking the Buy button.

Disclaimer: Crypto is still in its infancy and very high risk. I recommend setting the expectation that you will lose all of the money you invest as you’re learning and treat it as an education. Throughout the mini-course, I will indicate the cryptos I hold myself, so at least – if you do copy me – I’ll lose money if you do! This course is not investing advice, it is intended to teach you how DeFi works! You are reading these materials and taking any guided steps at your own risk. Crypto is technical and funds are easily lost if you are not careful with passwords as well as sending and receiving money to/from the right wallet addresses – not to mind bugs in nascent software platforms or bad actors. Learn with small amounts to minimise your risk.

Part 1, Wallet and MetaMask Setup

DeFi is built on Ethereum. What is Ethereum? What is Ether?

As of January 2021, most decentralised finance (DeFi) is done on Ethereum, which you can think of as a software platform. You may think that Ethereum is a cryptocurrency, so saying “on Ethereum” may be somewhat confusing to you. In fact, you can think of Ethereum as a software platform (kind of like windows for cryptocurrencies), and Ether is the native cryptocurrency of the Ethereum platform.

In order to do operations on the Ethereum platform – for example, move crypto from one wallet to another – you need to pay for those operations in what is called “gas”. The currency that you pay gas in is called Ether, which is why Ether is considered the “native” crypto currency of Ethereum.

The units of gas needed depend on the transaction complexity. For example, a simple transaction might use 21,000 gas, while a complex transaction might use over 1,000,000 gas.

Finally, because one Ether is worth quite a lot, it is common to use the unit of gigawei or gwei, which is one billionth of one Ether. The price per gas at the time of writing (Jan 2021) is about 40 gwei. It depends on the price of Ether and also on how busy the Ethereum network is. Users of the network (you!) set the price per gas you are willing to pay, and miners are incentivised to choose transactions that pay more per gas. So,

  • If the price of Ether (in USD) goes up – then the gwei is more valuable to miners, so price per gas goes down.
  • If the network is being used heavily, then price per gas gets bid up.

You can check instantaneous gas prices here: https://etherscan.io/gasTracker.

It’s all a bit confusing, but this calculation helps out for a sample transaction cost:

= 100,000 gas * 40 gwei per gas * 1e-9 gwei per Ether * 1 Ether per $1,200

=$4.8 in cost for the transaction

So, to do operations in Decentralized Finance, you need to have some Ether (or gwei), let’s buy some now!

*Set up MetaMask and Buy Some Ether.

To buy some Ether, you need to have somewhere to keep it aka, a wallet. What is most handy is to use a wallet that can interact with a browser. In blunt terms, you will run DeFi software in a browser tab, and need to be able for the browser to push and pull money from your wallet. One easy to use wallet that can do that is MetaMask.io , which is what I use.

*Head over to MetaMask and set up a wallet (either on your phone or browser). Make sure to keep your password and your wallet seed phrase in a password manager and/or written down safely.

*Buy $100 worth of Ethereum. Metamask will allow you to do that. I recommend a very minimum of $100 to get through this mini-course because you’ll need to cover gas costs to try out the DeFi platforms. Note: I hold a similar level of Ether to pay for gas costs. I don’t hold Ether as a long term asset. [If you experience issues using Metamask to buy crypto from a UK or Irish bank account, an alternative option is to create an account on Coinbase, purchase Ether, and then transfer that Ether into your Metamask wallet.].

*If you still need to pay for the mini-course, you can do that using Metamask to send crypto to the wallet address presented when you press the Buy button above.

*Lastly, in your browser (I use the Brave browser, but Chrome will also work), go to Extensions (brave://extensions/ or chrome://extensions) and then find & install the Metamask extension. This is what will allow you a browser tab to interact with your wallet.

Cryptos that are Built on the Ethereum Platform.

Ether is the native crypto of the Ethereum software platform. However, it is possible to build other cryptos using Ethereum. In fact, there are many such cryptos, including the following cryptos that we’ll take a look at in this mini-course:

  • Uniswap, a governance token for the Uniswap decentralized crypto exchange platform, i.e. you can swap one crypto for another (more on that in Part 2).
  • Aave or Compound, governance (and reward) platforms allowing you to loan or borrow cryptocurrencies to others (more on that in Part 3).
  • FARM, a governance (and profit share) token for a crypto hedge fund (more on that in Part 4).

Back to Metamask for a brief moment… Metamask is a wallet that allows you to hold Ethereum based cryptos (e.g. Ether, Compound, Farm, Uniswap). Bitcoin is not built on Ethereum so you cannot buy or hold Bitcoin using metamask.

Further Technical Notes – for the Quiz Bonus Section!

wBTC – There is actually a way to use Bitcoin on the Ethereum platform, and that’s called wrapped Bitcoin. The idea behind wBTC is to have an Ethereum based token that is backed by one Bitcoin for every wBTC token that is in existence. This allows you to transact in Bitcoin using Ethereum software. [Technically, wBTC is an ERC-20 based token – which you can think of as a software subset of the Ethereum platform.]

Buying non-Ethereum based tokens – If you do want to buy non-Ethereum cryptos like Bitcoin, you can use a service like Coinbase (my referral link earning you $10 if you sign up and invest $100 in crypto). You could even use Coinbase to buy Ether and then transfer it to your Metamask wallet. You may get a better exchange rate using Coinbase than Metamask (using Wyre) but you’ll then have to pay for gas to get your crypto from Coinbase to a Metamask wallet. For a $100 purchase, it’s probably easier to buy in MetaMask – unless you trust Coinbase more.

You’re now ready to engage in Decentralised Finance!

You now have a suitable browser (Brave or Chrome), you have a wallet that can engage with browser apps (Metamask) and you have $100 worth of Ether, which we will use to buy other Ethereum based cryptos, and also to pay for gas (sigh….) as we move crypto around.

Take the quiz below! If you pass, you’re ready to move to Part 2:

Once you’ve clicked Submit on the quiz, scroll up to see your answers!


A Simple Explanation of Why Home Antigen Testing is Effective


  • Isolation based on COVID symptoms is helpful and important, but far from perfect.
  • Contagion can happen before symptoms or without symptoms.
  • An inaccurate test (say, 30% false negatives) is likely more accurate than isolating based on symptoms.
  • Cheap quick (15 min) antigen tests accurately detect contagiousness (often 95%+).
  • Governments should make emergency allowance for antigen tests that are authorised for clinical use to be used at home. This would be much better than isolation solely based on symptoms.

The Problem:

Government policy is for people to isolate at home if they have symptoms. However:

1. People are contagious before the onset of symptoms.

2. Not everyone isolates because not everyone gets strong enough symptoms.The isolation based on symptoms policy is helpful, but far from perfect. For every person isolating, there is likely at least one other person unknowingly spreading.

A Solution:

> Cheap (less than 10 euro each) antigen tests are already available that are already EU approved for clinical use.

> These tests are effective at detecting contagion and can catch cases that are missed if you wait for symptoms.

> These tests could easily be self administered to give a result within 15 minutes – you can see my demo video here: https://ronanmcgovern.com/2021/01/08/i-tried-out-a-covid-antigen-test-for-myself/.

Simply by issuing an emergency authorisation for self-use – the government could allow these tests to be sold and used in Ireland. This would be a big improvement over isolation based on symptoms.

Research backs this up. There are US studies from Harvard and Colorado Boulder (here) showing that cheap and quick home antigen tests can greatly slow down the spread of covid in a population.


Test sensitivity is secondary to frequency and turnaround time for COVID-19 screening. https://advances.sciencemag.org/content/7/1/eabd5393/tab-pdf

Further Reading and Reference: www.RapidTests.org .

If you have concerns with the accuracy of any statements or arguments in this article, please e-mail me (you can do so by replying to my email when you subscribe) or by commenting below.


Alfa Laval Acquires Sandymount Technologies

As of December 31st 2020, Alfa Laval has acquired Sandymount Technologies.

The full press release is available here: https://www.alfalaval.com/media/news/investors/2021/alfa-laval-acquires-unique-technology-to-strengthen-its-position-in-sustainable-beer-production/

I have been able to work with many people since the idea for Sandymount started in a lab at MIT back in 2014. Thank you very much to:

  • The Sandymount Team – past and present, including advisors and board members, including those who helped formally and informally.
  • MIT – Prof. Lienhard’s laboratory, the Technology Licensing Office, the Translational Fellows Program, The Trust Center and MIT VMS – including mentors.
  • Our Clients at Sandymount who supported us through this journey – financially and through learning. I think that Alfa Laval will be able to provide you with even higher quality, more complete and better supported Revos solutions and I am very happy about this.
  • Our Suppliers and Service Providers, who helped us to scale from lab systems to commercial solutions.
  • Sandymount’s Investors, who believed in the vision for more sustainable beverage supply chains.
  • My Family and Friends who lived the startup experience with me.

I’m looking forward to helping Alfa Laval take advantage of the full potential of Sandymount’s technology, and – in time – to new business adventures.

Sláinte = Wishing you all good health!


What alternatives to Facebook and Twitter?


  • Whether you think Twitter/Facebook should ban users or not, the crux of the matter is that Facebook/Twitter have a lot of influence.
  • Mastodon.online provides a more decentralised version of Twitter.
  • The most robust option is for everyone to have their own website and to link websites together in feeds. This isn’t fully slick, but it’s a smart step from a business or audience building standpoint.

Are there good reasons to get off Facebook and Twitter?

Let me go through a few angles and philosophies that I see on this question. Hat tip to my friends Ste and Greg for discussions and ideas on the below.

A. Getting de-platformed.

If you’ve built up a big following (which I haven’t!), then getting thrown off Twitter is pretty annoying. Personally, I don’t think that I’ll get thrown off Twitter or Facebook, nor will many others. Although maybe I’m being complacent…

Clearly Trump getting thrown off Twitter is not arbitrary given his contribution to the US State Capitol being over-run by a violent mob. At the same time, it’s hard to find a rule – consistently applied across the Twitter platform – describing why Trump was thrown off. It is sad, but a reality, that there are many Twitter accounts inciting violence – including accounts of major national figures and governments.

My position – and maybe it is a naive and hopeless position, although I think not – is that the problem is in having a small number of platforms (aka Twitter, Facebook) with large reach. The pragmatic solution I see is in having more platforms. This is my review of some alternatives to Twitter/Facebook.

B. Having to pay for ads

From a business angle (my experience running http://www.Point5Brewing.com as an eCommerce business), the power that platforms like Facebook have, is huge. Even if you build up a large following, Facebook completely controls who sees your posts, and has the system set up to encourage you to pay for ads.

As a business, if you can develop your own distribution channel/audience, that’s always going to be better and lower risk.

Alternatives to Facebook and Twitter:

1. Social Media as a Public Utility

There is a substantial Wikipedia article on the topic. I won’t dwell further on the idea because I think that social media is too dynamic and amorphous to make a public utility. Myspace fell apart and was replaced by Facebook. Facebook is now increasingly driven (at least economically) by Instagram. In ten years time, who knows what the landscape will be. The landscape is changing too fast to make it a public utility.

2. A Slightly More Decentralised Version of Twitter

One candidate here is Mastodon.online . Mastodon (terrible name btw) is decentralised in the sense that there are different servers hosting Mastodon (think Twitter) accounts. You can join a server and, if they kick you off, you can go to another server and keep your data and followers (more or less).

I created an account here (https://mastodon.online/@RonanMcGovern) if you’d like to check it out.

One criticism of Mastodon is that the servers/hosts can still be targeted, so you can get shut down temporarily before you move to another server/host (if someone will accept you).

The larger issue with Mastodon seems to be that it’s too similar to Twitter. For a new platform to succeed, it needs to be substantially more unique than what is already out there.

3. A domain based approach to social media

Another approach is that everyone has their own website, and they only use that website to send out blogs or tweets and/or a newsletter. You can see how I’ve set mine up in this way at www.RonanMcGovern.com .

Balaji Srinivasan (www.balajis.com – you see how I did that!) is a big advocate of this approach and even has a guide on how to leave Twitter and bring your followers with you (start by putting your Twitter username as your website).

The benefit of each person having their own website is that it can be fully decentralised. You can control your own hosting of your website and you control distribution (your e-mail list) – you can make it very hard to be taken down. People can also subscribe to your content using your RSS feed (e.g. www.RonanMcGovern.com/feed/ ) using a feed reader (like this one – https://www.feedreader.com/) – so you can also control your distribution. A bit old school – but sometimes old is the new new.

My reservation with everybody having their own domain (read: website) is that it seems a bit clunky and not as slick as Facebook and Twitter. Perhaps if it was attached to a phone number instead of a website, it would be easier.

Conclusion – having your own personal website is a smart business decision.

If you’re worried about Facebook/Twitter having a lot of influence, I think creating and spending time on your own website and e-mail list is the way to go. Tools like wordpress.org make it easier to do this. That said, it’s not a fully slick approach.

If you’re not that worried about Facebook and Twitter, I understand that too. Still, it can be worth checking out personal websites from your friends who do have them and signing up to their feeds/newsletters.

If you’re interested in a monthly summary of my writing, you can just sign up here:

Success! You're on the list.


DIY Self Driving Cars


  • My 2019 Honda CR-V offers limited self driving via adaptive cruise control and lane assist.
  • The adaptive cruise is good – it controls speed while keeping you a fixed distance from a car in front.
  • Honda’s lane assist is pretty terrible, it’s supposed to keep you within the lane on a motorway.
  • For ~$1k, Comma.ai offer a plug and play dash camera that integrates into 92+ cars and provides much improved lane assist, plus a driver facing safety camera.
  • I bought and installed it. Here’s what I found.

The Basics – Adaptive Cruise and Lane Assist

Most new cars now have some level of self driving. For years, cars – especially automatic drive shaft cars in the US – have had cruise control, which allows you to fix the speed at which you drive. That’s a big help, but annoying if there is a car in front changing speed that forces you to disengage and control speed manually.

Adaptive cruise uses radar to detect objects in front of the car and thereby keeps your car at a fixed distance from the car in front. I’ve driven this feature on my Honda CR-V, but also on a Toyota Corolla, and it works very well. It makes motorway driving a lot less tiring.

Lane Assist is a different kettle of fish. For lane assist, there is a forward facing camera in the car that sees lanes on the road, and keep you within the lanes. Honda Sensing does this on the CR-V, but it really only works well on motorways. Annoyingly, it requires you to touch the steering wheel every 5 or so seconds or else it disengages. Moreover, if there is much bend at all in the road, the system loses track and you have to steer the car manually. In all, the benefit of lane assist is very minor with Honda (or Toyota’s) current capability (this article was written on Jan 1 2021).

Comma.ai – the next level

Now comes the DIY approach, which allows you to keep adaptive cruise, improve lane control greatly, plus add some safety features. The solution is a dash camera (the Comma 2), that adds a front facing camera and also a driver facing camera to your car. The Comma 2 also draws from the car’s native front facing camera and radar systems. Lastly, Comma 2 provides input to your car via the OB-II service port under the driver’s steering wheel so that it can control your accelerator, brake and steering wheel.

For regulatory reasons the Comma 2 is not a self driving product. It is “merely” a driver dash cam allowing you to record your journeys. However, by downloading some open source code (Open Pilot), you have the option – entirely at your own risk – of using it to run some pretty sophisticated self driving.

A quick review of the features of Comma.ai (comparing to Honda Sensing as the benchmark):

  1. Adaptive Cruise (adjusting your speed to cars in front) – Comma behaves largely the same as Honda Sensing. At some points Comma is smoother than Honda Sensing in slowing down, at other times it’s the inverse. One slick feature with Comma is that, when your cars slows all the way down behind another car at a stop, Comma will have the car take off again once the car in front moves, whereas Honda requires a prompt by pushing the accelerator.
  2. Lane Keeping – Comma is streets ahead of Honda Sensing on lane keeping. While Honda’s lane keeping frequently shuts off, Comma can take you along street roads – avoiding obstacles – and does very well on motorways. A big advantage for Comma is the driver facing camera – that starts beeping if you look away from the road. This means you can drive for long distances (at least on the motorway) without having your hands on the steering wheel at all.
  3. Safety – I’m repeating myself here a bit, but the driver facing camera adds a layer of safety that isn’t there in Hondas (or in Teslas). George Hotz (founder of Comma) is of the opinion that all cars will need to include a driver facing camera to achieve full self driving safely.

BTW, Lex Fridman’s two podcast with George Hotz of Comma are phenomenal – one and two – the best podcasts I listened to in 2020.

Some worked examples on Comma.ai !

Highway Driving

Here is a video of Comma 2 on motorway driving (not me, someone else on youtube). The one caveat I would give, at least for the Honda CR-V, is that the motorway needs to be quite straight in order for the system not to require you to touch the wheel. It’s not as bad as the Honda Sensing’s native system, which disengages very easily, but Comma Ai does disengage if there is a turn of roughly 20 degrees. I’m not sure if it’s an issue with insufficient steering wheel torque (the self driving system uses power steering to turn the wheel) OR whether the field of vision of cameras is not good enough when there is more bend in the road.

Street Driving in the Honda CR-V with my friend

This video gives a sense of what the car can do on streets from driving with my friend. Note that Open Pilot can definitely not turn around corners, but it can move the car side to side in order to avoid obstacles when driving along a straight street. For most of this video, I have my hands ready to intervene, but I don’t touch the wheel.

Comma.ai vs Tesla

I have driven a model 3 once, but I didn’t use much of the self driving. A few key differences, as far as I understand, between Tesla and Comma’s Open Pilot:

  • Tesla self driving costs $7,500, Comma costs $1,200 (including a connecting cable).
  • Tesla – from a regulatory standpoint – is likely a safer bet. As I said above, Comma is technically only selling a dashcam that you can add your own software to.
  • Comma is limited by the cameras and radar that the average new car has. Teslas have many more cameras – and so they can do things like parallel parking automatically, that Comma can’t do for hardware reasons.
  • As I gather from George Hotz of Comma, Tesla is more advanced (Elon Musk reckons they will have fully autonomous driving in 2021), but they obviously have spent tens of billions more than Comma, which has probably only spent a few million to get where they are.

That’s it for this week. Next week will be a technical review of Helium.com Long-fi systems – long distance wi-fi systems that you can set up in your home, and earn money on for providing the service. You can sign up for my Engineering Business newsletter if you’d like to get an update:

Signing you up…
Kindly check your e-mail to confirm your subscription.

I tried out a COVID antigen test for myself


  • Cheap COVID antigen tests are available with high specificity and sensitivity.
  • They are easy to use and give results in 15 mins.
  • In the EU, they are only approved for clinical use.
  • Government could greatly expand testing simply by allowing people to test at home.

I tried a test for myself (skip to 7 mins)

Demo of a simple antigen test

A few notes on these tests

Performance: Relative specificity of this test is 97.1% and 99.5% sensitivity in a lab setting.

Cost: About 7 euro per test. Could be much cheaper if purchased at scale.

Result: You get the result within 15 mins.

Approval status: Approved for clinical use in the EU. Not allowed for home use.

Why antigen testing makes sense

Read more on RapidTests.org

What are the public health benefits of home antigen testing?

  • Antigen tests are cheap and quick and good accuracy. This allows for widespread testing and earlier detection than having to wait for PCR tests.
  • Antigen tests are sensitive during contagiousness. This means you are less likely to keep testing positive after being contagious (as can be the case with a PCR test).

What are the risks of allowing home testing?

  • Not every test is 100% perfect. There will be a small number of false negatives (i.e. you test negative but are actually positive). We have to balance this against many more people currently not knowing they are positive and spreading the virus. Also, if people test positive at home, that frees up PCR testing (which is currently under too much pressure).

What can you do to help out?

  • Talk to your friends about antigen testing as a way to reduce spread.
  • Get in touch with your local politicians and ask them to push for emergency authorisation for home use.

You can get notified of all updates on my projects by signing up here:

Success! You're on the list.

Review: Sun Tzu – The Art of War

The Art of War by Sun Tzu

My rating: 4 of 5 stars

In this quick read, Sun Tzu shares advice on how to approach and prevail in conflict. Though this book is focused on a time where war was hand to hand, I can see how Tzu’s advice is seen as timeless. Weapons have advanced but human nature remains.

“Humble words and increased preparations are signs that the enemy is about to advance. Violent language and driving forward as if to the attack are signs that he will retreat.”

“If it is to your advantage, make a forward move; if not stay where you are.”

A worthwhile read, if only for the knowledge that a battle should be fought from where it is high, dry and sunny!

View all my reviews on Good Reads


Newsletters are all the rage.


  • Newsletters are in vogue, particularly as companies like Substack have made it easy to set up a newsletter on a freemium model.
  • Many journalists have left traditional media companies in favour of running their own newsletter.
  • Although e-mails are a “basic” format and quite old by internet standards, e-mail remains a highly valuable format for growing and reaching an audience.

My Subscriptions

To cut to the chase, here are the newsletters that I actually read every week:

Lorem Ipsum. This one comes out a few times a week and is a quick read – and free. Written by Margot, it covers food, tech and culture. She has a good take.

Talk Money with Mesh Lakhani. Free and roughly weekly, this one gives a summary of topics and links to click on. Good for pulling out nuggets of money/finance news.

The Diff. Written by Byrne Hobart, this one is a very very deep dive into tech companies and trends. It’s $20/month (aside from a few free articles) and it publishes a few times a week. The content is savage, and very well worth the $20.

Weekend Briefing. Kyle Westaway curates a summary of key news happenings, with a decent level of business focus. The newsletter is weekly. It’s a reasonably long read – which is good – and Kyle tends to pull out some worthwhile rabbit hole stories to go down. It’s free.

Default Friend. Another substack newsletter – this one focused on relationships and publishing a few times a week. No fear in this author to tackle out uncomfortable questions. I know what you’re thinking – like a newspaper advice column. No, not really, this deserves more credit than that. And it’s free.

Contrarian Thinking. Codie Sanchez has this substack and it publishes a few times a week. Codie always has some mad ideas on how to make money through hustling – like buying small businesses or pulling together paid networks. Also, it’s free.

Newsletters allow authors to control access to their audience

In a world of Facebook and Google and Instagram, the good old fashioned e-mail is one of the few ways to build an audience where you can control distribution.

At the simplest level, you can keep a list of e-mails and send them from gmail. You can also set up a mailing list using Mailchimp or MailerLite or many more.

I can’t emphasise this enough, but with an e-mail list you have free access to distributing to your audience as many times as you want! Once someone subscribes (and assuming you don’t piss them off and they unsubscribe), you control that communication channel. I know this is obvious, but compare that to Facebook:

  • You build up followers on your page, but, Facebook controls how many of your followers see your posts.
  • You can pay for ads on facebook, but you have to pay Facebook each time you distribute.

This is why e-mail (and text messages) are huge for eCommerce businesses and for anyone trying to grow a following. (Codie Sanchez above talks about buying and selling mailing lists and newsletters – a potentially lucrative business if done well and without spamming subscribers).

Newsletters as a Software Company

Then there’s Substack.com – they’ve taken the hassle out of getting a newsletter layout and landing page right, and replaced that with a slick ready-to-go package to set up your newsletter – with Substack taking a 10% cut of any monthly subscription that you charge.

As you can see, some folks writing these newsletters are making some serious coin, with thousands of subscribers paying $5-50/month, a very nice side-gig or gig indeed:

Substack top Business Writers

And here are some folks that have ditched larger traditional media (like Rolling Stone (Glenn Greenwald) and The Intercept (Taibbi). Ok, I’ve no idea what the Intercept is, but there was a big fanfare when Glenn Greenwald left it for substack. Who is Gleen Greenwald…?

Some substack folks, including some that ditched traditional media

And of course, to cap it off, if you’d like to sign up for a meta newsletter (actually mostly a newsletter about random stuff, with emphasis on business and technology), here I am:

Signing you up…
Kindly check your e-mail to confirm your subscription.

COVID Notes: Ventilation & Anyone know where to buy home test strips?


  • Vaccines are welcome, but viruses can mutate.
  • Quick and cheap home COVID tests should be a priority for individuals and governments as a complement to vaccines.
  • COVID likely spreads through air – ventilation is critical.

Quick and Cheap Home Tests

It’s good that vaccines are progressing. However, it is possible COVID will mutate, and the vaccine will become less effective. While the vaccine is rolling out, the higher number of infections, the easier for the vaccine to mutate. I see widespread home testing as more socially acceptable than lockdowns, and better for the economy.

It is already possible to make home COVID tests for cheap. They are also accurate – not quite as accurate as a lab – but much cheaper and quicker.

If everyone could test themselves at home each week for a few euro per test, that would allow for early detection and significantly slow the spread – even if the strips tests aren’t perfect (btw, they aren’t far from perfect!).

Please comment below if you have information on:

  1. The regulatory status of home strips tests in Ireland, UK or EU.
  2. Where I could buy home test strips.

Lastly, on this point – a podcast I highly recommend from Lex Fridman with Michael Mina.

The Importance of Ventilation.

There is not full consensus on whether COVID is transmitted through the air or through surfaces. However, it seems likely that transport through the air is dominant.

There is a worthwhile podcast with Dr. Martin Bazant of MIT covering transport of the virus through the air, and you can listen here. I still pay attention to washing my hands, but here are a few other takeaways from this podcast:

  1. If you are outside, your breath typically rises and the air is well mixed, making transmission more difficult.
  2. If you are in a room, the virus will accumulate in the room as people spend time in there.
  3. Masks help (not just to stop spit/spray) but in slowing the rate at which you release the virus from your mouth/nose into the room.
  4. The air in a room is generally well mixed, so staying on the opposite room is not likely to help as much as you would like. Consider someone smoking in the corner of the room – you’re still going to smell the smoke from anywhere in the room.
  5. What will help reduce the amount of virus in the air of a room are: having windows/doors open, good ventilation, or an air filter.

Sometimes it’s too cold to open the windows, so I got a small standup HEPA filter (Amazon affiliate link) (that can filter viruses), with a UV lamp on it. This small unit can change out the air about a few times per hour. It’s not perfect, but I think it’s better than just sitting at the opposite side of the room to someone. I don’t understand why more people aren’t buying air filtration units for their homes.

Comment below if you bought a filter, or if you see things differently on why a filter would/wouldn’t help.

If you’d like to subscribe to my monthly newsletter, here you go:

Signing you up…
Kindly check your e-mail to confirm your subscription.

What is affiliate marketing and how to use it to make money or increase sales?


  • Affiliate marketing allows you to get paid a commission when you link products from your website or social media accounts.
  • Many companies have an affiliate program, including Amazon, that is easy to sign up for. Companies – like Amazon – use affiliate programs as a form of advertising.
  • I set up an Amazon affiliate account and made videos for two products I like to see how affiliate marketing works. Here is what I learned.

Getting set up

Signing up for Amazon’s affiliate program is easy. You go to https://affiliate-program.amazon.com and put in your details. You can then generate custom links to any product on their website and put that link on your website or social media. You also need to put somewhere on your site (I did it on my about page) that you get paid for links to Amazon if there is a purchase. The amount you get paid depends on the category the product is in – it typically varies from 1% to 10%.

Where to share/put the link

This depends on where you are getting traffic. I have a personal blog (that you are reading now), and I also like to make videos. So I have created a web page specifically for my favourite products with a short Youtube video for each of my favourite products.


  • Do provide links to products that you already use and like.
  • Do put links in places that naturally make sense – like in a blog article where you are referecing a book, or in the description of a video where you are doing a review.


  • Don’t link link to random products that you don’t use or don’t like – that will hurt your reputation.
  • Don’t spam people with links in e-mails or by plastering links all over your website.

Example #1 – My video for a Dyson Cordless Vaccuum Cleaner

Example #2 – Bose Headphones

Example #3 – My Favourite Books

  1. Zero to One by Peter Thiel
  2. Risk Savvy by Gerd Gigerenzer
  3. Antifragile by Nassim Taleb
  4. What they Don’t Teach you at Harvard Business School by Mark McCormack
  5. Blue Ocean Strategy by Kim and Mauborgne

How to use an affiliate program to boost sales?

This is relevant if you are a product owner that wants to boost sales. A few options:
1. If you are on Amazon with your product (e.g. Point 5 Non-alcoholic Beer), then Amazon affiliates are already able to promote your product by default.

2. You can join an affiliate network like http://www.ShareaSale.com (I haven’t used this yet but plan to explore it in early 2021) that allows you to get in front of people that might want to promote/market your product.

From a product owner standpoint, the nice aspect of affiliate marketing is that you can set the commission rate so that it is less than your gross profit. This means you can generate added sales that are net profit generators – i.e. no marginal cost! Beautiful.

Let me know if you have any questions on affiliate marketing by commenting below, and consider signing up to my newsletter if you enjoyed this piece. Cheers.

Signing you up…
Kindly check your e-mail to confirm your subscription.


Why Digital Advertising Sucks


  • I have spent thousands of dollars on Facebook, Instagram and Snapchat ads. In some cases they resulted in zero sales, in other cases the cost per sale far outweighed my profit per sale.
  • The effectiveness of digital ads is impossible to measure because there is no way to conduct a randomised trial. AB tests that platforms promote are not statistically valid.
  • Digital ads are presented as having the superpower of “targeting”. It is hard to prove that targeting delivers a lower cost of customer acquisition. It is much more likely that targeting results in prices being bid upwards.
  • The concept of customer life time value is being used to encourage startups to spend more on customer acquisition via ad spending. Credit Jaffer Ali.
  • Where digital ads appear to generate a return, you have to ask the question of whether you would have had those same sales without doing any ads.
  • Many of my ads have been trolled in the comments. Others have annoyed would-be customers who feel they are being spammed. This hurts sales long term.
  • A healthier approach to advertising is a breakeven approach, whereby you breakeven on marketing costs on your first sale.
  • With digital ads it is very hard to break even on your first sale. For now, I believe it is best to consider other forms of advertising such as e-mail advertising.


The following three references are central to my discussion below and I highly recommend each of them:

  • Rory Sutherland’s Alchemy – An entertaining read covering the physchological aspects of marketing. To quote David Ogilvy – ” The trouble with market research is that people don’t think what they feel, they don’t say what they thing, and they don’t do what they say”.
  • Jaffer Ali’s interview on Risky Conversations. Jaffer has an eluminating perspective on the drawbacks of digital ads and typical eCommerce businesses, as well as unique approaches of his own on marketing.
  • Freakonomics multi part series on whether advertising works. Some may find portions of the series too academic, but the questions raised around the value of advertising are good ones.

I believe advertising works, but there may be a bubble in digital ads

As Rory Sutherland likes to say: “A flower is a weed with an advertising budget”. Our emotions can certainly be controlled by what we see and hear. Our emotions affect our purchasing patterns. I accept this and I accept that people would not be spending so much money on advertising if it was useless.

However, it is entirely possible that advertising works, but digital advertising is in a bubble. My direct experience with paid advertising, and my reading of the references above, have brought this to light for me, although I am far from the first.

My Experience with Digital Advertising

To date, I have spent a few thousand dollars, maybe even ten thousand dollars – through hand sanitizer and non-alcoholic beer businesses – on paid ads. I have:

  • Tried Snapchat ads for non alcoholic beer – yielding zero sales for hundreds of dollars in spend.
  • Tried Facebook ads – yielding some sales, but at a cost of over $30 per sale – way higher than the profit per sale.
  • Tried Google ads – yielding no sales for non-alcoholic beer and yielding some hand sanitizer sales (back when sanitizer was scarce) at a huge cost.

Digital Advertising as a Black Box

First off, you may argue that I am not doing it right. I accept that argument, but I am not going to address my own incompetence in this article!

By trying paid ads, it has become clear to me that here are aspects of paid advertising that fundamentally make it hard for advertisers to generate a strong financial return:

  1. The advertiser only has the appearance of having control over how ads are shown. Actual control is in the hands of the platform.

The problem is – you have no visibility into exactly how those targets are being chosen behind the scenes. One week you may hit a certain 10,000 users. Another week you may hit another 10,000 users. I have never run the same campaign twice and achieve similar results. I was like a lab scientist running tests but with no control over the test parameters to make things a fair comparison. Even with the AB tests that digital platforms allow, you can run the same AB test multiple times with different results.

When I go into Facebook (or other platforms), I select a group of people to target. I have access to parameters such as geography or job title or age. Based on these settings, I can define a target group – let’s say of 100,000 people. I then set an advertising budget (say $100 per day). That advertising budget determines how many people – within my target group – the advertisements will hit. The problem is, I don’t decide exactly how these groups are defined, and who within the group will be hit with my ads. The system is a black box that gives the appareance of control but can easily be designed on the back end to optimise for the advertising platform’s revenue.

2. The complexity of the domain in which advertising occurs is vastly underappreciated and is misrepresented.

My second criticism is a statistical one. It is very difficult in a complex system to obtain meaningful comparisons in performance between ad sets. Measuring the performance of a single campaign is easy. Comparing one campaign to another is – in most cases – statistically impossible. Platforms like Facebook are accidentally or deliberately giving the appearance of a measurable process by offering AB testing systems that give the appearance of control over the advertising process. The reality is that there are far too many variables involved in ad performance that cannot be held constant in conducting a comparison.

Targeting does more to increase the price than the effectiveness of ads

Precise targeting of ads creates discrete markets that allow advertising platforms to have advertisers bid against each other and drive up process. This is a clear benefit to the platform

Much less clear is whether a high level of targeting can generate a strong financial return for advertisers. The more I target, the smaller my audience and the higher my advertising costs per sale (especially if I have competition). I need to think more about this but, as I do more targeting, I see myself going up a price curve that is progressively steeper. In other words, I’m getting less bang for my buck in terms of advertising dollars as I do more targetting.

Furthermore, I think targeting is over valued because it is easy to measure performance based on the acquisition costs of your final target audience, but forget about the advertising costs you had to pay for in order to test and find what your proper target audience should be.

Lastly, from the perspective of the advertising platform, they love that I am spending money to find out who my target audience should be – because – once I get to my “target”, they are still the ones in control and can still charge me up to my gross profit in order to give my access to that target! Any advertising monies spent on testing is a sunk cost to me and a sunk profit to them! It’s not like I can have a preagreed arrangement where I say, “I’ll pay $100 to find my target audience but then you’ll agree not to screw me over on pricing once I get to that target. Ok?”. “No”.

Life time customer value as a business school philosphy to drive advertising spending

As Jaffer Ali points out, the “razor – razorblade” narrative pervades business school, corporate and startup thinking. Every investor is looking for the recurring revenue model where – yes, you may have to sell a razor upfront for cheap – but you’ll make it back selling razorblades. Ironically, for the original razorblade company (e.g. Gillette) this hasn’t worked out so well recently given Dollar Shave Club and Harry’s getting on the scene with their cheaper blades.

Lifetime Value or LTV is where you take all of the future value of a customer and discount it back to a present day number. As modern textbook thinking goes, so long as the lifetime value of a customer is signficantly in excess of advertising (customer acquisition costs, CAC), then you are golden. The problem with this narrative, is risk. Summing up the future value of a customer often assumes that the future will look like the present, and it often does not.

For advertising platforms, it is convenient to set risk aside as a real parameter. All that is needed is to push the narrative that their ads provide you with customers with great future value. If you have a customer worth $100 over the next three years – why wouldn’t you spend $50 on Google ads to acquire them today? Easy! Why not even pay $99?

Sometimes no ads are just as good as ads

It’s a very hard argument for a marketing department to completely turn off their advertising. However, without doing this, it is tricky to know whether ads make a difference. In many cases (for example, buying google ads when you are already ranking highly organically in searches), it turns out that you are paying to advertise to customers that would anyway have found you. It’s like giving people a discount after they have already checked out!

Rory Sutherland likes to make a similar point in Alchemy.

Advertising can lose you customers

When I was doing Facebook ads, there was a certain percentage (maybe 20%) that would get trolled in the comments. For example, for Point 5 Non-alcoholic beer, there would be comments asking “What’s the point?” or emoji symbols of puking down a toilet. (The first one I found funny, the second I didn’t like for obvious reasons). At best I might say that “there is no such thing as bad publicity”. At worst, I worry this damages my brand.

Add to that the fact that many people hate ads and use ad blockers. I use mozilla firefox with privacy badger in order to block ads. I also tend to use DuckDuckGo, even though it’s not as good as google. People often don’t like digital ads and tracking, and I feel that some of that sentiment rubs off onto my brand if I use paid ads.

A better approach to advertising

I think it makes sense to think about customers in a long term kind of way when it comes to branding, mission and service. However, when it comes to advertising spend, I’m coming around to the view that it pays to be short sighted. In other words, if an advertising strategy doesn’t pay off very quickly – and it involves spending cash upfront – then I don’t do it. Here’s how I think about it:

  • If I want a long term effect, then spend the money on making great content (videos, art, podcasts) where it isn’t possible to easily calculate a financial return.
  • If I want a shorter term effect, only spend money on advertising to the extent it is breakeven within a short timeframe (30 days).

I take the view that paid digital advertising is overpriced at present and best avoided. I’m glad that I don’t own any Facebook or Snapchat stock any more. Unfortunately I don’t have any easy alternatives other than creating good content, cross promoting with others (businesses, influencers) that are drawn to my brands, and building up a database/sales channel that I can control fully (like an e-mail list).

If you liked this writing, consider subscribing to my monthly newsletter. Please also comment below with your own thoughts on digital advertising and if/where there are areas that you disagree.

Signing you up…
Kindly check your e-mail to confirm your subscription.


What’s the story with Bitcoin?


  • Bitcoin recently passed the $20,000 price mark for a brief moment.
  • The value of all circulating bitcoin is about $371B (at a price of $20,000 per bitcoin).
  • The value of all gold above ground is about $11.5T (at a price of $1,800 per oz.), a factor of ~30X more than bitcoin.
  • If bitcoin even got to $200,000 (ten times the recent price) – approximately half of all billionaires in the world would owe their wealth to bitcoin.


Bitcoin provokes strong reactions – positive and negative. Warren Buffett thinks Bitcoin is a bubble (“rat poison squared”). Meanwhile, large public companies are starting to get on board with Bitcoin – Square recently bought $50M worth and Paypal are planning to add Bitcoin to Venmo in 2021.

Personally, I think Bitcoin could go to zero, but it could also go much higher. As I’ve written before, I hold ~5% of my liquid net worth in Bitcoin.

Passing $20,000

Bitcoin jumps around in price a lot and – like any speculative asset – people just love when it crosses a good round number, like $20,000! $20,000 is roughly the mark that Bitcoin hit in 2017. This past week, bitcoin went slightly beyond its 2017 record and reached an all time high.

How much Bitcoin is in the world?

There is a maximum supply of 21 million bitcoin. These have now almost all been released into circulation. At the time of writing, there are about 18.5 million bitcoin in circulation. At a price of $20,000, that implies that the value of all bitcoin (known as the market capitalisation or market cap) is about $370 billion.

How much gold is in the world?

Given gold is a financial store of value that is not profit generating in itself (in the same way that a business or real estate is), it provides one comparison for bitcoin, which is – like gold – limited in supply and not profit generating in itself.

As of the end of 2019, about 200,000 tonnes of gold are above ground. At a price of $1,800 per Troy ounce, that implies a market cap for gold of about $11.6 trillion.

So, the price of bitcoin would have to rise by a factor of about 30 in order for the value of all bitcoin to be about equal to the value of all gold. That’s a tall ask, but I think it is at least conceivable that bitcoin could increase more in value. [It’s also conceivable that it could go to zero.]

What if Bitcoin went up by 10X in price (say, to $200,000)?

Interestingly, as Balaji Srinivasan (@balajis on Twitter) points out, half of the billionaires in the world would be “Bitcoin Billionaires”. That’s kind of crazy to think that Bitcoin could re-write the wealth hierarchy of the world.


People love a good round number, hence the recent hype. Expect the same hype to play out again multiple times in the future.

My recommendation is to learn about Bitcoin and how it works for yourself. Then you can make your own decision. Nataniel Popper’s Digital Gold is an ok place to start, as is Wences Casares’ episode on Econtalk.

If you liked this article, you may like to sign up for my monthly newsletter:

Signing you up…
Kindly check your e-mail to confirm your subscription.


Who are You?

“Hi, my name is, what? My name is, who?
My name is, chka-chka Slim Shady”

Eminem / Slim Shady


  • Bio-metric verification (iris, face scans etc.) is one (likely) direction for the future of security and passwords.
  • There are risks around being identifiable by personal traits, and some of these risks may be mitigated by an increased use of pseudonyms.
  • Pseudonyms allow accountability but can be tossed if your data gets hacked or profile gets cancelled.
  • Pseudonyms are already common on Reddit, Twitter and on eBay – and (loosely interpreted) with Bitcoin. Perhaps they will become more common in the future as a form of protection of identity while allowing for accountability at some level.

Yourself as a form of verification:

It is ludicrous that we still use credit cards, passports and social security numbers as forms of verification. Why not use bio-metric information, such as iris scans or facial recognition? Why are these not more widespread?

Countries like China are going in this direction, as are companies like Clear – previously for expedited airport security clearance, but now more generally to replace your credit card. You don’t bring a credit card, you bring yourself!

While the benefits for convenience are there, bio-metric approaches raise the obvious concern of data security. Once someone has your iris scan, it’s not as though you can change your iris scan as easily as you can change your social security number. Having a data breach like Equifax is bad enough, imagine how much worse it could be to have a data breach of bio-metric data – especially as the data being recorded gets more and more detailled.

Accountability and risks in being yourself

Having people be accountable for their actions is something that I feel strongly about. There are few things more frustrating to me than working with an anonymous “Simon B.” at Amazon, who gives a half answer to my first question, knowing that it will be “Mary P.” who will be on the hook for the next follow-up.

Putting your name on the line in business and in life is something that I see as admirable – and that can be seen in many of the great brands today that are called after their founder or inventor, such as Wolfram Mathematica – after Stephen Wolfram.

Conversely, when you do put your name on the line, there is the risk that your character can be torn down. This disincentivises bad behavior, but assumes that the mob is always wise in deciding who to tear down. We know that truth is not always the victor on the battleground of social media, so putting your name out is not always wise – particularly if wading into politicized battlegrounds.

Pseudonyms as an alternative

What then, might be a solution that allows for accountability while also providing some protection against hacking of personal information and against bad actors/mobs?

The answer might lie in pseudonyms, where you hold a secondary name for public activities. That secondary name or identity can develop a brand and reputation of its own, and even an income of its own, but – if it is torn down or hacked – damage can be limited to that secondary level by starting a new pseudonym. Painful but not life ending.

Pseudonyms are already widely used on Twitter and have huge followings without their underlying account owners necessarily being known:

@mathtick @rudyhavenstein and @wrathofgnon are some examples of pseudonymous tweeters

However, the vision for pseudonyms can be much wider, with these pseudonymous characters earning income and taking on much fuller personalities than before. Credit to Balaji Srinivasan for this thinking and here is a great video exploring the concept:

Pseudonyms – back to the future

I suppose – to some degree – the reason we still have credit cards is because they are somewhat pseudonymous. The number of each card is a secondary identity you can have, and you can cancel and throw it away if there is fraud. That part works well – it’s just that hauling around a physical card is annoying.

Centralised or decentralised?

A credit card – while it is a pseudonym of sorts – is a centralised pseudonym because the card issuer defines the card numbers and . A decentralised pseudonyn could be a bitcoin wallet or password, where you generate these numbers yourself (with the wallet address publicly visible and the password/key private to you alone). So, pseudonyms can be centralised or decentralised in how they are used.

The question of centralised or decentralized is important because it affects how the information could be used for law enforcement and also how vulnerable the information is to hacking. In a centralised system, all of the information can be used to identify bad actors. With a centralised system, there is also the benefit of having a lot of information – so this information can be used to better understand the system – call it “artificial intelligence” or “machine learning”.

On the other hand, a centralised system is more vulnerable to hacking, i.e. you just need to break into one point to get all information (very roughly speaking). Plus, if law enforcement is a bad actor, that presents a risk too. So decentralised systems can be better for privacy and protect against abuse of power.

Bio metrics are communist while pseudonyms are libertarian?

To get this paragraph title, I’ve hacked up Peter Thiel’s view that “artificial intelligence (centralised deep information) is communist while crypto (decentralised, pseudonymous) is libertarian”. (Parentheses are my own)

An increased use of biometrics means putting personal (permanent) information in the hands of large organizations (governments or tech companies) with centralised control. Pseudonyms can be centralised, but they can also be decentralised and they can be self generated – so they can be more libertarian.

My view is that the increased sharing of personal information is inevitable and may have some benefits, but we should be looking at alternatives – where possible – that allow for accountability but provide better protection from data breaches, bad actors and cancelling.

If you like this article, you might also like my monthly newsletter:

Signing you up…
Kindly check your e-mail to confirm your subscription.