We're All Stupid

or, Don't Put all Your Virtual Eggs in One Basket

A Brief Thought Experiment

Imagine a world in which you did everything right, and then lost anyway. You saved money, contributing pre-tax to your employer’s 401k. You never took money out, you compounded your gains, you lived within your means. You bought index funds from Vanguard. You watched your friends buy flashy cars, take on debt, and live paycheck to paycheck. But you? No, you were playing the long game. You deferred satisfaction. You were setting yourself up for life. You did everything by the book.

Then one day, you woke up and it was all gone.

How would you feel?

The crypto equivalent of this happened to a friend of mine, and it should raise some questions about many of the core beliefs in the space.

Ideological Foundations

In the bitcoin community, one of the core principles that many hardcore bitcoiners espouse is self-custody. For those of you who are not crypto people, this is essentially “be your own custodian”, and would be the traditional financial equivalent of keeping your stock certificates locked in a safe deposit box instead of having them with your broker dealer, or maybe keeping all your cash under the mattress inside of a bank.

This is something the modern financial system mostly does not do. There will be a lot of gnashing of teeth, hissing, and screaming from some libertarians about how this is a plot to control everyone and create some sort of financial surveillance state slash techno-oligarchic feudal system, but the reality, as these things often are, is way more banal: actually people were just losing shit constantly.

If you look at the history of financial markets prior to electronification, one of the things that stands out is how big the inefficiency and deadweight loss is. You still see this today with cash. Has anyone here ever lost a stock in their brokerage account? As in, not sold it, not lost money on it, but literally misplaced it? Like one day your 100 shares of Apple stock were there, and the next day, they were not?

That happens all the time with cash. Much of the modern financial order is because of this problem, not some grand conspiracy. It turns out it’s helpful that all of the things be held in places where you can track it, it won’t be lost, and you can settle it efficiently.

However, in the bitcoin community, self-custody is often the default recommendation. Why? Because the community was founded on a deep distrust of all institutions, and a refutation that they might provide any value. In some places, this is probably correct; I am not exactly excited to leave my money in a Venezuelan bank, for instance. However, in many other places, what you are saying here is “I trust myself and my ability to assess safety more than I trust JP Morgan to treat me well”.

Is that a good bet? Many people seem to think so, despite all the evidence to the contrary.

This profile is sort of the same mental error that leads people to support communism over capitalism, despite the former dramatically underperforming the latter on both economic and human outcomes every time it has been tried. Why? The harms are easy to envision (picture a mustache twirling billionaire, and then the poor in the streets), and the gains are hard to envision (picture hundreds of millions of poor children who did not starve to death because of Archer Daniels Midland). Even so, if you take a structured utilitarian view and stack them up against each other, capitalism outperforms dramatically: it does a better job of lifting the poor out of suffering, and it does a far better job of not murdering tens of millions of people for not doing what the government says. In fact, basically no capitalist country has ever killed even 5% of the people communist countries routinely do. And yet people seem to hate capitalism and prefer communism repeatedly? Are people just stupid?

“Yes!” shouts my old behavioral economics professor.

Visibility Bias

Deep in the shadows of the past, I used to work in reinsurance, and one of my favorite behavioral economics findings comes from that industry. I got into the space because the kind of mathematics I was studying became particularly relevant in the post 9/11 era, as it turned out to be the kind of math that was being used to build terrorism risk models. Math is weird!

However, insurance is also a] hotbed of weird outcomes in behavioral finance, because of visibility bias. My favorite experiment is one where travelers were offered trip insurance, covering all sorts of outcomes that could disrupt their travel, for a fixed price. They usually declined. However, when offered the exact same price for trip disruptions due to terrorism, they would much more frequently say yes. Read that again. Terrorism is a subset of what the first policy covered; it is strictly better! You are protected from more things! For the same price!

Yet people would consistently decline that, but accept a more limited policy covering only terrorism for the exact same price? Now you see why my professor thinks people are stupid.

So what is going on here?

The people taking these outcomes are basing them on availability bias: they are scared of a thing they can easily imagine, and so protect against that. This is not a deep probabilistic analysis. You may think your brain is weighing all the possibilities in a matrix and then making a decision based on that, but unless you are quite neurodivergent, you are almost certainly not doing that. You are making a decision based on fear, and then rationalizing your way into why it “makes sense”.

Self-custody is the same thing. More people lose their bitcoin due to theft, due to hacks, or due to simply losing their private keys than lose them due to bad behavior by regulated custodians and exchanges. An important detail here: regulated. Nobody is saying to go trust FTX or some shady offshore exchange. Don’t do that! The track record there is also very bad.

However, if you think a custodian regulated by the NYDFS or JFSA is stealing all your funds, with no procedures in place, with higher probability than you just losing your own keys, you’re doing the terrorism insurance thing.

Coldcard

My friend from the introduction basically did this. The problem was knowable in advance, with certainty, and I say that because I’ve been telling him for years not to do the thing he did, which was keep almost all of his savings in bitcoin, and then keep that all on a single device.

His rationale was simple: institutions are not be trusted, because of both 2008 and the fact that the rich will steal from the poor and the political system in the United States is deeply corrupt. Therefore, his odds are much better holding his money himself than trusting a centralized entity. Decentralization, you see, was the optimal goal.

As an aside, you can tell we are in the terrorism insurance problem with that phrasing: if he truly believed that, he would have been decentralized in terms of assets and holding locations. Decentralization to him actually meant: “extremely centralized somewhere else, away from the problem I am imagining”, and the entire framework was backward rationalization to do what he wanted to do in the first place: take his money out of the US financial system.

So what happened? He did a bunch of research on this, and followed the gold standard recommendations in the Bitcoin space: a cold storage device, kept in a secure location, where it was hard even for him to access, and would likely be safe from any disaster. Only one other person, a parent, knew about the existence of this device. That device was also, unfortunately, a Coldcard device.

My friend did not do the coldly rational thing: diversify across gold, bitcoin, cash, stocks, and bonds, at a variety of institutions so even if one fails, many of the others are likely to survive. No, he was maximally centralized: one asset, one provider.

Trust Assumptions

Bitcoin is founded on being “trustless”, so the great irony of this situation is that it turns out you need a whole lot of trust to use it. In this case, my friend had placed his trust in the Coldcard device and designers, which if you have been following the news on this at all, turned out to be a terrible idea.

To paraphrase quickly, bitcoin is safe because of cryptography, and that only works if you do it properly. If you generate properly random key phrases, you’re great! But if you don’t, if they are easy to guess or replicate, someone else can steal your stuff. Guess which one of those Coldcard did, because of something that is either a bug or a deliberate insider exploit to allow theft? That’s right, the bad one.

So there has been over $100mm of BTC stolen from these devices, and more will be coming. It is the largest self-custodial hack I’m aware of at this point in time, and it raises an important point about the difficulty of actual self-custody, which is simple: unless you are doing the cryptography yourself, you are trusting someone. You’re trusting someone different than if you were trusting a regulated financial entity, but you are still trusting someone. In this case, you were trusting programmers who turned out to be some combination of incompetent or corrupt, but the outcome is the same as leaving a ton of money above the FDIC limit in a rural bank that fails: you are screwed.

The core problem with all of this is the terrorism insurance problem. That is to say, the bitcoin people were operating on religious beliefs and fear rather than deep, rigorous analysis. If you had actually been trust minimized, you would have been using open-source code, compiling it yourself, keeping it across multiple devices, and in what is probably a fatal error for the satoshi’s witness belief system, keeping it across multiple assets.

Lessons in Anti-Maximalism

I’m not writing this to dunk on these people. I’ve been warning for years that self-custody is less secure than institutional custody for most people. I’ve publicly stated the bitcoin that I own is held in ETF form, not in any personal wallet, because I trust Blackrock, Fidelity, Franklin Templeton, etc. to do a better job of this than me. Am I making a trust assumption? Yes, I 100% am. And I might make a different one if I was a political dissident in Russia, but being a college professor in the United States, I think this is probably the right call.

However, there is a bigger lesson here beyond “don’t trust a device you don’t understand”, and that is about actual trust minimization and decentralization.

Every point at which you concentrate things is a form of trust. That can produce outsized gains (if you own the one asset that goes to the moon, as many early bitcoiners did, you will be quite happy), but it also creates outsized risk. Your volatility is concentrated in both directions. So you need to understand what you are doing, and why you are doing it. You also need to step back and ask how well you understand things, which is hard. So here are some useful principles I use to troubleshoot this:

1 - If you have to assume the largest asset you hold goes to zero, what kind of shape are you in?

2 - If you have to assume the largest holding container you have is compromised, what kind of shape are you in?

3 - If you have to assume that your main form of making payments is disrupted, what kind of shape are you in?

In short, if you have a single point of failure, what happens if it fails? What is your central expectation in those cases?

As you can see from my friend, he fails on two of those (1&2). This is the folly of the bitcoin maximalist: by distrusting the system so much, you have created critical reliance on your alternative path, and if something happens, you are fucked.

If you want to make that choice with eyes wide open, I’m not going to stop you. You can. You should, if you truly understand it and think it’s the right one for you. My advice is simple: make sure you understand that.

So if you are a self-custody or bitcoin maximalist, you need to ask the three questions I said above, and you need to be comfortable with your answers. You don’t get to assume a bad thing happening is not possible. Bad things happening are always possible. Trust me, the world is endlessly creative about bad things.

Do your actual research, which means the first thing you need to research is yourself, your internal biases, the decisions you are making due to availability of mental images of risk, and the assumptions you have embedded.

Or, put in the words of my old behavioral finance professor: “Accept you are stupid, and work with that. We all are. I am. You are. Everyone is. If you deny it, you will make mistakes. If you accept it, you can begin to be less stupid.”

Harsh, but fair.

And if you know a bitcoin or self-custody maximalist still doing this stuff, send them this article and ask them to honestly answer those three questions. They probably won’t, but you will have at least made an honest effort to help.

A final note: most of my work now lives over at Zero In. You can find it here.