Michael Lewis (2023)

An absolutely riveting account of the rise and fall of Sam Bankman-Fried, his Alameda Research hedge fund and FTX exchange. the book came out before the trial at which Bankman-Fried was found guilty of fraud other financial felonies.

The person who emerges from this account is … strange. That’s not all that unusual in tech industries: as a computer science academic I encounter lots of people at varying points on the Asperger spectrum. Like many such people Bankman-Fried has a hard time relating to others and to his own feelings, and (also like many such people) is competely open about this and other aspects of his inner world. Unlike others he somehow manages to gain the trust of supposedly sophisticated investors.

Why is this? I suspect it’s got something to do with the archetypes of the Boy Genius and the Great Man of Science: people who are misunderstood but can nevertheless succeed beyond normal measure. That’s an attractive hook for investors wanting to profit off unlikely accomplishments.

Part of Bankman-Fried’s strangeness was his embrace of effective altruism and the philosophy of an analysed and quantified life. The basic premise seems to be that the probabilistic techniques that work so well in finance can be applied to all life decisions. The problem (I think) is that this is a mistake, and lots of pseudo-mathematical talk about “expected value” doesn’t disguise that: but it does throw up a smokescreen for people who aren’t themselves clear about what the maths actually says. They allow Bankman-Fried some astonishing lattitude, playing video games on reporting calls and simply not turining up for events because something else “with a higher EV” showed up. These aren’t characteristics of a reliable business partner – and would label someone with a more normal affect a sociopath.

(Why do I think the talk about EV is a mistake? In order to make an expected value calculation one needs the value of an event and the probability of that event happening. The value might be available, but what about the probability? How do we assign probabilities to events that lie a long time in the future, and whose course we don’t understand? A one-in-a-billion chance, for example, sounds impressively quantitative. But it actually only refers to three independent one-in-a-thousand events happening together. What are those events, how did we assign their probabilities, and are there really only three of them? If we don’t know all of this information, then the numbers are simply speculative, and only sound quantitative. That’s not to say that these calculations can’t be done – finance people do them regularly – but they only work under conditions of substantial information and substantial control over the environment.)

Lewis is also quietly scathing of many of the characters who emerge around FTX. The bankruptcy CEO, John Ray, comes out badly, as do the corporate lawyers. And there are some uncomfortable questions. Why did FTX file for bankruptcy in the US when it had no presence there and didn’t deal with US citizens? Why were so many senior people offered plea deals? Why were some not indicted at all?

The trial (which has happened at the time I’m writing) brought up some contradictions with Lewis’ account. He clearly takes Bankman-Fried at his word a lot of the time, especially with regard to his inner calcultions, but also with the naturalness of his unkempt image. Testimony from those close to him suggest this is naïve, and was subject to a lot of delierate design. It’s an open question whether the trial has achieved a more accurate view of Bankman-Fried’s psychology than Lewis did, though, and this book deserves to be read in that context.

5/5. Finished Sunday 19 November, 2023.

(Originally published on Goodreads.)