Long Take: The mistakes we made analyzing FTX, and how to correct them
Confirmation bias, cognitive dissonance, private information, and now deep fakes
Gm Fintech Architects —
Today we are diving into the following topics:
Summary: We attempt to diagnose the mistakes made in thinking about FTX by the industry and ourselves. How was everyone tricked into such a disastrous opinion? What made investors drop real diligence and write enormous checks into the company? We talk about the distinction between public and private information, as well as confirmation bias and the traps we all fall into confusing the map for the territory. Additionally, we should the difficulty of diagnosis specific outcomes arising out of complexity, even when you know the component pieces.
Topics: capital markets, crypto, psychology, investment management, complexity, deep fakes
Tags: FTX, Twitter
If you got value from this article, please share it. Long Takes are premium only, and we need your help to spread the word about how awesome they are!
Long Take
Emergent Complexity is bigger than any one mind
You might have seen this uncanny video.
A deepfake SBF, simulated by a machine learning algorithm that has been trained on the various media appearances of the now disgraced FTX CEO, is posted on Twitter the social network. The video claims to provide compensation to those who lost money on the FTX exchange, but is of course just a phishing scam to steal more of your money.
The open source algorithms that have powered up immense creativity in generative AI are used here to defraud retail investors, who have already been defrauded.
The account tweeting this out has a blue verified checkmark, fooling people into thinking it is “official”. This is the case because Elon Musk has arbitraged the desire of people to be seen as notable by allowing users to buy verification for malicious bots and various catfish scams. The arbitrage will go away as people understand the blue checkmark actually means nothing now, but confusion still lingers.
The attempt by Musk to re-architect Twitter, in some warped attempt to liberate free speech from advertising pressure, has resulted in impersonation and scams.
Is this outcome predictable? Obvious? Or is it unpredictable, emergent from the complex interactions of weird frontier technologies and unrelated events? AI that can simulate human appearance and voice. The collapse of a particular crypto exchange, and the ability of crypto assets to be digitally native. The hubris of a particular billionaire in buying Twitter, while the markets collapse from macroeconomic pressures, making tech companies historically cheap.
These are all dependent vectors emerging from our society fractal, and we can and have mapped these general trends for over 5 years. Permutations of thinking about AI, agents, authenticity, how the Internet is largely fake traffic, Musk’s motivations, all of this has floated across the newlsetter pages before.
And yet, predicting a particular symptom of the system is impossible. We can know the general types of outcomes that comes out of complexity, but we cannot know the exact one until it happens. In retrospect, however, all the dots causally connect, and we can say “of course!”. Of course A —> B —> C.
The millions of other permutations and probabilities fade into the background and become invisible and irrelevant, once we actually observe what has happened. It is the quantum wave function collapse.
And so it is with FTX.
The Mistake
It’s all so embarrassing. Just earlier this year, we were saying this about FTX.
Look how super wrong that is!
FTX was definitely not the winner of volatility, despite acting like one. It was absolutely not well timed in the long run, and its trading strategies — which included a massive exposure to Terra on the Alameda side that eventually led to the unauthorized “lending” of customer funds — was not a smart trading strategy. Perhaps “using retail distribution” was the only answer to subsidize running a bad prop trading shop, but that’s certainly not an aspirational statement. And further, whatever propping up of Solana that FTX was doing through “market-making” has indeed collapsed. The only thing we got right there is the word “collapse” itself.
And not only is it embarassing to had been wrong, but the effects of the overall market holding this set of opinions and beliefs about FTX is exactly what allowed them to get away with inappropriate risk taking, poor controls, and very likely financial malfeasance.
Real damage was suffered by the actual holders of FTX accounts, and further real damage spreads into Genesis, DCG, and Grayscale, as well as many other prime brokers and lenders.
Here’s something much closer to the truth. The restructuring CEO of the now bankrupt FTX entity had this to say —
Oops.
Public and Private information
The first issue to highlight the difference between public information, and inside information. What most of the media and analysts built their cases on is the performance that SBF put up on Twitter, in public meetings and appearances, and the top line financials shared by FTX and discoverable by Google.
Private companies are private. When we approach their analysis, we have to resort to publicly shared figures that are made available in press releases or given as part of articles written during the company’s fundraising. We have to rely on signals, like the work of dedicated institutional investors and their commitments, in order to ascertain how much to trust a particular company.
This is the bonkers part. FTX raised from Temasek, Tiger, Softbank, and Sequoia, among others. These are not bad investors. As far as growth and innovation goes, despite the occasional ribbing we like to give some of them from mis-pricing multiples and blitz-scaling prices from 5x to 100x revenues, these investors still identify the *right* companies that then become industry leaders. You could argue that it doesn’t matter that much if WeWork or Uber equity is overpriced, especially for the average consumer of their services, because the services are transformational and massively useful in the long run. Users are not investors.
So we would expect these investors, who have access to private documentation generated by their targets showing financial statements, legal formation documents, and the rest, to have done actual diligence. It is reasonable to rely on expert signals in a complex, multi-dimensional economy. Having seen plenty of data rooms for venture deals, we know there’s at least a balance sheet in there somewhere.
The first flag should have been the humiliating troll of the storied institutional investors that FTX had trapped.