Update: new 1 year poll is live! This poll https://forms.gle/A8iPjmNX4ZQz8zht5 asks participants: "What do you believe was the probability that SBF committed willful fraud in the FTX collapse?
Resolution
In 1 year (November 2023), I will create another instance of the same poll. Resolves to the average (mean) poll response at that time.
Motivation
In the immediate days after the FTX collapse, a lot of people have quickly concluded that it was willful fraud (e.g. that FTX intentionally loaned out customer deposits which they had committed not to). With the limited information available early on, I believe there is substantial but far from conclusive evidence in favor of this, and that the evidence and information is rapidly changing, and that many people are jumping to conclusions too quickly. You can find a copy of a post I made on this topic at https://manifold.markets/post/in-defense-of-sbf#jIGvMdkFDKDltmegirwJ tl;dr: FTX's failure to hold customer deposits as fully backed deposits seems like it could very plausibly be a combination of bad judgement, terrible accounting practices, and mistakes rather than fraud. I'm not claiming it's likely, just claiming that it's not beyond a reasonable doubt.
This market is an attempt to quantify the degree of certainty the community has on this question. Many people have been acting as if it is basically 100%, while a seemingly smaller minority believes it is substantially less. This is a type of prediction quesiton, and of course we can resolve different predictions by making a market on it.
Notes
Poll responses must be honest. Do not attempt to manipulate the poll - if I suspect manipulation then I reserve the right to restrict the poll to reputable Manifold users or resolve N/A.
I am aware that this poll would likely be highly subject to selection bias. I am considering some ideas to mitigate this in another variant of this question; let me know if you have suggestions.
Poll means a literal poll where you answer with a probability, it does not mean a market.
I chose 1 year somewhat arbitrarily. I could also create a version of this that settles when I or the community judge that no more significant evidence is likely to arise.
Related
1,000
3.00The responses to the poll averaged 92% (median 95%). For comparison, the poll 1 year ago averaged 76%, median 95%. https://manifold.markets/jack/how-much-do-we-believe-sbf-committe
However, this time there were far fewer responses (only 6) than last time. So I can't really make interesting histograms or do that much analysis. But it is apparent that people became much more confident in fraud over the last year, which makes sense.
It's been 1 year, the poll is now live! Please respond here: https://forms.gle/A8iPjmNX4ZQz8zht5
The poll will run for two weeks, so it closes December 5.
Today there was a description of how Alameda had eight different spreadsheets to hide their liabilities, each with different lies about their actual assets and liabilities. Sam directed Caroline Ellison to pick one of the non-truthful ones. The spreadsheets were evidence in court.
I mean maybe there is an explanation that does not involve Sam directing someone to lie to investors but it’s hard to imagine that.
https://twitter.com/molly0xFFF/status/1710718416724595187
Claiming to have an insurance fund of a particular value, but using a random number to determine the claimed value, is remarkably unsubtle fraud; any claim you're literally generating randomly is certainly not one you actually believe.
It isn't as if you could write it by accident, either. Insofar as you think Nishad Singh was following SBF's direction in implementing the "functionality", this would show an example of willful fraud.
It feels like EA is punishing SBF not for being unethical, but for being unlucky.
This is probably where I disagree the most, it's totally unacceptable to handle customer funds this way, and I would codemn it as strongly in a counterfactual world where this somehow all worked out. I don't think I have a clean way to explain this if you don't already have this intuition, but it's the financial equivalent of the greatest possible violation of trust. In the case of the Lehman Brothers, no customer funds were ever at risk and all funds were returned to customers. Stealing, gambling with, and losing customer funds is not only fraud but fraud of a level far beyond the fraud most people usually condemn. As far as accounting, I find it very difficult to simultaneously believe that FTX International was a successful, competently run business, and that SBF believed it was legitimate to list the assets FTX held to cover customer funds as mark to market when they were the primary holder of those assets and responsible for the vast majority of the trading volume of those assets. It should not be possible to make this sort of mistake in good faith, and it's the responsibility of the fund to ensure that these sorts of mistakes are not made.
But if you believe the latter — tell me, how much did you know about crypto, exchanges, or trading firms before last week?
Since I'm being asked, probably top tenth percentile? Not a whole lot, but I was certainly familiar with this type of fraud before FTX collapsed, it is very much classic fraud. It's not impossible that SBF sort of accidently stumbled into committing fraud totally in good faith, but it does seem to me like a pretty absurd claim which would require a whole lot of updates on my part about his intelligence, knowledge of finance, personal ethics, or mental state. Again, putting yourself in a position where it's possible for you to fat-finger billions of customer funds is not acceptable. You are required to not put customer funds at risk. Even if the actual loss of funds was somehow totally done in good faith, the actions leading up to it are still unethical and should still be condemned.
Risk-taking and ambition are two sides of the same coin. If you swarm to denouncing risks that failed, you do not understand what it takes to succeed.
I think one of the most credible and reasonable criticisms of Utilitarianism and specifically longtermism is that it becomes very easy to justify certain means that most people would find repulsive if taken too naively and too literally. 0.00001% of an astonishingly big number is still an astonishingly big number, and so what is $9B of customer funds but a pebble on the road to the long term future? Personally, I like to take my utilitarianism with a side of virtue ethics.
For what it's worth, I don't think much or any of the criticism from with EA is motivated by their changed financial situation. If EA has a track record of anything, it's continuing to do their best and reason in good faith even when large sums of money are involved. Fairly rapidly EA has moved to directing an order of magnitude more funds and it doesn't seem to have led to any notable corruption or dilution of morals. The whole thing is unfortunate and I'd generally prefer to let the trial end before criticizing anyone too heavily. My predictions, though, are 99% willful fraud, and that most of the people who have a fraud probability of under 90% will update significantly after the trial. No plea deal offered means the prosecution thinks they have an absolutely slam dunk fraud case, and I generally believe that the people in charge of prosecuting high profile fraud cases have access to the most information about the case in question and are generally competent and knowledgeable about fraud. Finally, while it's unfortunate and painful, I think it's necessary to criticize people who do things that are unethical even if you like and care about the person in question.
Addendum: I last looked at FTX's books when this happened so my memory may not be exactly clear on this front, the central point is that the mishandling of customer funds was so bad as to be unethical even if it was unintentional, and it was in fact so bad that it defies belief that it could have been unintentional. I'll probably read back into it once the trial starts, but again I expect that the trial will make most things clear and there won't be much ambiguity left to read into.
@NathanpmYoung Agreed, I also mentioned concern about selection bias earlier.
One thing that potentially helped with the current poll is that it was featured so that anyone on Manifold who was interested in the subject had a chance to see it and respond, instead of only people who were already following the market.
@NathanpmYoung The intent for this question was a poll open to anyone, as per usual Manifold poll practice. If you have ideas for what sort of select group might give better results, I'm interested in hearing them.
@NathanpmYoung That would be a potentially nice feature but I don't think it makes a huge difference - the poll here is already getting a decent number of responses, and I can gather more information this way than with a Twitter-style poll.
@NathanpmYoung There are two common ways I've seen to do polls like this that are lighter weight and don't require new Manifold features:
Poll by liking comments: https://manifold.markets/IsaacKing/at-the-beginning-of-2023-will-manif
Poll by embedding a 3rd party website poll: https://manifold.markets/Sinclair/which-hogwarts-house-is-the-best-po
But I'm not convinced getting more responses improves poll accuracy. In fact it might make it worse. I think to help mitigate selection bias, it would be better to get a random sample of the Manifold community (probably limited to those with some familiarity with FTX), and poll them.
I have committed not to reveal the current mean until the current instance of the poll finishes at the end of January.
Firstly, every question should have a wiki for info and a way to tip people who add it.
Here is my case for about .92
I don't know what the definition of fraud is, but SBF allowed money which customers had specifically earmarked to not be risked to be risked. And it was $8bn. That's not an amount you can do by accident. What's more Ellison's testimony said they were doing this longer than we previously thought. How can we argue at this point that this either wasn't fraud or wasn't willful?