Resolves YES if at the end of 2026, there was any successful or still-being-prosecuted attempt to claw back, from more than 2 individual recipients of an FTX-originating grant that went through a legal charity along the way, more than 5% of the money received; including but not limited to FTX Foundation, FTX Philanthropy, FTX Future Fund, "North Dimension" under Alameda, the FTX Future Fund regranting program, etcetera.
(This is intended to capture the spirit of "there were not just a couple of exceptional cases" and "there wasn't just a go-try-it letter with no chance of making it successfully through the legal system" and "they didn't just settle for 5%"; i.e., to measure the probability of a scenario that would pose strong danger to a lot of individual people who did work under FTX grants. Resolves YES if there's a protracted battle still going on into 2027.)
Added: Per original title, "individual" means "individual human worker".
1,000
1.00
(Also, assuming that the scenario "person voluntarily settles with the estate before getting a demand letter" doesn't increase the counter of people who would cause it to resolve YES?)
@JonasVollmer Judgment call on to what degree there was a credible chance of the money being successfully forcefully clawed back, vs. the bankruptcy trustees managing to find overly credulous targets who were easily bluffed, or particular EAs taking stands on conscience.
@EliezerYudkowsky Okay. I bet this to 86% assuming something like "if 3 people settle with the estate for 10% of the total grant they received because their lawyers (acting in their best interest) advised them to do so instead of potentially going to court, this resolves to YES"
https://forum.effectivealtruism.org/posts/W7AMKT8qssjS8WwjN/habryka-s-quick-takes?commentId=Fu3CgEF4uiagNbvZ3 but apparently that isn't too relevant to the market
Right, because that’s a voluntary return, not a clawback, right? So why the spike in early 2023? What changed from Nov. 2022, when [Scott Alexander](https://www.astralcodexten.com/p/mantic-monday-twitter-chaos-edition) said, “Everyone is panicking about this possibility, but it looks like it’s not too likely”?
I’m betting No out of sheer curiosity.
@OllieBase They’re recovering the money quickly!
an increase of more than $800 million since January
They reported having $5.469bn in liquid assets on Jan 17th, and $6.107bn at the start of March, so $800mn sounds like an understatement if they now have $7.3bn?
It sounds like they took the January-March increase ($638mn) and multiplied it by the recent appreciation of crypto (it’s up over 20% since March).
Either way that still leaves about an $8bn dollar hole if I’m reading things correctly.
The big news here is that they’re thinking of relaunching the FTX platform, that’s a very positive sign.
(For some reason, I can't comment on the thread by @berealistic and @jack.) I'm not privy to many private conversations on this issue, but there are a few factors that are going to make it harder for a public Manifold market to be valuable here. Many of the people who have the most valuable insights at baseline -- i.e., practicing lawyers -- have likely pulled their punches on the public analysis out of concern that going in too deep could be seen as offering legal advice. I certainly have more opinions than I've actually shared. Second, there are good reasons for the EA community not to have too much of this discussion in public where attorneys for the FTX debtors can see it. I certainly would not want my predictions and strategic advice in my day job as an attorney to leak to the opposing party. That being said, this market was worded in a way that I think buying YES doesn't leak any real information. It's really obvious to me that there are more than two grantees for whom a 6% clawback would be a really, really good deal. The only plausible NO outcomes are that FTX decides the individuals didn't get enough money to bother with (very unlikely), that FTX will not negotiate (very unlikely), or that all-but-one of the grantees decide to hold out rather than take a deal (fairly unlikely).
@Jason Thanks for this comment! I agree with your assessment. One clarification:
or that all-but-one of the grantees decide to hold out rather than take a deal (fairly unlikely).
If they hold out and FTX is still attempting to get the money, it counts towards YES ("successful or still-being-prosecuted attempt to claw back"). If FTX gave up on their attempt, then it wouldn't count.
The other thing that makes me hesitant to bet too high is the fact that some grants came from bankrupt entities, and others came from non-bankrupt entities (most notably, FTX Foundation is not one of the bankrupt entities). I don't know enough about which grants came from where, although I think it's very unlikely that all the relevant individual grants came from non-bankrupt entities.
@jack Not enough money in play in suits vs individuals for me to assign much probability to suits against individuals continuing into 2027. I don't think the identity of the grantee matters as much as others do where section 550 exists allowing recovery against subsequent transferees under many circumstances.
I’m interested in following the quality of people’s epistemics within EA, but should I assume that there’s not much interest in this question? Maybe my judgement is poor and this isn't a big deal, but there’s a lot of relevant information coming out that’s not being shared on here. FTX released lots of information 2 days ago:
In a press release they said "their books and records are incomplete and, in many cases, totally absent" but, from their presentation (PDF download), they seem to have detailed information on the 90 day period from August 14, 2022 to November 11, 2022:
So FTX presumably knows who received funds from a debtor entity during the bankruptcy period. Do people think individual workers got paid before this period, or in particularly small amounts, or entirely from the FTX Foundation (and not debtor entities)?
Huge withdrawals from customer funds started in June, so grants given before June are less likely to be deemed to be fraudulent transfers:
Side point: this is pretty good for EAs to know about, right? I can't see anyone on the EA Forum talking about it
It looks like they've found $0.6b more in liquid assets (now $6.1b, up from $5.5b) but this could just be because crypto is up
It looks like there's a shortfall of around $9bn (unless I'm looking at the wrong numbers) so there's definitely a massive amount to recover
And shortly after the last comments were made on this market, there was confirmation of how expensive Sullivan & Cromwell are (from CNBC on Feb 8):
FTX’s top bankruptcy, legal and financial advisors have billed the company more than $19.6 million in fees for work done in 2022, according to Tuesday bankruptcy court filings. More than $10 million of that was for work done in November 2022
Anyway, is this stuff not shared because it's not deemed worthwhile/helpful? Should I be looking elsewhere for assessing the quality of EA epistemics?
Anyway, is this stuff not shared because it's not deemed worthwhile/helpful?
No, posts like this are helpful! Even if it doesn't directly change my forecasts, it's useful background information at least. I would say primarily just because of the people interested in any question, only a small fraction make predictions, and an even smaller fraction take the effort to post information.
Re "Huge withdrawals from customer funds started in June, so grants given before June are less likely to be deemed to be fraudulent transfers" - I don't think this is an issue because clawbacks are restricted to 90 days before the bankruptcy filing (see https://forum.effectivealtruism.org/posts/o8B9kCkwteSqZg9zc/thoughts-on-legal-concerns-surrounding-the-ftx-situation), which would mean grants given before August are shouldn't be affected.
Should I be looking elsewhere for assessing the quality of EA epistemics?
There's also Metaculus, which has a lot of EA related people and forecasts.
However, Manifold is the only site I'm aware of with forecasts on this particular topic.
I would say you should adjust your expectations to reflect the fact that there is an extremely wide space of questions, and furthermore that nobody is being paid significant amounts to forecast on this question - it looks like the traders with the biggest positions might profit the equivalent of $10-50, maybe. I think epistemics can be greatly improved when there is focused attention, e.g. popular markets or people who are dedicating time to the forecast as part of their job.
I don't think this is an issue because clawbacks are restricted to 90 days before the bankruptcy filing
This is for one kind of clawback (which is very likely to happen) but there are others that can be made outside of that window. From Molly Kovite's post:
The second clawback process (called a “fraudulent conveyance claim”) targets transactions that happened up to two years prior to the bankruptcy filing. The root of these claims is an allegation that the debtor moved assets out of their organization for the purpose of frustrating future creditor claims. These types of claims are more complicated to prove, less commonly brought, and more individualized to the specific transaction.
My suspicion is that people reading that will immediately think this is unlikely to apply to funds sent to EAs (because it seems extreme or something).
But yeah they shouldn't think that because the only clawbacks we know about so far are of this type; many of the donations being clawed back from politicians were sent well before the 90-day period (including some from 2021). And the amounts are often less than $3000: https://unusualwhales.com/politics/article/senate_ftx
@jack And thank you for your comments on epistemics. I should clarify that I'm not really interested in whether people associated with EA can form accurate beliefs on arbitrary questions; the project of 'doing good better' and the prospect of a community of people that can actually solve important problems is what's interesting to me. That involves asking the right questions that are decision-relevant, not just performing well at making lots of inconsequential forecasts.
nobody is being paid significant amounts to forecast on this question - it looks like the traders with the biggest positions might profit the equivalent of $10-50, maybe.
I have a lot of thoughts about this, but will try and keep it short:
This market reflects the community's beliefs on something that they should intrinsically care a lot about (this goes for the other markets I'm active on too)
I'd expect those with accurate beliefs to share the info they have, as it seems like a pretty obvious public/club good when you have 100s (?) of Future Fund grantees uncertain about what to do (presumably)
Here and Molly's Forum post appear to be the only places where this community are sharing info with each other, yet not much visible info sharing is happening
Plus, isn't this the value proposition of Manifold Markets? They've received ~$2m in EA money because info is deemed to be very valuable, but maybe I've misunderstood the point?
It seemed like EAs valued the info being communicated on here pretty highly (over the value of equiv. charitable donations) since there didn't seem to be a huge sell-off of Mana when the end of the charity program was announced? Not sure the financial argument really makes sense; seems like people aren't in it for the financial reward of directing money to their favoured charities.
Accurate beliefs here seem very important for reasoning about funding allocation, project planning, and making sensible financial decisions
If the incentives are important and it's also an important question (which maybe it isn't?) then why aren't the incentives better? That seems like a flaw in EA's epistemic abilities.
@berealistic I wasn't aware that clawbacks for political donations had been attempted before that period, interesting. My impression had been that grants to honest charitable orgs aren't fraudulent conveyance claims, as far as I understand. It would be something like: if SBF made a grant to a friend and this was done just for the purpose to pay the friend some money without the friend doing work of "reasonably equivalent value", I think that would be fraudulent and subject to the fraudulent conveyance claim, but they'd still need to prove it. But of course, they can send a letter to try to claw back money from earlier, even if they have no expectation of succeeding.
@berealistic I agree with many of your points, but I think my main point in response would be: Markets aren't magic, they can serve two major purposes: quantifying the information people already have, and providing incentive to research more. On this question, I would claim it is almost all the former. The amount of incentive provided for the latter on Manifold is currently very small relative to the amount people were already interested in this question (of course, this ratio can look very different for different questions and different prediction platforms).
I think those people are probably sharing what info they have. I think the baseline amount people would be motivated to research and share info on the topic if the market didn't exist is roughly what we see in Molly's posts, and I think the market only increases the motivation to research the question a relatively small amount. I think the main thing the market does do is provide a place to share updates, and try to quantify the information that the community already has.
I'm not saying that people are trading only based on the value of their mana as charitable donations. I just used that to give a rough proxy of how much incentive there is to trade on the market for profit alone. I would expect most of the people predicting here to be mainly motivated to do so because they were already interested in the question, and therefore the market mainly serves to quantify the information that has been researched.
I do think the quality of the predictions on this market was disappointing. In my earlier comments here I said that I didn't see any reason for individuals to be treated differently in terms of clawbacks, and I bet a small amount accordingly, but I didn't know enough about the topic to bet more. There are easy ways it could have been improved, with more investment of research effort. E.g., I suggested that looking at precedent (aka base rates) would be the best way to get a good prediction on the question, but that would take research effort that apparently nobody did.
Accurate beliefs here seem very important for reasoning about funding allocation, project planning, and making sensible financial decisions
Manifold has run some forecasting projects like https://manifold.markets/group/clearer-thinking-regrants/about and https://manifold.markets/group/predict-cep/about which I think produced reasonably accurate predictions.
Making questions that are decision-relevant and getting enough forecasting interest on them is tough, but certainly something that Manifold and Metaculus are working on.
@jack Yeah that perspective makes sense, but aren't the politicians also honest recipients? This normative perspective - that Future Fund grant recipients don't deserve to have their grants clawed back - might be true, but also seems like wishful thinking that might prevent clear thinking about the real consequences. People like to think that their legal system is always fair and just, but my impression is that sympathetic reasoning like this has limited sway over due process.
Also, I think FTX is (by default) not receiving 'reasonable equivalent value' when they give out donations, because it's a gift rather than a service they're paying for? Idk how fraudulence gets proven or not though, and I've ended up looking into this stuff when I was trying to just observe how the community approached these questions..!
@jack RE markets/forecasting/info sharing in EA: I think I'm just looking for something in EA that isn't there. But thank you for sharing your thoughts, I appreciate it!
@jack I realise I misunderstood you when you said:
My impression had been that grants to honest charitable orgs aren't fraudulent conveyance claims, as far as I understand.
and that you think this applies to the politicians too. I think FTX does in fact expect to get that money back (even if it was received before the bankruptcy period) because they deem them to be fraudulent transfers. I've made a market to try and target this: