@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.

@berealistic

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.

@jack

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:

@berealistic I agree it seems to have been sitting too low at 30-50% until very recently. I'm not convinced it should have been as high as 90% that whole time, given the complexities of the network of FTX entities, and the fact that FTX Foundation is not one of the bankrupt entities (although some FTX grants did come from other bankrupt entities); and the many ways clawbacks might not succeed.

Yet if I sold my shares this market would sit at 50%.

I'm pretty sure if you sold your shares now the market would buy it right back up.

@jack You're right they probably would buy it right back up, but I would have hoped that EA had enough collective intelligence to price this correctly (without needing a snarky pseudonymous account slowly buy up shares months after the relevant info was out).

I personally think it should've been at ~90% this whole time since, as far as I know, there hasn't been any new information about this (except for the news about letters being sent out to politicians yesterday).

@berealistic 88% for an event that would require lots of political will to happen (not just to send the letter, as market explicitly calls that out) is "unrealistically low"? You seem to think very highly of bureaucracies' effectiveness. (Although, of course, four years is a long time.)

@b575 What political will is needed for the recovery of funds from a bankrupt company to continue to progress?

@berealistic I've been a small creditor of a large bankruptcy exactly once, so I'm dumb mana on this market. You're welcome.

John J Ray III is expensive and I doubt it is in the interests of the creditors to pay him to go after individuals at several steps remove. At some point the easy money has been recovered, the accountants are rich, and the creditors are mollified.

@berealistic It's low still. One weak case for "no" is that litigation is expensive and the expected recoveries from "individual workers" as defined here are not high enough to justify litigating out individual avoidance actions. For disbursements more than 90 days prior, there is maybe a defense based on the funds running through a charity and maybe a defense on the grantee having given value in good faith under section 550.

But I strongly suspect at least two such workers would settle at 6 percent...

@berealistic "We will persecute individuals even if it's more work than what money we're gonna gain". In other words, as @Jason said.

@Lauro See my comment here: https://manifold.markets/EliezerYudkowsky/will-5-of-an-ftx-grant-be-clawed-ba#2VOLDviqmamunPvBZCY6

@NathanpmYoung I mean it's an individual who got money in exchange for doing grant work, not their host organization. I'm not sure what about this was ambiguous so I don't know how to be less ambiguous for you.

@EliezerYudkowsky That's helpful thanks.

@EliezerYudkowsky I think it was a vibes thing, but it just seems likely that money is gonna be attempted to be clawed back wherever it can be, so your bar is pretty low. so 55% seems like a price to move to.

Based on reading https://forum.effectivealtruism.org/posts/o8B9kCkwteSqZg9zc/thoughts-on-legal-concerns-surrounding-the-ftx-situation it appears that transactions that occurred in the last 90 days before the bankruptcy filing are typically subject to clawback (these are called "preference claims"). However, there are many reasons such a claim might not succeed (see the comments).

Nothing I've read indicates any difference in treatment between organizations and individuals.

Note that this is specifically about grants that came from the bankrupt entities, which notably does NOT include FTX Foundation. However, not all grants were paid out from the foundation (according to a comment on the above post).

The intent is to talk about individuals who got grant funding, including by way of an org, and are now worried that the money they already received will be subject to legal action. An employee of an org that got the grant funding, who opens a letter addressed to the org, is not an intended subject of the question.

OK, thanks!

I'm still not entirely clear on this. To clarify, of these two cases:

• A) grant from FTX-related foundation to an individual person

• B) grant from FTX-related foundation to a non-FTX-related organization, which pays its employees using the grant money

Which of these cases is included in the question? I think your response above suggests B?

Ok, sorry, after rereading I think I understand that you intended to include the case of individuals getting a grant directly from an FTX-related entity, and also including the case where the non-FTX org regranted to an individual, but not including the case of an employee of a non-FTX org.

From most broad to least broad:
Spencer's version of the market is about demand letters (which may not proceed as far as a court case)
Nathan's version of the market is about account freezing (which would require a court case and would certainly imply demand letters were sent), but freezes could be temporary if a court later decides the funds can't be taken.
Eliezer's version of the market is about funds being actual taken (or tied up in litigation past the end of 2026), which implies the previous two.

Actually scratch part of that, Spencer's version is stricter than just a demand letter because it restricts it to a letter from a court (implying at least a motion) or from an FTX entity (which won't happen unless compelled by a court). In particular it excludes demand letters coming from FTX's creditors.

@JamesBabcock I intended letter from creditors to be included as a yes, so I updated the description. Thanks for catching that!

@JamesBabcock Correct, I'm talking about going after individuals.