Summary
The EA and AI Safety communities have broadly condemned the actions of SBF and FTX, and expressed regret that they find themselves in possession of funds that seem to have been misappropriated from FTX users. However, at the same time, messaging within the community about the need to return these funds is somewhat mixed. I argue that this mixed messaging fails to take the correct stance on this issue, and risks significant instrumental harm to the community.
I write as an outsider to the EA community, although I'm fairly familiar with the philosophical foundations (have read a lot of LW, GW, Peter Singer for the last ~15 years). I've never used FTX and have no financial stake in this.
Comment from Eliezer
Important notice to readers. Please vote up even though it is not very carefully argued here, because it may be important to some readers to read it immediately.
DO NOT FOLLOW THIS POST'S ADVICE. IT IS PROBABLY VERY BAD ADVICE FROM A LEGAL STANDPOINT. IF IT DOESN'T GET YOU IN TROUBLE IT WILL ONLY BE BECAUSE PEOPLE IGNORED YOUR LETTERS.
NEVER FOLLOW ADVICE LIKE THIS FROM PEOPLE WHO ARE NOT LAWYERS.
ONLY DO ANYTHING REMOTELY LIKE THIS IF YOU READ A POST FROM OPEN PHILANTHROPY'S LEGAL COUNSEL TELLING YOU TO DO IT.
I've edited this comment in as I can see he thinks it's important that people see this notice. I've also edited in a comment in the section that I think he's referring to.
Regarding Eliezer's November 12 post
Probably most of you have read this November 12 post by Eliezer on the subject of clawbacks: https://forum.effectivealtruism.org/posts/FKJ8yiF3KjFhAuivt/impco-don-t-injure-yourself-by-returning-ftxff-money-for
I would summarise this post as primarily addressing the question of FTX funds that have already been spent, with particular reference to the situation where the funding was "spent" as compensation for services the grantee has already rendered. The post argues that if you provided the services, you shouldn't view yourself as morally obligated to return the funding. It's clear this comes from a place of compassion, based on knowledge about the likely mindset of many blameless people who suddenly find themselves in a difficult position.
However, the post is a nuanced take on a nuanced set of circumstances. It's basically the "Level 2" reply to the "Level 1" conversation that Eliezer assumes people have already had with themselves. The problem is that I look around now and I don't see any clear statement or consensus about the much more basic case, addressing funds which haven't already been spent.
Clawbacks are good actually
It was bad that funds were misappropriated from FTX users, and it will be good for as much of that money to make its way back to its owners as possible. Any unspent money should be set aside until the relevant bankruptcy proceeding can request it. I suggest also writing to the new FTX CEO to let him know the funds are being set aside in this way. I think it would be especially welcome if this could be done in an organised, collective way. Edit: I agree that this part was worded poorly; it's definitely not a good suggestion to just go ahead and do that without any kind of advice or consultation. Rather what I'm hoping for is that the community can be proactive about cooperating with the bankruptcy proceedings, rather than taking the stance that the more money that can be retained, the better. I don't want to offer any specific advice about how that should be carried out.
Leaders of the EA community have repeatedly denied that their ethical philosophy entails an extreme act utilitarian "ends justify the means" position. For instance, William Macaskill was repeatedly clear on this before the FTX revelations, and he's been asked it almost incessantly ever since.
The basic argument is that attempting to define ethics strictly over individual actions ("act utilitarianism") leads to worse consequences than defining ethics over rules or policies ("rule utilitarianism") instead. There's an epistemic case against act utilitarianism (the temptation to fool ourselves) that I think is very compelling, but I think the strategic or game theoretic case applies especially well to these events. We all need to live in a society --- there's very little utility to be found in the state of nature. In order to live in a society we have to agree to cooperate with each other, and agree to withdraw our cooperation from, or even actively punish, defectors. Naive act utilitarianism simply isn't compatible with this. Individuals can't be expected to agree on individual act-based decisions, and you can't form a coalition with people who are making it up as they go along.
A basic respect for property rights is a good ethical rule, that we will all be much worse off without. Support for earning to give in no way entails support for stealing to give.
There are lots of situations, either constructed or actually occurring, where it's unclear how to reason about the priority of different rules or policies. The actions of SBF are not any sort of boundary case. You can't run a society where it's okay for your trading partner to abruptly decide it would be better for the world if your goods were sent elsewhere.
The need for the funds to be returned to their owners therefore completely overrides any considerations of what good those funds would be doing for the purposes SBF tried to allocate them. If it wasn't good to steal the money, it must be good to return it if you still have it. You can't just continue on with some prior plan to deploy the funds, that you had agreed with SBF or his delegees. That agreement was illegitimate.
Lack of consensus and leadership about returning funding
I view the ethical argument above as following fairly straightforwardly from the stated positions of EA's leadership. However, browsing the forum discussions, the attitude to the clawbacks seems much more mixed. For instance, here's what the FTX FAQ on the forum states about the clawbacks:
If you got money from FTX, should you give it back?
- You probably shouldn't, at least for the moment. If you gave the money back, there's the possibility that because it wasn't done through the proper legal channels you end up having to give the money back twice.
If you got money from FTX, should you spend it?
- Probably not. At least for the next few days. You may have to give it back.
I feel bad about having FTX money.
- Reading this may help.
- "It's fine to be somebody who sells utilons for money, just like utilities sell electricity for money."
- "You are not obligated to return funding that got to you ultimately by way of FTX; especially if it's been given for a service you already rendered, any more than the electrical utility ought to return FTX's money that's already been spent on electricity"
These answers also link repeatedly to Eliezer's post for reference to how people should view the question of the clawbacks. However, Eliezer's post is targeted to the specific situation where the funds have already been spent for services that have been rendered. It is not a general statement about how the community should view the need to return money to FTX. To me this is a good illustration of the lack of clarity in the community about this question.
I find the answers to If you got money from FTX, should you spend it? and If you got money from FTX, should you give it back? particularly inadequate. The answer to the should question interprets "giving it back" very narrowly. I certainly agree that nobody should just transfer the funds out right away. But I would expect the consensus position to be that it's straight-forwardly wrong to spend FTX money that could instead be preserved for its rightful owners.
The advantages of a clear public stance
This is more a question of strategy, so you can take it for whatever you think it's worth. It's an outsider's perspective.
There's obviously the risk of lasting reputational harm to the EA community from FTX. The first and most memorable thing many people will have heard about EA is that SBF stole a bunch of money and this is where some of it went.
Everyone can understand an innocent victim, but being an innocent beneficiary is a much more difficult position to explain. EA's role in unambiguously endorsing SBF and helping to open doors for him makes things doubly awkward.
The EA community will find it much easier to distance itself from SBF's crimes if you're able to say hey, we didn't want this. We worked proactively to get as much money back as we could. SBF gave out $X and we got $Y back --- everything that hadn't already been spent by the time the news broke. We gave it all back voluntarily, even if it would have been hard to track down or claw back.
As I argued above, I think this approach is the right thing to do. But even if you doubt my appeal to rules and principles, I think the instrumental case for this approach is pretty strong as well.
I know that the EA community believes very sincerely in the urgency of the problems it is working on. So I can imagine seeing this as some sort of key turning point. I can imagine seeing returning the funding as a setback that you just can't afford. However, there's no Solutions Store that you just have to make it into with the right amount of cash in your hand. The money is just one step in the long road of working towards solutions.
You can't make any progress on the problems you care about without strong, practical organisations full of people working together in predictable ways. And those organisations will need to be part of a wider coalition. You need to be trustworthy.
I think this post is making some important points, though I think it's overstating the need for urgency. I agree it's very likely a good idea to try to avoid spending the FTX money that you've received, though if you have no choice (since e.g. you have unavoidable ongoing financial expenses and need time to arrange other sources of funding or pivot your organization), I think people should probably spend the money instead of fully shutting down their organization or defaulting on ongoing liabilities they have, which seems overall likely to cause greater harm than to keep operating for a few months until the situation clears up more (though I also think it's a good time to cut spending if you can, and therefore avoid spending the unspent FTX money).
I think it's a good idea to try to make sure that depositors get their money back, though I am not as confident as you seem to be that the clawback procedures are the best way to achieve that. I am currently holding out to see whether I expect the clawback procedures to actually result in an equitable outcome here, or whether we should expect some enormous fraction of the clawback amount to be spent on legal fees, or be siphoned off to e.g. the Bahamas government, or institutional investors beyond their initial investments, which seems pretty plausible to me, and which would make supporting the clawbacks itself a somewhat morally dubious act.
If the clawbacks turn out to be a pretty bad way to actually get money back into the hands of the depositors, then I would like us to spend a decent amount of time trying to figure out some better way to actually achieve that, so I am on board with the overall call for "actively supporting giving stolen money back to FTX depositors".
Separately, I think it's also pretty unclear what fraction of the money that was given out by the Future Fund was actually stolen. It's plausible to me that via various clawback mechanisms (independently of the Future Fund) the vast majority of the funds that were lost to depositors will be recovered, and that only a fraction of the money given out by Future Fund should be modeled as stolen, and where e.g. the fair thing to do is for people to return 20% of the funds given to them by the Future Fund (like, as far as I can tell FTX and Alameda also made a bunch of money and produced a bunch of legitimate value, and in as much as everyone can end up even with only a fraction of the money returned, I think that's a fairer outcome). I have a ton of uncertainty on this, and expect we won't know for quite a while, though I think we should continue to actively try to reduce uncertainty on this as the bankruptcy proceedings go on.
I sincerely care about doing right here, and I want us to seriously try to put depositors and debtors to FTX and Alameda back in the green, but I also think the facts of the situation are genuinely uncertain.
Some other questions I have uncertainty about are:
I think settling this issue will take many years, and I think I want us to put more emphasis on strapping in for the long haul and righting the wrongs that we might have been involved in, than immediate calls to action for people to do things right now.
I also think it's a time a lot of people are paying attention to us, so I don't think it's a terrible time to make commitments to somehow do things right here, but I do have enough uncertainty about how to actually do that that I don't currently feel comfortable making any super precise and strong commitments besides a general "I will really try pretty hard to make things right somehow, even though I don't know how yet".
Thanks for such a thoughtful reply.
I think the right thing to do is to follow the official, legally authorized bankruptcy procedure (whichever that is --- I do hope it's the Delaware proceeding, but I guess we'll see).
I don't think it would be right for people who happen to be in possession of the funds to be making decisions that second guess things like the amount of legal fees. That's a pretty involved judgment call, and trying to set up some sort of alternative, direct solution will be difficult to implement and execute. You haven't been entrusted with any of that responsibility by the people who actually own the money, so I don't think it would be right for you to intervene in that way, even trying to do the right thing.
I can also say for sure that I don't think it would be right to invest the money in any way. It should just be kept in a deposit account. Investments come with risk of losing some of the principal, and nobody entrusted you with the authority to make that risk calculation.
This is just my outsiders guess, and in the end who knows, but...I predict that if the EA community did try to be sort of heroic, and tried to do some direct-for-depositors complicated thing that worked around the bankruptcy procedure, that would end up with a lot of people mad at EA, and some takes along the lines that this all confirms some of the initial criticisms.
Instead, I think the approach that's actually better, and I think also has much better optics, is to just work with the official process. There's a legal process for returning money to creditors and you're expected to just go along with it, rather than trying to invent your own alternative to try to get what in your view would be a better outcome.
And while I'm not a lawyer, I think you might be pleasantly surprised about how bankruptcy works. (This is all assuming it goes through the Delaware process -- maybe the Bahamas is odd or shady or outright corrupt.)
The basic idea is that the administrator lists out all the people who lent money to the company, which includes the depositors, but can also be other companies. Then they pool up the assets of the company, selling whatever they can for the best price they can manage, and try to pay it out to the creditors. If all the creditors can be paid back, congrats, the company is solvent! Now it can be returned to the shareholders, who might have something left that's worth more than $0. If creditors can't be made whole, they receive some fraction of what they're owed, and the company is wound down.
In practice there's almost always a seniority ordering of creditors, where you have some loan that was secured against some asset. So it's not necessarily the case that all the creditors will get the same fraction back out. Like, in the basic case, maybe a "creditor" is a bank who gave a mortgage for some property. They get to liquidate that property, so maybe that creditor gets all its money back out, while the others don't.
It's definitely weird and unideal that the depositors are just unsecured creditors. If it were a bank or a brokerage or something, it would be handled differently. Nobody knows what sort of deals FTX might have had, with what sort of creditors, secured against what.
On the other hand, definitely no loans would have been secured against user deposits! An no loans would be secured against like, money to grants. So it's not like you're returning assets that some institutional loan will be secured against. I think this does just mean whatever funds are returned will go into the pool for unsecured creditors. I'd definitely be happy to be corrected on this though.
The other thing to note is, a lot of the other unsecured creditors can be other crypto firms or other sorts of counter-parties. And maybe their holdings with FTX mean that they went broke and their own depositors are out of money. Who knows.
I don't think it's right or good to second guess any of this. There's a law and process for how assets are parcelled out to creditors in bankruptcy. You can trust that if more money is kicked back into that pot, creditors will get a slightly higher amount back.
I'll put it this way. Let's say there was some money that FTX didn't pay out in a grant, and it instead sat in its bank account like it was supposed to. That money will be swept up into the bankruptcy proceeding to be allocated to creditors. Now, consider your situation: instead of sitting in FTX's bank account, that money finds itself in your account. It shouldn't have been transferred to you; FTX wasn't solvent when it made that transaction, it needed to keep all of its money to try to pay back its creditors. So the right place for the money to go to is into the bankruptcy proceeding. That's where it would have be if FTX had never transferred it to you.
As for legal fees, yes they'll definitely make the pie smaller. On the other hand they don't scale linearly with the amount of money, and this is a lot of money -- so hopefully they don't take up that big a percent.
Assisting the process where possible will hopefully help reduce the fees too.
Is that true of grants made back in Jan - Feb? I read somewhere (Forbes, I think) that this was when the big grants to EA orgs were made, whereas it seems maybe the solvency issues didn't arise until after the Luna crash in May?
For a neater, hypothetical version of the question: consider some honest profits FTX made several years ago. If still in their accounts now, it would need to be used to pay back the creditors. But suppose instead that they immediately granted out those profits (several years ago), which seemed an intrinsically legit transaction given the circumstances at a time, and the recipient org for some reason hasn't gotten around to spending those funds (not sure exactly what that means, in accounting terms, since money is fungible and the org would surely have had some expenses during this time; but maybe it was earmarked for a specific purpose that hasn't yet eventuated). Do you think the org is obligated to return the funds in this case?
While none of know exactly what went on or what the state was at any time, my mental model is basically that there was never a time where FTX genuinely had "honest profits" it was free to disperse as it wanted.
It sounds like they intermixed user deposits with operating capital from day 1, and never accounted for anything well enough to have a responsible estimate of whether they had money to donate.
It could be that if you went back to some particular snapshot in time, you could find various points where yes their actual assets exceeded their liabilities (which doesn't count having as "assets" a bunch of shitcoins marked to market). But even at that point, I think if you go back a further they'll have passed through points where they weren't solvent --- where they traded on user funds. This is basically what Shkreli went to jail for: he dipped into one fund to rescue another. The trades happened to work so everyone was in the black, but this still isn't legit.
However, let's grant the premise of your hypothetical, and imagine a world where FTX circa 2020 had always been in the black, and it granted out some of its rightly gained profits. The recipient of that grant shouldn't need to give anything back. In that transaction FTX did have the right to give the grant, so there's no issue.
But I really don't think this hypothetical has much relation to the actual situation. I think it's better for recipients of money from FTX to assume it wasn't legit, and set anything aside that hasn't been spent.
I think this is plausible, but I would currently take a bet against this. My best guess is that customer deposits were safe until earlier this year.
Fair enough, thanks for the explanation!
What's your thought process on this? It's a tough question, but any sense of the rough likelihoods?
Manifold estimates 19%. I don't think this is particularly principled or reliable, but seems better than my personal guesses.
Sorry, I think this Manifold market is answering a different question, it's what percentage of assets will they get back in-expectation.
I do actually disagree with the Manifold market a good amount, so I am updating downwards here. I was assigning like a solid 20% chance that almost all of the funds will be recovered. For example, for Bernie Madoff, of the $18 billion in estimated losses, ~$14.4 billion were recovered, so I was assuming a roughly similar rate here.
I don't think the outcome in Madoff, the most successful Ponzi-scheme recovery in history, would be a good predictor of the outcome here. Madoff's scheme was rather simple and inert at heart -- he basically put incoming money in the bank and took it out when his customers wanted to make withdrawals. So little money was "lost" to bad investment decisions in Madoff. There was merely a transfer from clients who were net depositors into the scheme to those who had withdrawn more than they had deposited. Although there has been much legal wrangling about the clawbacks, there were obvious targets (people who had flat out received more than they put in, in other words people who got totally fictious profits). And those people did not, as a general rule, have good legal defenses.
Although we don't know how FTX/Alameda lost billions in customer monies, all signs so far point to speculative trading activities. For example, if FTX/Alameda bought billions in magic beans (i.e., certain forms of crypto) at fair market value and then that crypto later crashed in value -- that money is gone. You generally cannot claw back transactions that were fair at the time they were made just because they don't look so great in retrospect. Because the transferee gave then-equivalent value for the funds received, an avoidance action isn't going to generate funds for the estate in those sorts of losses.
Although each massive fraud is sui generis, I think the outcome will be closer to Enron (about 7% recovery) than to Madoff due to the lack of good targets. Of course, for anyone who feels that I or the Manifold Markets consensus is wrong -- there's a nice potential profit to be made, as distressed FTX debt is for sale for really cheap last I checked (e.g., 5-8 cents on the dollar).
These are good arguments! I have updated that recovery of most of the money is less likely than I previously thought, having anchored too much on the Madoff case.
However, some random googling caused me to believe that Enron creditors received back 53% of their defrauded assets (a total of $21 billion): https://www.bloomberg.com/news/articles/2012-01-13/enron-creditors-pocket-21-8-billion-in-cash-stock-1-
So I am curious where your 7% is coming from.
I meant the shareholder victims, who had their $100B in stock evaporate and got about $7B from banks. That was intended as an example of a defrauded class who didn't have good targets.
As for the smaller Enron creditor class -- Enron's fraud was substantially more complex than FTX's (which, from initial reports, may have only required a few people to execute without any outside help). Outside corporations with hefty balance sheets were caught up in the fraud -- not with fraudulent intent per se, but with at least somewhat unclean hands that negligently assisted with fraud. It's not clear to me that there are equivalent third parties here (or that those third parties are going to be able to pay judgments). Alameda's counterparties likely didn't have anything to do with draining customer accounts, for instance.
So I put the expected recovery as closer to Enron-investor than near-full recovery, although I freely admit I could be wrong.
Huh, this feels pretty non-analogous to me. Shareholders don't seem to me like victims that are much at all in the reference class of FTX depositors (like, still victims, but much less obvious ones).
When I buy a stock, I am totally taking into account that the company might go bust for various reasons, this includes a probability that it's not reporting its books accurately (like, if a company lies about its books and I nevertheless make money, nobody comes to take my money away from me, and I think many investors totally try to buy into frauds, just to get out earlier than the others).
Indeed, shareholders seem partially morally culpable for having helped Enron defraud other people, by providing funds to run their operations. There are still laws about shareholders being able to sue to get their money back, but the situation strikes me as very different from the depositor case (who were just using a product and had a specific contract that specified that FTX was merely entrusted to keep the money safe, not to do things with).
Noth the ethical and legal rights of these shareholders seem much more limited than the rights of the depositors who I expect to be in the FTX case much earlier in line for having their funds returned than the shareholders in the Enron case.
The Enron shareholders were last in line for any distributions and surely got nothing through the bankruptcy process. However, they owned their own claims for violations of the securities laws designed to protect investors. Those claims were not part of the bankruptcy estate because they were not owned by Enron. However, they did not have many strong targets for the vindication of those claims, so they only got 7%. If there had been a clearly liable and super-deep-pocketed target -- say, someone central to committing securities fraud was secretly a trillionaire -- they would potentially have collected 100%.
My point is that the rate of recovery is often primarily determined by (1) how deep the pockets of potential defendants are and (2) how good those defendants' defenses are. I am suggesting that the FTX depositors -- no matter how clearly their rights were violated -- may not have many deep pockets to sue who lack good defenses. I haven't heard any identification of any deep pockets without good defenses, or information that would lead me to believe they likely exist. Hence my suggestion that the depositors' fate will be closer to pennies on the dollar than to near-full recovery.
Reporting from Bloomberg suggests that depositor claims are being sold at 5-8 cents on the dollar, which suggests that the market agrees that substantial recovery is unlikely. The market could of course be wrong, but these folks are experts in purchasing debt claims against bankruptcy companies. They have doubtless thought harder about potential avenues for recovery thanme or anyone else posting on an EA forum.
Yeah, sorry, I buy the overall claim that FTX debt is sold cheaply, and this is strong evidence that recovery is unlikely, but I still think reasoning from analogy from the Enron shareholders to the FTX depositors seems wrong to me.
I worry that this could go badly wrong if an affected individual were to do this, in the following way:
(1) An affected grantee makes statements to FTX which imply that they believe all of the money that they were given rightfully belongs to FTX's creditors, or could be interpreted in court to imply this
(2) The bankruptcy trustee sends out their routine demand letter. The grantee, having set the money aside, originally thought that they could repay GRANT_SIZE - AMOUNT_SPENT. Unfortunately, their government has treated their grant as income, so they had or will have to pay taxes on the amount that they were unable to deduct. Also, if they aren't planning on trying to negotiate this settlement personally, they will need to enlist a lawyer, who will want their own chunk of any settlement. So they have to hope that the trustee will settle for (GRANT_SIZE - AMOUNT_SPENT - TAXES_OWED - LAWYER'S_FEES), as they will face financial ruin otherwise.
(3) The bankruptcy trustee, who might otherwise have been perfectly happy to settle for that amount, thinks they can do better, because they have the statement that grantee provided in (1) to wave around in court. Grantee has a much harder time arguing in court that they do not legally and morally owe this creditor GRANT_SIZE, given that the bankruptcy trustee has a document in grantee's own words seeming to imply that grantee themselves believes they do legally and morally owe this amount. By the way, the creditor the bankruptcy trustee represents is not an original FTX retail or institutional investor at all, but is instead a Wall Street firm that has bought up FTX customers' bankruptcy rights for pennies on the dollar, looking to turn a profit: https://www.nytimes.com/2022/11/18/business/ftx-assets-wall-street.html . In the end, the grantee might face bankruptcy.
The risks are clearest for individuals. Not sure to what degree an EA community statement similar to (1) would raise analogous risks for grantees or the community as a whole, I don't really have the legal background to know how such a statement might wind up being deployed in court.
Fully agree that just continuing on with some prior plan to deploy the funds seems like a bad idea.
I hear what you're saying about this being uncertain and scary to deal with as an individual, and if someone really doesn't have any sense of the laws that might apply to their situation or what actions might be committal, that puts them in a really tough spot.
For donations that have been made to a literal individual who doesn't have any practice operating a business, I really hope there can be some sort of organisation on top who can take care of this. They at least need to get together and work collectively, it's prohibitive for someone who got like $15k or something to try to sort this out themselves.
But for organisations that are a bit larger (let's say, something like over $1m per year operating budget), I'd expect them to be able to get some basic legal advice about where things stand, and then be able to reach out in a way that starts a conversation without committing them to some sort of disastrous position if things go wrong.
It's a good thing that there's a secondary market for distressed debt. That market gives the creditors valuable options. They can get money out sooner if they need it, and trade from a position of low risk appetite with someone who has the funds and means to take on a larger risk.
The institutions who purchase the debt should be viewed as having the same ethical claim on the funds as the people who sold it to them. They're the rightful creditors at that point.
By the way this also means a statement from the EA community could be relevant to some creditors, because it might affect their pricing. The FTX future fund is obviously not that big a pool of money, overall. But the investment in Anthropic could be 10% or more of the shortfall. If they announce that they'd prefer to unwind the investment if possible, that would be pretty relevant to the price of the debt.
They clearly have the same legal claim, but why do you consider them to have the same ethical claim to me? It seems completely different to me - whether they get more or less back doesn't matter to the people to whom harm was done (it advantages future defraudees, maybe, but that doesn't feel like EAs business)
Fair point re publicly TALKING about this maybe helping, but even then things are a drop in the bucket (Anthropic are a decently large chunk, but VC investments in for profit companies are wildly different situations than the one your post is about)
Regarding the secondary debt holders, my position is that FTX didn't have the right to disperse the funds, and so the grantees who might have custody of the money currently don't have any right to try to intervene in the outcome. Now, I use the term "rights" here, but really that's just a shorthand for the rule utilitarianism I mentioned in the original post. I don't believe in natural rights or something. I just think that society needs to coordinate around clear principles for things like property, and actions which go against those principles are almost always net bad.
So, the people with the title to the debt via the secondary market have the right and proper claim to the property. The people who are holding the property shouldn't try to do anything but get out of the way.
When it comes to Anthropic, I've just seen from your profile that you work there. I understand that you might know much more about the situation, and also that you might not be able to comment on it.
Based on the information that's available (which might paint a misleading picture), I'd say that SBF also didn't have the right to make the investment into Anthropic. So it seems to me that the right thing for Anthropic to do would be to offer to buy back their shares. I think they should not prefer to have $500m in stolen money as investment capital. They should prefer to get their equity back and find another investor.
Again, I accept that there could be a lot of specifics about the situation that aren't public, or where the information that's available is wrong or misleading. But I do strongly believe that Anthropic needs to come forward and explain its position. So far the company hasn't said anything at all in response to this.
Re the secondary debt holders, to me there are two different questions here. One of them is obeying whatever outcome of legal proceedings happens, where I think we both agree that people should fully obey the law. But the second is whether you should go above and beyond and be unsolicitedly cooperative with the legal proceedings, try to proactively allow money to be clawed back, etc. I agree that living in a society with strong, functional property rights is important, but don't think that people going above and beyond in a complex situation like this is a core part of what makes property rights work (and, indeed, any system relying on that would have less reliable property rights!)
I vibe with this, but to me this implies the first but not the second.
Re Anthropic, I used to work there, but left well before the FTX crisis, and have no particular position on those questions. I just think they're a sufficiently different category to be worth clearly distinguishing from Future Fund donations to non-profits.
Anthropic is in a different category for a number of reasons. I believe FTX got something of value (equity in Anthropic) in exchange for the $500MM. If $500MM was in fact a reasonable market value for that equity at the time of the transaction, I don't think Anthropic has any moral obligations here. It gave something of equal value to FTX for what it got. No one would argue that Pepsi has a moral obligation to repay FTX or its creditors if FTX had invested in Pepsi stock at fair market value and then that stock lost value. Of course, if the Anthropic share is worth more than $500MM, the estate can sell it and make some money.
On the other hand, if $500MM wasn't a reasonable value for the equity share at the time of transfer, then I would view the portion of the $500MM that exceeded FMV as a de facto gift that needs to be returned.
Important notice to readers. Please vote up even though it is not very carefully argued here, because it may be important to some readers to read it immediately.
DO NOT FOLLOW THIS POST'S ADVICE. IT IS PROBABLY VERY BAD ADVICE FROM A LEGAL STANDPOINT. IF IT DOESN'T GET YOU IN TROUBLE IT WILL ONLY BE BECAUSE PEOPLE IGNORED YOUR LETTERS.
NEVER FOLLOW ADVICE LIKE THIS FROM PEOPLE WHO ARE NOT LAWYERS.
ONLY DO ANYTHING REMOTELY LIKE THIS IF YOU READ A POST FROM OPEN PHILANTHROPY'S LEGAL COUNSEL TELLING YOU TO DO IT.
I'm happy to edit the post to prepend this notice.
Edit: Done.
I agree this post has some kind of bad advice, but I also don't believe this statement. I think there are many non-lawyers whose advice I would trust more than lawyer advice here, and I've generally found lawyer advice only a relatively weak guide to whether something is actually a good idea.
I do think it makes sense to say something weaker like "Do not follow advice like this from people who have not done pretty thorough legal research, and seem to have good judgement".
I think the sort of people who look at this advice and find that it sounded plausible to them, might want to first follow the rule of only taking advice that originated in actual lawyers, because they couldn't tell which nonlawyers had done real legal research. IDK, I don't know what it's like from the inside to read the original post and not scream.
Seems plausible to me, though the notice you posted didn't really seem to distinguish between different people (and I prefer the world where we don't say things like "never do X" when actually we only want some fraction of people to never do X, but hope that those people will learn to ignore the notices at the right time).
Hm, I think you may be reading the comment from a perspective of "what actions do the symbols refer to, and what would happen if readers did that?" as opposed to "what are the symbols going to cause readers to do?"[1]
The kinds of people who are able distinguish adequate vs inadequate good judgment shouldn't be encouraged to defer to conventional signals of expertise. But those are also disproportionately the people who, instead of feeling like deferring to Eliezer's comment, will respond "I agree, but..."
For lack of a better term, and because there should be a term for it: Dan Sperber calls this the "cognitive causal chain", and contrasts it with the confabulated narratives we often have for what we do. I think it summons up the right image.
When you read something, aspire to always infer what people intend based on the causal chains that led them to write that. Well, no. Not quite. Instead, aspire to always entertain the possibility that the author's consciously intended meaning may be inferred from what the symbols will cause readers to do. Well, I mean something along these lines. The point is that if you do this, you might discover a genuine optimiser in the wild. : )
On a matter of significance, one shouldn't take legal advice for their specific situation from someone off an Internet message board -- whether they claim to be a lawyer or not. Even if the poster is a lawyer, they are not your lawyer, they are not speaking to your individual situation, they are probably speaking outside their field of expertise, and their Internet musings are likely not up to the standards of rigor they would apply in their day jobs. If someone is giving you what sounds like specific legal advice about your specific situation (other than to consult a lawyer), they are probably not a lawyer.
People should consult their own lawyer before taking action on a matter of significance. Ask the lawyer why they are giving the advice they are giving. People are of course free to disregard their lawyer's advice for any reason they find appropriate, but they should at least know what the "orthodox" advice is and why it is being given.
This seem far too conservative to me. I think the advice from Molly (Open Phil's lawyer) will likely be substantially better than the advice by random people trying to find their own lawyer (but not having much experience with choosing a good lawyer), even without access to context. Separately, law is not a magical magisterium, and "legal advice" is not a natural category. Many parts of law can be understood to a totally sufficient degree by laymen, and can be explained to each other, and indeed is often superior to talking to lawyer who are often notoriously bad at communicating certain aspects of the law (like the likelihood of enforcement of various laws).
???
What suggested action or claim warrants this emergency-like statement? I can't find it in this post.
Overall, this contributes to the squalid character of these events, that this whole thing is essentially 16 year olds who read too much online blog posts.
As an aside, I don't think or don't know if FTX grantees should give back all the money, but Yudkowsky's post about is badly argued, intellectually and morally, and it's disappointing, and amazing really, it got the upvotes and credibility it did without the obvious counterarguments appearing.
This sort of behavior is obvious to outsiders.
I think he was referring to the idea of getting in touch with the bankruptcy proceeding to let them know about the funds you've set aside.
I really wasn't trying to advocate this as an immediate action, but when I went back and read the relevant paragraph, it does say that. I don't think there's actually so much danger of people taking my post as a recipe to follow --- I hope it's clear that I'm focussing more conceptually on what the right thing to do is, and that there's still a significant gap between that and the exact way to carry anything out.
Another thing I'm saying is that the public discussion and leadership on this topic has been greatly lacking. On that much I have some agreement with what you've said. However I really think you should reconsider the way you've worded your sentiments. It's fine to register an anonymous account to say something that you wouldn't readily put your name to. But using that as an opportunity to insult is definitely not something that can make anything better.
Hi, I now found and I agree that the advice is bad, directionally.
However, I expect LT people who receive large amounts of funds, to be personally competent and responsible enough to say/write/prepare to set aside these funds. They would be looked down upon if they needed to be scolded on an online forum to navigate the moral and legal issues in the most basic way. Polite disagreement would have been adequate.
(I'm not anonymous, and this setup is intentional, but this is wildly hard to explain. )
More to the heart of the issue, unfortunately, the situation is exactly the opposite as I believe you perceive.
Outside of voting/writing on the EA forums, many parts of EA is treated with absolute contempt and seen as noxious, and this was before November and held by multiple senior EA people across all cause areas, people that you would respect.
As for one example, see Matt Yglesias.
https://www.slowboring.com/p/some-thoughts-on-the-ftx-implosion
What is especially bad and broken is that many people do actually act with great conscientiousness and care online, on the EA forum and Lesswrong, but this is effectively harvested by active people who want access to the resources, power structures, that has been built up by conscientious, unrelated work.
I literally suspected this was deliberate or at least tolerated, in part because this kept the related worldviews relatively weak. However, in the wake of the FTX collapse, this situation and the weakening of MacAskill and non-AI establishment, these latent issues might result in extremely bad states for EA.
I believe large parts of online EA discourse is intellectually bankrupt and dysfunctional. I believe I can decisively articulate why. This would really depress a lot of people without a solution, so I haven't written it up.
I think that on the specific legal questions around clawbacks, this seems super messy, and I expect the correct thing is to do nothing for now. Legal proceeding are not your friend! And I do not think that the standard norms of cooperation, clear communication and good faith clearly apply in a legal situation. Obviously, I think that people should fully obey the law and any results of legal proceedings, but I think that the ethical status of things here is unclear, and I am not comfortable deferring to a court to figure out the most equitable solution. There's a lot of messiness on the part of a money recipient (income tax on the money, legal fees, potentially covering living expenses, etc). And there's a lot of messiness around who would receive clawed back money - eg, I think that ethically a poor, uninformed creditor who invested in crypto because of FTX marketing and lost a lot of their net worth is far more deserving of renumeration than a hedge fund who deposited money on FTX, or a firm who loaned FTX money. But I expect that the legal status of who gets paid money back first/to what ratio is much less clear. And I expect that many of the larger orgs and actors who are owed money will be selling those claims to other financial firms (at a massive discount), and that any actually clawed back money will go to these other firms. Having the expectation that these claims are to be paid back more will give the creditors a higher price to sell the debt, but this is much fuzzier and more diffuse.
On the PR questions, this is unclear to me and I don't have strong takes.
On the ethical questions, I think that "money spent vs unspent" feels like the wrong distinction. To me, the core argument of Yudkowsky's post was that non-profits are not some fundamentally different category to standard capitalism. Either way, you are selling a service, just that in a non-profit situation the goal is for value to go to the world. Pretty much everyone who works for a non-profit could have worked in a standard role instead, and I don't think that choosing to work for a non-profit should significantly change their rights to their labour here. (Similar arguments apply to academics, individuals, for-profits, etc who received FTX money).
If someone received money for work they have since done, but they had enough savings that they didn't actually "spend" any of the money, it just went to savings. But I don't think they have an obligation to return this, in the same way that an FTX caterer with significant runway and significant profit margins is obliged to return the money.
If you interpret unspent funds as payment for labour that has not yet been done, this is less clear to me. But I expect most people in this situation to continue to do the task they had been paid for, which makes it feel analogous to the above - if an academic lab received a grant to work on a research project, I think it is unreasonable to expect them to stop working on that project and to set aside all of the money.
To be clear, I agree that what FTX did was morally reprehensible! But I think this is a confusing and messy situation ethically, and want to push back on a framing that non-profit work is fundamentally different from people selling their labout.
(Disclosure: I was doing independent research on a FTX grant, though for the most part did not receive the money even for research I had already done)
I think when most people say unspent, they mean something like "payment for labor not yet performed or for expenses that can be avoided."
I think the observation that non-profits are selling a service to the grantor and should be treated the same as for-profits actually leads to the conclusion that unspent/unearned funds should be returned. Let's take the various stadium-naming and similar deals as an example, and let's assume that FTX had transferred all the money for those deals upfront. Would it be reasonable and ethical for the stadium authorities, sports teams, etc. to keep all the money if they are ready and willing to perform the services for which they had been contracted (whether continued performance would be of any value to the bankruptcy estate / creditors or not?)
Although I feel bad for the grantees, I can't find any way to distinguish them from any other vendor who was paid in advance.
Thanks for sharing this. I think a good bit of this depends on further factual development -- mainly figuring out when FTX became insolvent. If everything was OK at FTX when a transfer was made, I don't think the grantee has any ethical obligations because FTX ran aground later.
From a rule utilitarian approach, it is rather undesirable for insolvent debtors to be allowed to preference their preferred charities over their legitimate creditors. Credit and insolvency are significant parts of modern economic life, and honoring the reasonable expectations of creditors is an important part of growing the economic pie for everyone. Creditors come in all shapes -- from the employee who has performed work but hasn't been paid yet, to the supplier whose invoice you haven't paid yet, to someone who got burned when the toaster you manufactured blew up due to a manufacturing defect. To the extent a business or individual can get away with just dumping money to a charity at the cost of their creditors, it is going to be a lot harder for businesses and individuals to access credit (which is an oil that makes the economic system run). That is an undesirable state for everyone (including charities which are dependent on corporate donations).
To give two simpler examples:
Unfortunately, I don't think there is any way to effectively prevent creditor-screwing charitable transfers other than a system of clawback liability. Experience teaches us that requiring proof of an actual intent to defraud creditors allows a lot of problematic transfers to stand. The corporate executives should have known that the corporation was insolvent -- and making charitable donations while insolvent is itself a problematic behavior even if there is no affirmative intent to defraud (as it may prove to have been the creditors' money the executive was handing out). Thus, applying the clawback rule to generally all corporations that were insolvent at the time of charitable transfer makes sense.
The bankrupt corporation is dead -- it will either be carved up (liquidation) or sold as a going concern for the benefit of the creditors (reorganization). If you saddle the reorganized corporation with liability for the old corporation's misdeeds, you merely reduce the amount the reorganized corporation can be sold for by that amount (which defeats the purpose of selling it). So "punish FTX for making inappropriate charitable transfers" is not a viable option.
You could argue for a system of personally punishing the corporate officials who approved the charitable transfers while the corporation was insolvent . . . but everyone is worse off under that scenario (corporate officials will be too conservative if they fear facing personal liability for erroneous corporate decisions, meaning they will err strongly on the side of never approving charitable transfers).
Since neither of these options is viable, I don't see any other way to sufficiently discourage these kinds of transfers other than clawing them back from the charities.
Thanks, this is a really great summary of the principles.
I definitely agree that it will depend on further factual development. But I think the facts that are out imply that the insolvency period could go back a very long way. It sounds like the user deposits was intermingled with FTX and Alameda operating capital from the very beginning, and the accounting practices were such that they wouldn't necessarily have known their overall position.
Do you happen to know whether the bankruptcy court just looks at when the executives knew (or negligently failed to find out) that the company was insolvent? It seems kind of weird to expect the court to decide the true value of all the assets at each point in time.
The most relevant statutory text is 11 USC 548(a)(1)(B)(ii), which requires that one of four financial criteria be met for a constructive fraudulent conveyance -- one of which is that the debor "was insolvent on the date that such transfer was made or such obligation was incurred, or became insolvent as a result of such transfer or obligation." That does not seem to require knowledge/negligence on the part of the executives.
I would argue that making corporate charitable contributions when you have any reasonable doubt about whether the company is solvent is itself irresponsible behavior. So I'm not that worried about possibly sweeping in a few cases in which the executives non-negligently believed the corporation was solvent. As a general rule, I don't think most pre-bankrupt corporations are rapidly fluctuating between solvency and insolvency, so I don't think the Code's directive to look at the date of transfer/obligation is that burdensome to bankruptcy courts.
Also, I was using "insolvent" as the primary example of circumstances in which it is utility- maximizing to have a rule to discourage and clawback transfers. I think the other three financial criteria in 548(a)(1)(B)(ii) meet that description too. As relevant here, to the extent that FTX was misappropriating customer funds that did not belong to it, it arguably also meets the criteria that it was "engaged in business or a transaction . . . for which any property remaining with the debtor was an unreasonably small capital."
Thanks! It really should have occurred to me to just look this up and read the statute, it definitely makes things a lot clearer.
I'll be interested to see what value gets ascribed to the various cryptocurrency assets.
Let's say I'm running some business and it's maybe going under. On behalf of the business, I create 101 finger paintings, sell one to my friend's uncle's golf buddy for $10,000, and book the rest as $1m in fine art assets. With my balance sheet shored up, I go on trading, but eventually things don't work out and I file for Chapter 11.
Does the court have to accept the value I put on the paintings at the time, and regard me as solvent for that period? After all, sure, eventually it turned out I couldn't sell the paintings. But that could just be because by then my name was in the news and that tanked the market for my art.
Nope. See 11 USC 101(32) for a statutory definition of insolvency -- a "financial condition such that the sum of such entity’s debts is greater than all of such entity’s property, at a fair valuation . . . ." (emphasis mine).
" You can't make any progress on the problems you care about without strong, practical organisations full of people working together in predictable ways. And those organisations will need to be part of a wider coalition. You need to be trustworthy. "
Great points. This event should help movements including effective altruism, longermism, and those focused on a flourishing future grow and mature. FTX has served as a lightning strike, jolting a community of bright-eyed optimists into a real world of tradeoffs and consequences.
Long term, this might spearhead a valuable transition into integrating with the 'real world' a bit more more. Further integrating with businesses, across disciplines, and government will require breadth of perspective. Self-referential and insular communities, such as the organization that runs this blog, will benefit from reflection and further integration with the >99% of the world with minimal understanding of EA principles.
' We all need to live in a society --- there's very little utility to be found in the state of nature. In order to live in a society we have to agree to cooperate with each other, and agree to withdraw our cooperation from, or even actively punish, defectors. Naive act utilitarianism simply isn't compatible with this. Individuals can't be expected to agree on individual act-based decisions, and you can't form a coalition with people who are making it up as they go along.
A basic respect for property rights is a good ethical rule, that we will all be much worse off without. Support for earning to give in no way entails support for stealing to give.
There are lots of situations, either constructed or actually occurring, where it's unclear how to reason about the priority of different rules or policies. The actions of SBF are not any sort of boundary case. You can't run a society where it's okay for your trading partner to abruptly decide it would be better for the world if your goods were sent elsewhere. '
In this specific case, the majority of recoverable funds should naturally be restored to customers. The world is a bit more grey. Governments tax and allocate a significant portion of GDP and worldwide resources. Property rights, corporate power, and overall influence vary significantly across many of the world's biggest economies.
When it comes to the double edged sword of technological development, and the associated existential risks that could wipe out significant swaths of observable sentient life, our modern conception of just property rights may need to evolve alongside the commensurate risks. As technology has rapidly expanded and accelerated, so have threat vectors and risks.
Hopefully this crisis will prove a valuable opportunity to strengthen and grow as we trod forward.