site banner

Scott Alexander on Sam Bankman-Fried, FTX and Effective Altruism

astralcodexten.substack.com

I made this a top level post because I think people here might want to discuss it but you can remove it if it doesn't meet your standards.

Edit: removed my opinion of Scott from the body

15
Jump in the discussion.

No email address required.

That's a fairer defense, and the Ontario Pension Plan admins should be facing serious scrutiny, if not potential review of their licensing (if they have any), as should any who make serious crypto investment with other people's money and no extremely clear disclosure.

Should they though? I think the standard should be higher for institutional investors than for charities accepting donations, but that doesn't necessarily mean the standard for investors should be significantly higher than it already is. They're an easy target because they're partially sponsored by the government, but they were just doing the same thing that the entirely private investors were doing. And the private investors have an appetite for risk because ones that were too risk averse would get outcompeted and replaced in their roles by ones that pursued a more successful strategy. Sequoia Capital is a 50-year-old firm managing $85 billion, and while you could speculate that their employees have recently become less competent or too reckless it seems perfectly plausible that their decision-making here was just the same kind of decision-making that led to these investments:

Notable successful investments by Sequoia Capital include Apple, Cisco, Google, Instagram, LinkedIn, PayPal, Reddit, Tumblr, WhatsApp, and Zoom.

Meanwhile, charities accepting donations both have less to lose, since rather than outright losing an initial investment there's just any money/time wasted by planning around future funding that doesn't come and vague reputational concerns potentially affecting future donations, and more to gain, since you're outright getting money for nothing rather than trying to get a return for money you already have. There's a direct tradeoff between the two, if it's 35% of your funding you risk having wasted more money if it evaporates, while Sequoia obviously doesn't invest that much in a single company - but if you refuse you know you're out a whole 35% of your potential funding, whereas Sequoia can just invest their money in something else. If it's 100% of your funding because you've been soliciting funding for your new charity and they're the first donors to say yes, there's certainly a risk the money will dry up and destroy your charity if you can't find a substitute, but if you refuse there's a risk you won't find enough donations to begin with. You talk about it killing charities, but if a sudden loss of funding can do that how much is because of "less funds than expected" vs. just "less funds, same as if you had refused"?

An EA organization has to... look at other funders? Which, according to Scott's claims here, were already thirsty for good causes to give money to?

The reason why there was more funding than EA charities knew what to do with in the short term was because of FTX suddenly showing up and throwing around a bunch of money, if everyone had refused that wouldn't have been the case. If those other donors don't materialize for the current funding crunch would they have done so to begin with?

It seems like the tradeoffs here pretty strongly favor not being particularly picky about who you accept donations from. Sure if you know someone obtained money from criminality you don't accept the money, but if a dozen institutional investors and the police/SEC don't have a problem then why should you? Now, you could try to mitigate risk in other ways than refusing money outright, like saving more of the money rather than finding ways to spend it immediately, or better yet persuading them to give you a larger endowment rendering you more self-sufficient. But obviously this might not be possible and carries significant disadvantages, for one donors (especially EA donors) want to see actual results from their donations and evaluate your performance, not "we'll do some charity with this money someday". It transfers the risk of the donor having problems to a risk of the charity having problems, like becoming The Wikimedia Foundation with an enormous pile of cash and a huge stream of donations coming in while meanwhile only a tiny fraction gets spent on anything of value. That is after all one of the big problems EA sought to address, and unlike an incompetent/fraudulent for-profit company which eventually collapses to remove the problem, an incompetent/fraudulent charity can continue to waste people's donations indefinitely. I'm not saying that no improvement is possible, for instance maybe there are measures to be more resilient in case funding is lost, but I don't think it justifies extremely costly measures like outright refusing funding because the donor is in a risky field, and I don't think it reflects some deep problem with EA.

Should they though? I think the standard should be higher for institutional investors than for charities accepting donations, but that doesn't necessarily mean the standard for investors should be significantly higher than it already is.

Part of my argument is that this failed to meet the existing standards for institutional investors, at least for the pension funds -- I'm not (just) encouraging delicensing for encourage les autres, but because I'm extremely skeptical that this investment could have been made while honestly and completely complying with the Canadian principles for pension fund investments. (though I am not a lawyer, this is not legal advice, so on)

Sequoia Capital is a 50-year-old firm managing $85 billion, and while you could speculate that their employees have recently become less competent or too reckless...

Funny example! The current Sequoia Capital page on FTX is iso-standard lawyer speak, but the previous page was hilarious. I think I can make a pretty strong argument that the people investing 85 Billion USD to a guy who has a business plan worthy of the underpants gnomes and who plays League of Legends was not competent or prudent.

((During an investment meeting, specifically? Well, that too.))

There's a direct tradeoff between the two, if it's 35% of your funding you risk having wasted more money if it evaporates, while Sequoia obviously doesn't invest that much in a single company - but if you refuse you know you're out a whole 35% of your potential funding, whereas Sequoia can just invest their money in something else.

If you refuse 35% of your potential funding, you can go look for other funders. Very few charities are in a field where they just need or are capable of taking in infinite dollars, because either their goals are limited, or their ability to scale up is limited. This is pretty well-understood in EA and GiveWell circles: indeed, a number of GW top charities have been 'capped out' for years.

This also strikes the other direction. Sequoia isn't just investing less in any individual company than a charity was receiving from FTX Future, but it's devoting a lot less in FTX specifically. Sequoia put something like 210 million USD into FTX, which is absolutely embarrassing, and also well less than 1% of the specific Sequoia fund. The Ontario Teacher's Pension is almost an order of magnitude smaller on a percentage-of-assets measure. I don't think they did adequate due dilligence at this level of expenditure, and there are significant limitations to using models like the Kelly criterion even for investment funds, nevermind for charity, but the scope of the difference is a marker.

((Though not all 'conventional' hedge funds were that sober: Galois put nearly half of their bank on FTX, which would have been irresponsible to do for something like Apple.))

You talk about it killing charities, but if a sudden loss of funding can do that how much is because of "less funds than expected" vs. just "less funds, same as if you had refused"?

There are serious problems specific to having committed funds disappear on short notice, that are not present to not having the funds to start with. In the extremes, you've already bought crap from vendors and have to pay for it, signed leases, hired and/or moved people. Short of that case, there's still a bunch of softer commitments that have significant reputation or financial costs.

The reason why there was more funding than EA charities knew what to do with in the short term was because of FTX suddenly showing up and throwing around a bunch of money, if everyone had refused that wouldn't have been the case. If those other donors don't materialize for the current funding crunch would they have done so to begin with?

If we were talking about FTX-screwed charities as those which focused more on the Earning to Give side than the Rationality one, I think this would be a more plausible argument. My worry is not that someone got snookered: my worry is not even that someone focused on avoiding these risks got made a bad bet. But right now we're combining all of that with a 'whoops, shit happens, nothing we could have done'.

I think I can make a pretty strong argument that the people investing 85 Billion USD to a guy who has a business plan worthy of the underpants gnomes and who plays League of Legends was not competent or prudent.

I do find it funny how LOL is catching so many strays in this scandal. Like, its a popular video game. Some hopped up fake billionaire likes it, like the rest of the world of hopped up people.