Epistemic status: community conversation starter.
Content warnings: mental illness, IQ, drug abuse, speculation about unproven events, the concept of permanent mental degradation, flashing lights / loud sounds on a linked video, a link to the LessWrong tag about (discussion of the meta-idea of) infohazards, League of Legends.
If any ideas have gaps / are poorly-developed, click the nearby links and read them.
Boring Context Part
I saw this link to this post and also saw this comment, and I wanted to respond.
Time Is Not Skill
"SBF is an idiot because he's bad at League of Legends" is the wildest argument I've seen in a while.
I want to first unpack this explicitly, as the article went into minor depth about it:
While this [Sam Bankman-Fried being bad at the game League of Legends] may initially seem like a frivolous observation, it is actually shocking that SBF was unable to rank higher than the Bronze or Silver league after years of regular play across hundreds if not thousands of individual games. It is reflective of an incredibly impaired level of cognition, much like someone who was unable to learn how to ride a bicycle after an entire month of practice or someone who never progresses past the level of an average elementary school student after five years’ of daily piano practice. To me, this is one of the strongest indications possible that SBF is, to be blunt, “not entirely there.”
I have over 2,000 hours in the FPS Team Fortress 2. Shouldn't I be a master at the game?
Many (most?) people who are addicted to a video game, are doing it to relax, not to think.
Relatedly, "dark flow" can happen, where we get a "flow state" where our abilities (at the moment) are matched exactly to the challenge (at the moment)... but "our abilities" are "I'm tired/understimulated", and "the challenge" is "play the game somewhat poorly". This is why "mindless" activities are usually more addictive than non-mindless ones; for most people, most of their gaming time is not high-energy, so high-challenge flow is partly achieved by just playing the game at all.
The analogy used in the article doesn't work, since it compares deliberate-practice hours (piano practice, riding a bike) with those dark-flow hours of League .
Many of my hours in Team Fortress 2 were spent idling, as I left the game running as background noise/imagery while I did something else. I've even done homework in between deathmatch rounds!
Sam Bankman-Fried was, in addition to likely drug use, also:
...attempting to function on low amounts of sleep.
...in the kind of executive position that might've been more "sit at meetings" than "solve technical/risk challenges".
...semi-young and extremely online? (CW: that's the link with the epilepsy/loudness warning)
The "Gamer" Word
Paragraph about ADHD, risk-taking, drug usage, short-term gratification, personality, and gaming .
See what I did there? The point of the above paragraph was to put the ideas in your head, and make it clear that some connection may exist between the above ideas, each other, and Sam Bankman-Fried. The point was not to signal my intelligence or conscientiousness by spending extra time on the frills of prose. Also I'm currently mildly sleep-deprived. Also I used all my sources in the rest of this article.
This key example, getting tired and playing LoL, is not just one way to look at SBF. It's a key idea I've been grasping at for years, and imho one of the central conflicts within EA.
Cognitive Impairment and EA
There are people with varying mental attributes, somewhat changable but mostly fixed . Some of these attributes have tradeoffs that suck most of the time, in most situations... but maybe, if given enough resources (without evidence-of-goodness upfront) and inconveniently-specific requirements and strange modern scenarios, can do well enough to justify the extra hassle.
Is this biased? Yes.
Is this cope about my own abilities? Probably somewhat.
Do I still think it's true? Yes with >50% confidence, though of course my probability's not as high as it was a week ago.
The thing is, EA attracts unusual people. Maybe it attracts people who are weird enough to quickly "get" the ideas, but who also don't have all the capabilities to use them to their fullest. Futhermore, some mental conditions appear to be better for different situations than others, and some (like ADHD/related disorders) seem to be worse in lots of situations.
This should be investigated more thoroughly, by people with more dopamine and slack, especially people who can do it full-time.
The Really Important Paragraph
Say you're a longtermist with https://applieddivinitystudies.com/longtermism-irony/. The things most needing to be done, from technical research to operations to high-end earning-to-give, require cognitive capacity. More, perhaps, than you personally can give at-will. You're not getting any younger (nobody is), but you've heard that young people make better mathematicians/techies/whatever. Maybe on that "high-working-memory-while-having-novel-ideas" thing. So you take the risks, take the drugs, and try to pour yourself into it. Will you burn out permanently? How much of your portfolio-of-time-and-energy should you allocate towards avoiding that? Do you have enough time/energy/slack to do anything without a massive tradeoff between your health and the fate of the world? Are you already too late?
How do we reconcile these observations with SBF’s MIT degree, Jane Street pedigree, successful execution of international Bitcoin price arbitrage, and the founding of Alameda Research and FTX (which, even if ultimately failures, still required some degree of executive function to actually start)? It is not clear, but one could speculate that excessive drug usage may have literally “fried his brain” to some extent.
This is the headspace of a lot of us, whether we admit it or not.
Fraud is bad. This is not legal advice.
Stimulants might still be good, but only in moderation. This is not medical advice.
League of Legends is bad.
Don't(?) feel bad about being bad at video games.
Don't ignore the Kelly criterion . This is not financial advice.
I still have hope that people with some cognitive impairments can make valuable contributions to EA, even on complex tasks like AI alignment and hedge-funding.
That last point is what I wanted to start discussion on. What do you think? Are there places for people who are neurodivergent-but-not-in-the-fun-way? Should EA have better mental health setups in place, considering how many neurodivergent people are attracted to it? Can someone think usefully while taking longer?
Can we get clarity about these things, soon enough to start making a difference? Can we get precision, not platitudes?
Appendix about utility
Overall, it seems very likely that SBF’s propensity to take extreme levels of risk was elevated to preternaturally high levels via heavy, habitual usage of amphetamines and selegiline, causing him to rationalize nonsensical strategies as optimal via convoluted “linear wealth utility” reasoning.
... In particular, once customer deposits were raided, he may very well have tried to “make it all back” in increasingly desperate attempts that just dug the hole deeper and deeper ...
There is one (possible, speculative) intellectual justification for even negative-EV risk-taking: if you think "making it all back" is the only way to get acceptable outcomes. Like, your bet is dumb across possible timelines, but you're sorting yourself into the only timelines where you get what you want. Differing intuitions about this are probably behind Newcomb-boxing disagreements and questions about anthropic reasoning and information hazards. I eagerly await the arrival/development/winning-out of decision theory good enough to clarify this. Of course, in real life this situation probably doesn't happen that way very much at all, and we'd expect to be fooling ourselves if we thought our situation was like that.
I wasn't sure where else to put this, but: Being bad at games while tired/sleep-deprived/etc. is not good evidence of cognitive impairment. However, if you're bad at games on a full night's sleep and while on amphetamines, that is stronger evidence of cognitive impairment. ↩︎
Every vaguely-related thought that enters my mind must become a footnote here. EDIT 16 Nov 2022: I was reading Kelsey Piper's interview with SBF, and his simplification of reality to "win vs lose" (the 9:51pm texts) sure does fit the stereotype/generalization/idea that hardcore gamers are trained to be worse people in some way. Sucks but I'm updating to give that idea more credence. OR, much more likely, excessive competitive gaming selects for people who already simplified life into competition / are sociopathic to some degree. ↩︎
Until/unless we get to some kind of positive genetic/brain-computer-enhancement tech level. ↩︎
As noted in the main article I'm replying to: "SBF is on the record as denying the applicability of the Kelly criterion for bet sizing ... this simply does not make sense in the case of repeated bets." ↩︎
I have really different priors than it seems like a lot of EAs and rationalists do about this stuff, so it's hard to have useful arguments. But here are some related things I believe, based mostly on my experience and common sense rather than actual evidence. ("You" here is referring to the average LW reader, not you specifically.)
P.S. I realize you were trying to make a more general point, but I have to point out that all this SBF psychoanalysis is based on extremely scanty evidence, and having a conversation framed as if it is likely basically true seems kind of foolish.
I agree with the first two points and partly the 4th point. Also the P.S. (I tried to hedge with words like "likely" but I didn't really proofread this a lot).
The 5th point seems like it could apply to me specifically, but like... I don't really know how I'd solve my ego problem, and it's still not clear how bad or whether that's bad in my situation (again, one of my broader points). I know this is likely to be a defense mechanism but... I'm okay with it? Is there decision theory about whether whether I should try to become less inhuman?
If that resembles you, I don't know if it's a problem for you. Maybe not, if you like it. I was just expressing that when I see someone appearing to do that, like the FTX people, I don't take their suggestion that the way they are going about it is really good and important very seriously.
Alright, well thanks for engaging with it and me!
Commenting only on the last paragraph, existing decision theory addresses this adequately: if only a big win is acceptable and everything lesser is essentially equivalent, then according to pretty much any decision theory, you should maximize your probability of achieving that big win even when probability is small.
Kelly betting maximizes long-run expected growth rate, but not probability of reaching a threshold. The criteria for maximizing chance to reach a (large) target are more complex depending upon the context, but are generally less conservative.
The claim about LoL has some merit. At my peak I was top 1% on the ranked ladder, and the best advice I would give to new players is: play 3 games with every champion so you know what they do. The most important skill in LoL is being able to simulate a few seconds forward in time so you're not surprised by anything. As you play longer, you will naturally get better as you encounter more situations.
If you haven't played much, you might not understand that Bronze is an extremely low rank for people that play seriously. When I started playing, I was sorted into mid-Silver after my first 10 matches. I've had numerous friends, including one who rarely plays games, who have quickly gone from no experience and low Bronze to Silver and sometimes Gold. I don't know a single person who's remained in Bronze and plays ranked games with any regularity. SBF's performance is unusually, though believably, bad -- and not entirely due to poor sleep and drug use; he played thousands of games at MIT and wasn't much better then.