Bankman-Fried was wrong, but I sympathize with him
Nina McCambridge, April 8, 2024, 6 minutes
Editorials featured in the Forum section are solely the opinions of their individual authors:
I tend to feel bad for high-profile, white-collar criminals. Admitting this probably reflects poorly on me, and I of course think that stealing and lying are wrong and should and will continue to be against the law. However, I found that I was relieved when Sam Bankman-Fried didn’t get a life sentence. Perhaps this is because he seems like someone I’d know. My high school friend liked to tell us about how her parents were friends with Bankman-Fried’s parents — this was before the news began to focus on the complicity of the parents. I was not born like Bankman-Fried (I always appreciated style) but I was inducted into the regional subculture that produced him. I, too, am from the Bay Area. I, too, like math. Here’s something that differentiates us:
Bankman-Fried didn’t like decorum. He dressed badly. He played video games while talking to people. Was it the case that he had no appreciation for aesthetics (as he said), or was it the case that a seeming forgoance of aesthetics was his preferred aesthetic?
Bankman-Fried described himself as someone who does not take pleasure in life experiences, an anhedonic. I am not sure whether this is true or whether he was trying to be edgy. While the former case is an appealing thing for us to believe — it would make him less like ourselves — it seems like a common enough teenaged countersignal. (Consider Morrissey singing about how he wishes he could feel anything at all, when obviously he is a very emotional person.) At least, he either was an anhedonic or he was attempting to become one.
Bankman-Fried was optimizing his life towards something. Most of us are not so self-conscious. We have a great variety of values that often conflict. There are a few ways of dealing with this. One is not to examine our values, lying to ourselves about them or perhaps not thinking about them at all. This is what most of us do, and it ends up working out decently well. If you just do what seems normal, you will have normal outcomes.
The world is full of suffering. The world is full of existential risk. Think about this too much and it will drive you crazy. Most people don’t really care about what’s going on outside of their community. This makes sense, because most people have no way of predictably, substantially affecting whatever’s happening far away. Some people, like Fried, don’t think they have to remain helpless in the face of material evil. They have come up with a method that might actually be predictable and substantial.
After you’ve come up with something like this, what else matters? If you can help people on a global scale, what does it matter if you are harming people on a local scale? What does it matter if you aren’t following any of the old laws? Those laws were made with the assumption that no one could be as powerful as you currently are.
Most of us, of course, distrust that such magical methods exist, because radical changes are usually fundamentally ephemeral. Maybe Fried had a grand unified utility function and probability function which he actually used. Most of us don’t find that very comforting. We learned that he was stealing, and we all know that more important than any charity work is the law: Stealing is wrong.
This law makes things better in general. It probably shows up in a lot of specific metrics; I’m sure that if you took two identical countries and made it legal to steal in one of them, their GDP would fall and so on relative to the other country. This isn’t why we decided that it’s wrong to steal. This law just seemed right, so we kept following it — this law is far more fundamental than any metric.
Optimizing too hard can make things really horrible. Our values are too inscrutable for this to work. Fried probably knew a similar argument in the context of AI safety.
I am often frustrated that we enforce a lot of laws very selectively. If you didn’t want to enforce the law in that particular case, why didn’t you just amend the law? Of course, in our republic, laws are done by committee, so you don’t always get what you want. But even if you were a dictator, you wouldn’t be intelligent enough to have perfect foresight into all the little moral situations and unintended consequences of your law. You would still end up being selective. In fact, you would be much more selective than any liberal government, which is designed for relative universality. However, this sort of thing only applies to laws that aren’t extremely fundamental. It takes a lot more chutzpah to excuse a murder than a violation of the Americans with Disabilities Act.
If you read through the entire legal code of your government and found yourself convinced that all of it was good, even if you had trust that the government and its laws would remain stable and good, you would likely still opt out of the government putting cameras in your kitchen. You’re a law-abiding citizen and generally appreciate state capacity (right?), but you don’t want the government to be perfectly optimized.
Something similar can happen with your values. Introspection and non-hypocrisy are usually good. If you apply them too zealously to the strange values that you chose with your own grand intelligence, you might end up much worse off than if you didn’t think about it at all. Maybe not — society is often wrong. (In the case of Bankman-Fried, society is definitely correct: No scamming allowed!)
Does any of this mean anything? Does it bear relation to Bankman-Fried’s error? Was Bankman-Fried actually trying to be effective? I am not sure. He definitely made a lot of effective charity contributions. He also made a lot of strange and inefficient-seeming donations; he donated vast sums of money to Democratic politicians and then later claimed to have also donated to Republican politicians. He donated to high-profile but meaningless projects.
He did give villainous monologues justifying these choices in the name of public relations. Was this necessary to his charitable goals? Perhaps these donations were motivated by similar urges — to increase his standing in the eyes of two different charitable subcultures. This is much more human: To be driven by decorum. Still, with the aesthetics of optimization, you might get to the same bad end anyways.
