I think I might be missing what’s distinctive here. A lot of the traits listed — strong knowledge of the field, engagement with the community, epistemic humility, broad curiosity — seem like general predictors of success in any many fields.
Are you pointing at something that’s unusually clustered in EA, or is your claim more about how trainable and highly predictive this combination is within the EA context?
I agree they’re generally useful. I claim[1] they’re especially useful in EA. But that’s not enough to make this interesting.
There are many generally useful correlated traits where the correlation with general mental ability is stronger than the correlation between those traits. So being a good writer is good, and also being good at solving math problems is good. And they’re positively correlated. But most things are, just through general intelligence.
What becomes interesting is identifying a cluster of traits that are more correlated with each other than with intelligence, and also predictive of success. These are rare, and usually wrong.[2] If such a cluster can be identified, than knowing about it can help you identify talent that will do well.
If such a cluster is also trainable, well then, you’ve got a real prize.
I thought the same. Curious whether you think these traits are unusually clustered in EA (vs. just “general predictors of success”), or whether the claim is mainly that the bundle is more trainable/predictive in EA contexts.
Also, this is my first week in the EA In-Depth Program, so I’m excited to notice this “in the wild.”
Nice, I hope you train this bundle well! Was it linked somewhere in the In-Depth Program?
Here are 4 hypotheses for what could be going on:
Any generally smart person, in an EA context, will develop a bunch of these traits, to their advantage
There’s some factor that’s not general intelligence, but not specific to EA, that will cause these to be more correlated than expected, but would also apply to someone if they went into another field like biotech
There’s some factor that causes these traits to be more correlated in the EA context in particular, such that noticing this allows you to find people who are unusually good fits for doing EA-style work
That factor is trainable
1 & 2 are kinda interesting, but you don’t need a post on the EA Forum to tell you about intelligence and something like “competence for intellectual work.”
3 starts to get interesting because then you can take someone’s ability to both speak fluently about fish welfare and to use probabilities as a sign that someone will be able to understand and improve your organizations strategy, and even to be unusually cooperative.[1]
If that last clause sounds dangerous to you, I don’t disagree. I think over-reliance of the cooperativeness of other people sharing these traits has caused more than one problem. I nevertheless think this is one of the things that makes EA extremely powerful as an idea.
4 is where you might start getting really hyped about projects like an In Depth Fellowship or an EA Forum. 🙂
That’s what correlation means. Learning about correlations means you can update your bayesian prior about one fact when you learn about a correlated fact.
I think I might be missing what’s distinctive here. A lot of the traits listed — strong knowledge of the field, engagement with the community, epistemic humility, broad curiosity — seem like general predictors of success in any many fields.
Are you pointing at something that’s unusually clustered in EA, or is your claim more about how trainable and highly predictive this combination is within the EA context?
I agree they’re generally useful. I claim[1] they’re especially useful in EA. But that’s not enough to make this interesting.
There are many generally useful correlated traits where the correlation with general mental ability is stronger than the correlation between those traits. So being a good writer is good, and also being good at solving math problems is good. And they’re positively correlated. But most things are, just through general intelligence.
What becomes interesting is identifying a cluster of traits that are more correlated with each other than with intelligence, and also predictive of success. These are rare, and usually wrong.[2] If such a cluster can be identified, than knowing about it can help you identify talent that will do well.
If such a cluster is also trainable, well then, you’ve got a real prize.
With the noted caveats.
Citation needed.
I thought the same. Curious whether you think these traits are unusually clustered in EA (vs. just “general predictors of success”), or whether the claim is mainly that the bundle is more trainable/predictive in EA contexts.
Also, this is my first week in the EA In-Depth Program, so I’m excited to notice this “in the wild.”
Nice, I hope you train this bundle well! Was it linked somewhere in the In-Depth Program?
Here are 4 hypotheses for what could be going on:
Any generally smart person, in an EA context, will develop a bunch of these traits, to their advantage
There’s some factor that’s not general intelligence, but not specific to EA, that will cause these to be more correlated than expected, but would also apply to someone if they went into another field like biotech
There’s some factor that causes these traits to be more correlated in the EA context in particular, such that noticing this allows you to find people who are unusually good fits for doing EA-style work
That factor is trainable
1 & 2 are kinda interesting, but you don’t need a post on the EA Forum to tell you about intelligence and something like “competence for intellectual work.”
3 starts to get interesting because then you can take someone’s ability to both speak fluently about fish welfare and to use probabilities as a sign that someone will be able to understand and improve your organizations strategy, and even to be unusually cooperative.[1]
If that last clause sounds dangerous to you, I don’t disagree. I think over-reliance of the cooperativeness of other people sharing these traits has caused more than one problem. I nevertheless think this is one of the things that makes EA extremely powerful as an idea.
4 is where you might start getting really hyped about projects like an In Depth Fellowship or an EA Forum. 🙂
That’s what correlation means. Learning about correlations means you can update your bayesian prior about one fact when you learn about a correlated fact.