As a first step, let me just enumerate some hypotheses inconsistent with yours to see if they stick: (all intended to explain why higher-status people seem smarter)
Higher status increases the amount of face you lose when you continue to believe something obviously untrue, or increases the cost of losing face.
High-status individuals were less intelligent when they were young; the observed disparity is due solely to the wisdom that comes with age.
High-status individuals spend more time on dinners and politics, and less time on problem-solving and reading; they exercise their minds more.
High-status individuals are under more pressure to perform, in general.
High-status individuals are just as smart as they ever were, but when you or I try to approach them, the status disparity makes it harder to converse with them—they would sound less intelligent if we had higher status ourselves.
High-status individuals feel more social pressure to listen to your arguments, respond articulately to them, or change their minds when their own arguments are inadequate, which increases their apparent or real intelligence.
High-status individuals get more honest advice from their friends, especially about their own failings (and have better friends).
Great comment, makes it wonderfully clear that Eliezer’s list was all just-so-stories. Now my head spins and I can’t even tell whether the effect is real! Do higher-status people actually make stupider decisions on average? Has anyone ever measured that?
Not all; some of them really don’t work for me the other way round! I feel I would have objected to the following claims even were they presented to me before their counterparts:
Higher status increases the amount of face you lose when you continue to believe something obviously untrue, or increases the cost of losing face.
High-status individuals spend more time on dinners and politics, and less time on problem-solving and reading; they exercise their minds more.
High-status individuals feel more social pressure to listen to your arguments, respond articulately to them, or change their minds when their own arguments are inadequate, which increases their apparent or real intelligence.
High-status individuals get more honest advice from their friends, especially about their own failings (and have better friends).
On the other hand, I find this reversed claim more plausible than the original:
High-status individuals are just as smart as they ever were, but when you or I try to approach them, the status disparity makes it harder to converse with them—they would sound less intelligent if we had higher status ourselves.
I certainly buy that higher-status people have better friends. Status tends to make one’s social network grow, and you can pick the best of them. But ‘best’ might end up meaning something unuseful to intelligence.
The second one you list seems plausible to me (on the face of it). Talking to academics over dinner has exposed me to a lot of interesting arguments that I wouldn’t have encountered in my undergraduate nerdery.
And the first seems true, though it’s tempered by the difficulty in cashing out ‘obvious’ and the ability for high-status people to obfuscate their ignorance.
The large majority of these proposals simply don’t stick, at least for definitions of “higher status” along the lines of 99.9th percentile rather than 99th percentile. Many of the same proposals are very plausible when talking about the 90th percentile rather than the 50th percentile. Sadly, we don’t have grammar rules which encourage us to inconspicuously quantify values in English. Could be useful in a constructed language.
As a first step, let me just enumerate some hypotheses inconsistent with yours to see if they stick: (all intended to explain why higher-status people seem smarter)
Higher status increases the amount of face you lose when you continue to believe something obviously untrue, or increases the cost of losing face.
High-status individuals were less intelligent when they were young; the observed disparity is due solely to the wisdom that comes with age.
High-status individuals spend more time on dinners and politics, and less time on problem-solving and reading; they exercise their minds more.
High-status individuals are under more pressure to perform, in general.
High-status individuals are just as smart as they ever were, but when you or I try to approach them, the status disparity makes it harder to converse with them—they would sound less intelligent if we had higher status ourselves.
High-status individuals feel more social pressure to listen to your arguments, respond articulately to them, or change their minds when their own arguments are inadequate, which increases their apparent or real intelligence.
High-status individuals get more honest advice from their friends, especially about their own failings (and have better friends).
Nice exercise.
Great comment, makes it wonderfully clear that Eliezer’s list was all just-so-stories. Now my head spins and I can’t even tell whether the effect is real! Do higher-status people actually make stupider decisions on average? Has anyone ever measured that?
Not all; some of them really don’t work for me the other way round! I feel I would have objected to the following claims even were they presented to me before their counterparts:
On the other hand, I find this reversed claim more plausible than the original:
I certainly buy that higher-status people have better friends. Status tends to make one’s social network grow, and you can pick the best of them. But ‘best’ might end up meaning something unuseful to intelligence.
The second one you list seems plausible to me (on the face of it). Talking to academics over dinner has exposed me to a lot of interesting arguments that I wouldn’t have encountered in my undergraduate nerdery.
And the first seems true, though it’s tempered by the difficulty in cashing out ‘obvious’ and the ability for high-status people to obfuscate their ignorance.
The large majority of these proposals simply don’t stick, at least for definitions of “higher status” along the lines of 99.9th percentile rather than 99th percentile. Many of the same proposals are very plausible when talking about the 90th percentile rather than the 50th percentile.
Sadly, we don’t have grammar rules which encourage us to inconspicuously quantify values in English. Could be useful in a constructed language.
Well done—these sound just as plausible as their uninverted forms, suggesting additional hidden assumptions behind the original hypotheses.