Rationality tip: Predict your comment karma

For the last few months I’ve taken up the habit of ex­plic­itly pre­dict­ing how much karma I’ll get for each of my con­tri­bu­tions on LW. I picked up the habit of do­ing so for Main posts back in the Visit­ing Fel­lows pro­gram, but I’ve found that do­ing it for com­ments is way more in­for­ma­tive.

It forces you to build de­cent mod­els of your au­di­ence and their so­cial psy­chol­ogy, the game the­o­retic de­tails of each par­tic­u­lar situ­a­tion, how in­for­ma­tion cas­cades should be ex­pected to work, your over­all memetic en­vi­ron­ment, etc. It also forces you to be re­flec­tive and to ex­pand on your gut feel­ing of “peo­ple will up­vote this a lot” or “peo­ple will down­vote this a lit­tle bit”; it forces you to think through more speci­fi­cally why you ex­pect that, and how your con­tri­bu­tions should be ex­pected to shape the minds of your au­di­ence on av­er­age.

It also makes it eas­ier to no­tice con­fu­sion. When one of my com­ments gets down­voted to −6 when I ex­pected −3 then I know some part of my model is wrong; or, as is of­ten the case, it will get voted back up to −3 within a few hours.

Hav­ing pow­er­ful in­tu­itive mod­els of so­cial psy­chol­ogy is im­por­tant for nav­i­gat­ing dis­agree­ment. It helps you re­al­ize when peo­ple are agree­ing or dis­agree­ing for rea­sons they don’t want to state ex­plic­itly, why they would find cer­tain lines of ar­gu­ment more or less com­pel­ling, why they would feel jus­tified in sup­port­ing or crit­i­ciz­ing cer­tain so­cial norms, what un­der­ly­ing ten­sions they feel that cause them to re­spond in a cer­tain way, etc, which is im­por­tant for get­ting the max­i­mum amount of ev­i­dence from your in­ter­ac­tions. All the in­for­ma­tion in the world won’t help you if you can’t in­ter­pret it cor­rectly.

Do­ing it well also makes you look cool. When I write from a so­cial psy­cholog­i­cal per­spec­tive I get sig­nifi­cantly more karma. And I can help peo­ple ex­press things that they don’t find easy to ex­plic­itly ex­press, which is in­finitely more im­por­tant than karma. When you’re tak­ing into ac­count not only peo­ple’s words but the gen­er­a­tors of peo­ple’s words you get an au­to­matic re­flec­tivity bonus. Ob­vi­ously, look­ing at their ac­tual words is a pre­req­ui­site and is also an ex­tremely im­por­tant habit of sane com­mu­ni­ca­tion.

Most im­por­tantly, gain­ing ex­plicit knowl­edge of ev­ery­day so­cial psy­chol­ogy is like ex­plic­itly un­der­stand­ing a huge por­tion of the world that you already knew. This is of­ten a re­ally fun ex­pe­rience.

There are a lot of sub­skills nec­es­sary to do this right, but maybe do­ing it wrong is also in­for­ma­tive, if you keep try­ing.