Some thoughts on EA outreach to high schoolers

[A bunch of these points came from Claire Za­bel. Thanks to var­i­ous peo­ple who pro­vided feed­back.]

TL;DR: I think that lots of EAs have up­dated against out­reach to high school stu­dents based on ev­i­dence that isn’t ac­tu­ally en­tirely rele­vant. I also think that there are some rea­sons to think that out­reach to high school stu­dents could be com­pet­i­tive with out­reach to un­der­grad­u­ates. There are a va­ri­ety of down­sides to out­reach tar­get­ing younger peo­ple, none of which seem de­ci­sive to me.

EAs seem to be too pes­simistic about high school outreach

There have been sev­eral posts about un­suc­cess­ful (ac­cord­ing to the au­thors) at­tempts to get high school­ers into EA. E.g. see here and here. When I’ve talked to peo­ple about differ­ent re­cruit­ing pos­si­bil­ities (in which these posts were of­ten men­tioned if I brought up the pos­si­bil­ity of try­ing to re­cruit high school­ers), I’ve got­ten the sense that many EAs are pes­simistic about try­ing to en­gage high school stu­dents. But I think these past in­ter­ven­tions were in­effec­tive for rea­sons un­re­lated to their tar­get au­di­ence, and that other in­ter­ven­tions aimed at high school stu­dents seem com­pa­rably promis­ing to work­ing with uni­ver­sity stu­dents.

Th­ese posts doc­u­ment at­tempts to en­gage high school­ers that were rel­a­tively short and un­tar­geted (they didn’t strongly se­lect their au­di­ence, other than by age, and didn’t get to se­lect from a very big group). If you imag­ine the analo­gous kind of in­ter­ven­tion for other age groups where EA re­cruit­ment has had no­table suc­cesses, we pre­dict the re­sults would be (and has been) similarly dis­ap­point­ing. E.g. if you took a ran­dom group of a few hun­dred uni­ver­sity stu­dents or re­cent grad­u­ates, se­lected within that group for EA-ness, then showed them a few videos or had them listen to a few hours of talks about EA, we pre­dict the re­sults would usu­ally be similarly lack­luster (and, it’s my im­pres­sion that they have been, when that kind of thing has been tried).

In con­trast, many of the biggest EA groups are at top uni­ver­si­ties, where they can se­lect from thou­sands of stu­dents, and where the stu­dents have been some­what pre-se­lected for traits that seem cor­re­lated with EA-ness, like in­tel­lec­tual cu­ri­os­ity and open­ness.

The only some­what-similarly-tar­geted analogues I know of for high school­ers are SPARC and ESPR (which re­cruit from peo­ple with ev­i­dence of tal­ent in STEM fields, e.g. by look­ing for high school stu­dents that have done well in STEM Olympiad com­pe­ti­tions). I know a de­cent num­ber of SPARC and ESPR alum have gone on to do di­rect work in top cause ar­eas, some of which seems re­ally promis­ing. It doesn’t seem easy to com­pare the hit rate of SPARC to e.g. the Yale EA group, and es­tab­lish­ing causal­ity is always hard, but the story of SPARC seems to­tally differ­ent from the lack of trac­tion SHIC seemed to get.

Ba­si­cally, I think we should treat en­gag­ing with high school EAs more similarly to how we treat en­gag­ing with older EAs: we should look for places with par­tic­u­larly high den­sity of peo­ple who have a chance of con­tribut­ing to high pri­or­ity causes and en­gage them over the course of weeks or months rather than us­ing rel­a­tively short means of en­gage­ment, and also do struc­tured types of en­gage­ment where they can build up con­nec­tions with other peo­ple in­ter­ested in this. If we do that, I don’t think there’s a good rea­son to be more pes­simistic about in­ter­ven­tions to en­gage high school­ers than uni­ver­sity stu­dents.

Benefits to en­gag­ing with younger people

I see a few big up­sides to work­ing with younger peo­ple:

  • It seems harder to re­cruit peo­ple the older they are, at least af­ter peo­ple get into their thir­ties, and so maybe it gets even eas­ier if you go younger than the point at which most re­cruit­ing efforts start. I’m sure this stops be­ing true at some point — I don’t know if one could effec­tively re­cruit a 10-year-old to EA — but the lots of anec­do­tal ex­pe­rience above + my pri­ors in­di­cates that plau­si­bly, a large frac­tion of peo­ple who were con­verted to EA could have been and wish they were en­gaged ear­lier. If more peo­ple could be drawn to EA that way, re­cruit­ing efforts fo­cus­ing on en­gag­ing younger peo­ple might have greater coun­ter­fac­tual im­pact, be­cause they en­gage be­fore some win­dow of op­por­tu­nity elapses.

  • I think the EA and ra­tio­nal­ity com­mu­ni­ties have lots of tools that help peo­ple be­come over­all bet­ter at think­ing, and po­ten­tially vastly in­crease their life­time im­pact. Hav­ing ac­cess to those ear­lier seems like it might be very use­ful for prac­tic­ing us­ing them and thus be­come more skil­led; per­haps it al­lows peo­ple to be­come more skil­led than is pos­si­ble if one doesn’t en­counter these ideas un­til later in life.

    • Re­lat­edly, hear­ing about these ideas ear­lier might cause peo­ple to make some de­ci­sions bet­ter.

  • Anec­do­tally, a large num­ber of the most ded­i­cated and promis­ing longter­mist EAs I know heard about EA in high school (at a work­shop I ran for a small group of newish longter­mist EAs, if I re­mem­ber cor­rectly about ⅔ raised their hands when asked if they’d heard about EA be­fore age 18), and most oth­ers I know that I’ve asked this ques­tion to wish they had, and think their im­pact would be greater if they’d heard about EA ear­lier.

  • EA Sur­vey data im­plies peo­ple who are re­cruited ear­lier change cause ar­eas more fre­quently (see here, the sec­tion called ‘Cause Prefer­ence Change’), which prob­a­bly causes them to land in high-pri­or­ity causes at higher rates (as­sum­ing EA helps with cause se­lec­tion).

  • There’s less com­pe­ti­tion for the at­ten­tion and time of younger peo­ple.

  • Younger peo­ple might be eas­ier to con­vert for a given level of re­cruiter qual­ity, be­cause they have lower stan­dards for whether some­one is knowl­edge­able or in­ter­est­ing.

  • Younger peo­ple of­ten have more free time, es­pe­cially in the time be­tween high school and col­lege, to ex­plore ideas and think broadly about their lives, com­pared to older peo­ple.

Down­sides and failure modes:

It’s harder to se­lect for the most tal­ented peo­ple. There’s less con­cen­tra­tion of in­tel­lec­tual tal­ent among high school­ers than col­lege stu­dents. So you’d prob­a­bly want to put rel­a­tively less em­pha­sis on out­reach at spe­cific high schools (though there are some high schools that are prob­a­bly as dense with tal­ent as top col­leges). Also, it’s prob­a­bly harder to as­sess tal­ent among high school­ers than older peo­ple.

This down­side might seem ex­tremely im­por­tant if you think that stu­dent groups at top col­leges are by far the best in­ter­ven­tion for re­cruit­ing un­der­grad­u­ates.

Get­ting into ra­tio­nal­ity com­mu­nity stuff and EA too early might be bad for peo­ple:

  • I think that ra­tio­nal­ist com­mu­nity stuff can make peo­ple feel more iso­lated and alienated from so­ciety, which seems good in some re­spects but might make peo­ple less mo­ti­vated to be tra­di­tion­ally suc­cess­ful by drudg­ing them­selves through var­i­ous dumb sys­tems; eg they might do worse at col­lege ad­mis­sions or get­ting good grades in dumb col­lege courses.

  • Peo­ple who get into EA some­times seem to lose in­ter­est in in­tel­lec­tual hob­bies (the stereo­typ­i­cal ex­am­ple here is some­one who stops think­ing about math for fun be­cause they need to fo­cus on learn­ing im­por­tant things like AI safety). I’m wor­ried that peo­ple who get into EA while re­ally young might end up not go­ing down a va­ri­ety of in­ter­est­ing in­tel­lec­tual rab­bit holes and end up as worse thinkers for it.

Anna Sala­mon’s post Real­ity-Re­veal­ing and Real­ity-Mask­ing Puz­zles talks about how peo­ple some­times get di­s­ori­ented by think­ing about AI risk; my con­cerns here are pretty similar to those she de­scribes. I think that younger peo­ple might be more vuln­er­a­ble to those pit­falls than older peo­ple.

Also, ra­tio­nal­ity com­mu­nity stuff seems to some­times re­ally mess peo­ple up—they go down weird rab­bit holes and then de­cide to be­come monks or some­thing—and I feel a bit of hes­i­ta­tion about recom­mend­ing that young peo­ple get too in­volved with the com­mu­nity given the risk of that hap­pen­ing with them. I don’t have this same con­cern about EA.

It’s longer be­fore they can do use­ful EA work. This is most se­vere if you think that EAs work­ing now can do much more good per hour than EAs in the fu­ture. High school stu­dents are four years younger than uni­ver­sity stu­dents on av­er­age; so if you have a 10% dis­count rate per year, this leads to think­ing that re­cruit­ing a high school fresh­man is 66% as valuable as re­cruit­ing a col­lege fresh­man. This fac­tor seems likely to be offset by some of the ad­van­tages listed above; in par­tic­u­lar it seems plau­si­ble that re­cruit­ing high school stu­dents is 35% cheaper than re­cruit­ing col­lege stu­dents.

Two other brief points that would de­serve at­ten­tion among peo­ple try­ing to do work in this area:

  • It’s more poli­ti­cally del­i­cate to do out­reach to high school­ers than older peo­ple.

  • If there were more orgs do­ing this, there’d be the risk of abuse work­ing with minors if in-per­son.