Contra the Giving What We Can pledge

Yes­ter­day, there was a Face­book thread dis­cussing ar­gu­ments against the Giv­ing What We Can (GWWC) pledge, where peo­ple promise to donate 10% of all fu­ture in­come to char­ity (https://​​www.giv­ing­whatwe­can.org/​​pledge/​​). The thread didn’t seem very pro­duc­tive, but I do think there are strong ar­gu­ments against the pledge, at least as it ex­ists now. Hence, I thought I’d write up some of the ar­gu­ments against, and try to have a bet­ter dis­cus­sion here. Of course, I only speak for my­self, not any of the thread par­ti­ci­pants (though I’d wel­come their en­dorse­ments if they agree). There are also ar­gu­ments in fa­vor of the pledge, but since many oth­ers have already cov­ered them, I won’t be in­clud­ing them here.

First, and most ob­vi­ously, the pledge recom­mends a flat 10% dona­tion, re­gard­less of a per­son’s in­come. The gen­eral con­sen­sus is that util­ity of money goes as log(in­come), so giv­ing a fixed per­centage is more painful per unit of good done at lower in­comes than higher ones (hence, eg., pro­gres­sive in­come taxes). Differ­ent pro­fes­sions also have dra­mat­i­cally differ­ent ra­tios of di­rect im­pact to money gen­er­ated. Eg., an Amer­i­can con­gress­man’s salary is $174,000, but their votes in Congress are so im­por­tant that even an 0.1% im­prove­ment in vot­ing skill out­weighs a $17,400 dona­tion; hence, spend­ing even a tiny amount of effort donat­ing the 10% is likely not worth it. On the other ex­treme, a high-fre­quency trad­ing job pro­duces lots of money, but al­most no di­rect im­pact. There­fore, the best dona­tion per­centage will vary hugely from per­son to per­son. Given the high hu­man cap­i­tal and low in­come of the me­dian effec­tive al­tru­ist, my guess is that for many peo­ple here, the best per­centage is <1%; on the flip side, for a typ­i­cal billion­aire, it might be 90% or more. The GWWC pledge en­courages most to donate too much, while low­bal­ling a smaller num­ber of large donors.

Se­cond, the GWWC pledge uses the phrase “for the rest of my life or un­til the day I re­tire”. This is a very long-term com­mit­ment; since most EAs are young (IIRC, ~50% of pledge tak­ers were stu­dents when they took it), it might of­ten be for fifty years or more. As EA it­self is so young (un­der five years old, de­pend­ing on ex­act defi­ni­tions), so rapidly grow­ing, and so much in flux, it’s prob­a­bly a bad idea to “lock in” fixed strate­gies, for the same rea­son that peo­ple who take a new job ev­ery month shouldn’t buy a house. This is es­pe­cially true for stu­dents, or oth­ers who will shortly make large ca­reer changes (as 80,000 Hours en­courages). Peo­ple in that po­si­tion have very lit­tle in­for­ma­tion about their life in 2040, and are there­fore in a bad po­si­tion to make bind­ing de­ci­sions about it. In re­sponse to this ar­gu­ment, pledge taker Rob Wiblin said that, if he changed his mind about donat­ing 10% ev­ery year be­ing the best choice, he would sim­ply un-take the pledge. How­ever, this is cer­tainly not en­couraged by the pledge it­self, which says “for the rest of my life” and doesn’t con­tem­plate leav­ing.

Third, the num­ber of GWWC pledge tak­ers is used as a very promi­nent met­ric within EA. It’s listed in bold, 72-point font on the GWWC home­page, was the very first thing men­tioned in Will MacAskill’s monthly CEA up­date, and is found in many other places dis­cussing “the state of EA” or EA’s growth rate. This is prob­le­matic be­cause, as psy­chol­o­gist Dan Ariely says, “you are what you mea­sure”. GWWC pledge count is a bad met­ric for EA as a whole, be­cause:

  • it doesn’t ac­count for effi­cacy of dona­tions; while EA/​GWWC en­courages donat­ing effec­tively, an in­effec­tive donor is still in­cluded in the total

  • it doesn’t ac­count for amount of dona­tions, so five small donors count more than one big donor, even though the big donor prob­a­bly gives more

  • it doesn’t ac­count for di­rect work (eg. dis­cov­er­ing a cure for can­cer as a re­search sci­en­tist) at all

  • it cre­ates weird bi­ases re­gard­ing timing; pos­si­ble fu­ture dona­tions through ~2060 are to­taled on the GWWC home­page, but no ad­just­ment is made for the dra­matic differ­ences be­tween 2016 EA/​hu­man­ity vs. 2060 EA/​hu­man­ity, cre­at­ing the illu­sion of a “per­pet­ual pre­sent”

It might be OK to use GWWC pledge count as one met­ric, mea­sur­ing one as­pect of EA, along with a suite of other met­rics that cap­tured what it missed. How­ever, as far as I know, those other met­rics more-or-less don’t ex­ist right now (I think 80,000 Hours is track­ing “num­ber of ca­reer changes” in­ter­nally, but not sure if that’s been pub­lished any­where). Quan­tify­ing and high­light­ing this one met­ric, while not quan­tify­ing other things, cre­ates quite a large bias. [EDIT: 80,000 Hours has in­deed been pub­lish­ing this met­ric, eg. here. How­ever, I still think that it gets much less promi­nence than the GWWC pledge count. Even within 80K’s own post, it’s listed be­low other (IMO much less im­por­tant) met­rics, like web traf­fic and newslet­ter sub­scriber count.)

Fourth, al­though this isn’t ex­plicit in the pledge it­self, I think many peo­ple tak­ing the pledge in­tend to donate their 10% to GiveWell-recom­mended char­i­ties. (The GWWC pledge was origi­nally spe­cific to global poverty, and GWWC’s char­ity recom­men­da­tions largely over­lap with GiveWell’s.) This seems like a failure to prop­a­gate up­dates. For ex­am­ple, sup­pose your friend Joe is go­ing camp­ing in Ne­vada next week; he packs his RV with tents, clothes, food, wa­ter, and other equip­ment. The day be­fore, Joe says he’s changed his mind, and is ac­tu­ally camp­ing in the moun­tains of Alaska. That’s all well and good, but now that he’s made this change, Joe needs to prop­a­gate that change through the other parts of his plan. He can’t just buy a new map and drive to a differ­ent spot. A change like that will af­fect what clothes he needs to bring, how he should equip his ve­hi­cle, what emer­gency prepa­ra­tions he should take, what he’ll do for fun, and prob­a­bly even things like who will come with him or what food he car­ries.

Of course, I don’t speak for GiveWell, but my im­pres­sion is that the ini­tial GiveWell fo­cus was on up­per-mid­dle-class peo­ple mak­ing four-figure dona­tions ev­ery holi­day sea­son. This has a bunch of im­pli­ca­tions, but the biggest is that the tar­get au­di­ence is mostly busy with work, rel­a­tively risk-averse, and is giv­ing away “spare cash” that won’t be missed that much (if there’s a lean year, they’ll just donate less). In that con­text, the ini­tial GiveWell model (in ~2007-2010) made a lot of sense. How­ever, the GWWC au­di­ence is in­trin­si­cally differ­ent; al­most all pledge tak­ers are mak­ing a very se­ri­ous com­mit­ment, since it’s a sub­stan­tial fixed frac­tion of in­come ev­ery year for decades. And since tak­ing the pledge is still rel­a­tively “weird”, the av­er­age pledge taker will be much less risk-averse. Given those as­sump­tions, it makes much more sense to do a lot of re­search your­self, rather than “out­sourc­ing think­ing”; this is es­pe­cially true given the cur­rent deep dis­agree­ments in EA on what “the most good” even means. (I also ex­pect it would be much more mo­ti­vat­ing for the donor.)

Added: Michael Dick­ens also posted about this last month; I think the ar­gu­ments largely over­lap, but that he fleshes out some of them in more de­tail. H/​T Kit