New and improved Candidate Scoring System

I’ve sig­nifi­cantly im­proved the Can­di­date Scor­ing Sys­tem which looks at the 2020 US pres­i­den­tial elec­tion.

There is now a per­ma­nent link to re­duce the con­fu­sion of differ­ent ver­sions. When mak­ing up­dates, I will save over the same PDF. And within this doc­u­ment, the links to sup­ple­men­tary CSS files (the Ex­cel model, the draft for shared edit­ing and up­dates, and sup­ple­men­tal col­lec­tions of ev­i­dence/​ar­gu­ments) will re­main per­ma­nent as well.


The re­port it­self has been im­proved in a va­ri­ety of ways—more com­pre­hen­sive, more ac­cu­rate, etc. The con­clu­sions have changed but not to such an ex­tent that our pre­vi­ous recom­men­da­tions could be con­sid­ered harm­ful. Some ma­jor up­dates are: (1) Very helpful anal­y­sis from FiveThir­tyEight which sug­gests that we should fo­cus much more on ma­jor can­di­dates and much less on minor can­di­dates, as the former seem much more tractable. Thus But­tigieg be­comes the top recom­men­da­tion now. (2) Lower opinion of An­drew Yang, due to plac­ing less weight on do­mes­tic tech policy and more weight on in­ter­na­tional re­la­tions for the pur­pose of miti­gat­ing x-risk. This was not an ar­bi­trary change of mind, but an im­pli­ca­tion of a very sen­si­ble re­vi­sion of how the is­sues were di­vided. If some­one wants to in­crease the weight of “Emerg­ing Tech” then we can have that de­bate.

I am still not con­fi­dent that it is ro­bustly ar­gued and defended well enough to be worth broadly shar­ing as out­reach to mass au­di­ences (e.g. “vote for so-and-so, here’s the EA proof”) lest they find prob­lems with it and then con­demn EA to the detri­ment of other causes like global poverty, etc. How­ever, for EAs and any EA-ad­ja­cent peo­ple, it should be use­ful and at least some­what per­sua­sive—not just for this elec­tion but also as a broader sum­mary of ideas on poli­ti­cal policy. It’s also prob­a­bly good to share with peo­ple who will hap­pen to agree with some of its con­clu­sions.

As always, I will take a care­ful look at all in­put and am open to col­lab­o­ra­tion.