Longtermist reasons to work for innovative governments

Epistemic sta­tus: speculative


Longter­mists have good rea­sons to be in­ter­ested in in­sti­tu­tional in­no­va­tion. Broad longter­mist in­sti­tu­tions could fo­cus on im­prov­ing the poli­ti­cal rep­re­sen­ta­tion of fu­ture gen­er­a­tions; more tar­geted ones could fo­cus on par­tic­u­lar cause ar­eas such as AI gov­er­nance.

The com­mu­nity’s ap­proach to in­sti­tu­tional change so far seems to be two-pronged:

  • Re­search new in­sti­tu­tions. Ex­am­ples of such work in­clude Will MacAskill on age-weighted vot­ing and Gillian Had­field and Jack Clark on reg­u­la­tory mar­kets for AI safety.

  • Get into in­fluen­tial po­si­tions in the most in­fluen­tial gov­ern­ments, so that we are bet­ter placed to take im­por­tant de­ci­sions, in­clud­ing re­gard­ing new in­sti­tu­tions. For in­stance, to my knowl­edge, CSET was es­tab­lished partly to help build the poli­ti­cal cap­i­tal of thought­ful longter­mists in D.C.

Here’s an­other ap­proach which might be promis­ing:

  • Get into in­fluen­tial po­si­tions in the most in­no­va­tive gov­ern­ments (i.e., those most will­ing and able to cre­ate and change in­sti­tu­tions), so that we are bet­ter placed to test in­sti­tu­tional in­no­va­tions.

Many of the in­no­va­tions which are in­ter­est­ing by longter­mist lights can only be im­ple­mented by gov­ern­ments (e.g. most of the ideas in this post, reg­u­la­tory regimes); and, pre­sum­ably, gov­ern­ments vary in their will­ing­ness to test new in­sti­tu­tions. There’s also no clear rea­son why the most in­fluen­tial gov­ern­ments should be the most in­no­va­tive. Field-tri­als would help es­tab­lish which ideas work well em­piri­cally, and run­ning them in in­no­va­tive gov­ern­ments would al­low for more ideas, and stranger ones, to be tested. Ac­cord­ing to Robin Han­son, “the key re­source needed for in­sti­tu­tion adap­ta­tion efforts is ac­tu­ally real or­ga­ni­za­tions will­ing to risk dis­rup­tion and dis­trac­tion to work on adapt­ing promis­ing in­sti­tu­tion ideas.”

Suc­cess­ful im­ple­men­ta­tions could also prop­a­gate new in­sti­tu­tions to more in­fluen­tial gov­ern­ments. A cur­sory look at the liter­a­ture on the diffu­sion of poli­cies and in­sti­tu­tions sug­gests that in­sti­tu­tions set up in one coun­try of­ten quickly spread to oth­ers. In the con­text of en­vi­ron­men­tal in­sti­tu­tions, for in­stance, “diffu­sion mechanisms con­tributed to a sig­nifi­cant ex­tent to the in­ter­na­tional spread of en­vi­ron­men­tal ministries and agen­cies, par­tic­u­larly in the 1970s.” One rea­son for this might be the re­duced risk for those fol­low­ing the in­no­va­tor: it’s already been shown that the in­sti­tu­tion can work well.

If the above is cor­rect, per­haps some frac­tion of longter­mists aiming for a gov­ern­ment ca­reer should look to­wards the most in­no­va­tive gov­ern­ments. Some ques­tions that would help clar­ify the value and fea­si­bil­ity of this ap­proach in­clude:

  • Which gov­ern­ments are the most in­no­va­tive?

    • Could EA’s work for these gov­ern­ments?

  • Is the diffu­sion of in­sti­tu­tional in­no­va­tions a ro­bust phe­nomenon?

  • Which coun­tries have his­tor­i­cally had the most suc­cess diffus­ing their in­no­va­tions?


Thanks to Maxime Riche, Jeremy Per­ret, Adam Shimi, and Laura Green for feed­back.