The Political Prioritization Process

This is a pro­posed doc­trine for poli­ti­cal pri­ori­ti­za­tion. It is the set of ex­plicit and im­plicit prin­ci­ples that guide the Can­di­date Scor­ing Sys­tem; here I col­lect them in com­pact form that can be cited and used much more broadly. I figured it might be helpful be­cause EAs (like ev­ery­one) seem pretty con­fused about how to ap­proach poli­tics in a rigor­ous man­ner, and be­cause it would be good to have a doc­u­ment which can show other peo­ple what they’re of­ten do­ing wrong on the ob­ject level when they in­cor­rectly set poli­ti­cal pri­ori­ties. It can be scaled down to the level of in­di­vi­d­ual think­ing about poli­tics, or scaled up to the level of ma­jor or­ga­ni­za­tional pro­jects.

It draws upon my fa­mil­iar­ity with eco­nomics, poli­ti­cal sci­ence, moral philos­o­phy, de­ci­sion the­ory, philos­o­phy of sci­ence, cause pri­ori­ti­za­tion, the EA com­mu­nity, poli­ti­cal so­cial me­dia, and mil­i­tary doc­trine, and it is also shaped by my ex­pe­rience pro­duc­ing the CSS re­ports and view­ing peo­ple’s feed­back. It could prob­a­bly be adapted to pri­ori­ti­za­tion in other con­texts if you want.

Foun­da­tions and definitions

Poli­ti­cal pri­ori­ti­za­tion is the pro­cess of se­lect­ing, rank­ing and scor­ing poli­ti­cal ob­jec­tives and match­ing the ap­pro­pri­ate re­sponse to them, con­sid­er­ing the abil­ity of the au­di­ence to in­fluence poli­ti­cal pro­cesses.

An ob­jec­tive is an out­come of poli­ti­cal pro­cesses that may be be achieved by the efforts of the au­di­ence par­ti­ci­pat­ing in these pro­cesses. Poli­ti­cal pro­cesses can be elec­toral, ex­ec­u­tive, leg­is­la­tive, ju­di­cial, or so­ciopoli­ti­cal. Poli­ti­cal pro­cesses can be in­fluenced by votes, ac­tivism, or poli­cy­mak­ing.

The pur­pose of poli­ti­cal pri­ori­ti­za­tion is to en­able the au­di­ence to take the ac­tions that best meet the goals of the Effec­tive Altru­ist philos­o­phy. The em­pha­sis is on iden­ti­fy­ing ac­tions which will have the great­est pos­i­tive im­pact on poli­ti­cal sys­tems.

Prin­ci­ples Guid­ing the Poli­ti­cal Pri­ori­ti­za­tion Process

Provide use­ful con­clu­sions to the audience

The sole pur­pose of poli­ti­cal pri­ori­ti­za­tion is to en­able the au­di­ence to di­rect their ac­tions to­wards valuable goals. Con­clu­sions should fol­low five crite­ria in or­der to be use­ful.

Con­clu­sions should be ac­cu­rate—this is self-ex­plana­tory. Con­clu­sions should be de­ci­sion-rele­vant—they should have con­crete im­pli­ca­tions for the choices of ac­tions taken by the au­di­ence. Con­clu­sions should be com­plete—they should not re­quire the au­di­ence to en­gage in fur­ther eval­u­a­tion efforts in or­der to se­lect the right ac­tions. Con­clu­sions should be ac­cessible—they should not re­quire the au­di­ence to draw upon any par­tic­u­lar ed­u­ca­tion or knowl­edge base in or­der to se­lect the right ac­tions. Fi­nally, con­clu­sions should be un­am­bigu­ous—they should not sug­gest con­flict­ing ac­tions for those with plau­si­ble differ­ences in point of view.

The pri­ori­ti­za­tion pro­cess must be struc­tured to pro­duce con­clu­sions that can satisfy these crite­ria. Those con­clu­sions must then be writ­ten and pre­sented in a man­ner that does satisfy the crite­ria. Of course, the pri­ori­ti­za­tion pro­cess should not ig­nore gen­uine un­cer­tainty in or­der to pro­duce a definite con­clu­sion. If the pri­ori­ti­za­tion pro­cess can­not pro­duce con­clu­sions which meet these goals to an ac­cept­able de­gree, then it is bet­ter to recom­mend no ac­tion.

Fol­low a co­her­ent con­cep­tion of EA goals

The pri­ori­ti­za­tion pro­cess must be driven by a co­her­ent con­cep­tion of eth­i­cal goals. This in­volves ax­iol­ogy, the judg­ment of what is valuable or dis­valuable and by how much. It also in­volves nor­ma­tive de­ci­sion the­ory, the judg­ment of how we should re­spond to val­ues and prob­a­bil­ities.

Com­plete the­o­ries with spe­cific judg­ments for ev­ery­thing are rarely nec­es­sary and of­ten difficult to jus­tify. How­ever, in­cor­rect as­sump­tions do not au­to­mat­i­cally ren­der con­clu­sions in­valid—they merely make them pro­gres­sively less ac­cu­rate as the true as­sump­tions differ from the as­sump­tions used in the pri­ori­ti­za­tion pro­cess. There­fore, it is ac­cept­able for the pri­ori­ti­za­tion pro­cess to be driven by a com­plete spe­cific nor­ma­tive the­ory as long as the lead­ers of the pri­ori­ti­za­tion pro­cess are sen­si­tive to how things might change with plau­si­ble vari­a­tions in as­sump­tions.

The pri­ori­ti­za­tion pro­cess can pur­sue a va­ri­ety of goals. How­ever, cer­tain prin­ci­ples are uni­ver­sal or near-uni­ver­sal in the Effec­tive Altru­ist move­ment. First, im­prov­ing well-be­ing is gen­er­ally the biggest pri­or­ity. Se­cond, the well-be­ing of a wide range of pa­tients (typ­i­cally in­clud­ing for­eign­ers, an­i­mals and fu­ture peo­ple) must be con­sid­ered and ag­gre­gated with­out scope ne­glect or parochial­ism. Third, se­ri­ous at­ten­tion must be paid to un­likely out­comes, and the de­ci­sion the­ory should be closer to ex­pected-value max­i­miza­tion than peo­ple’s naive in­stincts are.

Defer to ex­per­tise when it is rele­vant and available

The pri­ori­ti­za­tion pro­cess should have a the­o­ret­i­cal frame­work of epistemic mod­esty, us­ing con­sen­sus among ex­perts as the high­est stan­dard of ev­i­dence on em­piri­cal mat­ters. Par­ti­ci­pants in the pri­ori­ti­za­tion pro­cess must be able to judge ex­per­tise at the in­di­vi­d­ual level and be able to eval­u­ate the col­lec­tive ra­tio­nal­ity of groups of ex­perts.

Par­ti­ci­pants should have a nu­anced, eclec­tic view of ex­per­tise; they should give vary­ing amounts of weight to peo­ple and groups with vary­ing de­grees of ex­per­tise rather than draw­ing ex­plicit thresh­olds. They should also have a flex­ible view of ex­per­tise, where ex­per­tise only ex­ists rel­a­tive to par­tic­u­lar top­ics. When­ever ex­pert sur­veys are lack­ing or in­com­plete, the pri­ori­ti­za­tion pro­cess should look at meta-re­views, stud­ies, blogs/​com­men­tary/​jour­nal­ism or bare ar­gu­ments.

Par­ti­ci­pants in the pri­ori­ti­za­tion pro­cess must be able to in­te­grate these differ­ent types of ev­i­dence into an over­all pic­ture to judge an is­sue. They should be aware of the im­plicit value judg­ments and gaps be­tween ex­ter­nal sources and the so­cial good tar­geted by the pri­ori­ti­za­tion pro­cess. Rarely does any study or ex­pert sur­vey di­rectly tar­get global well-be­ing, rarely do ex­ter­nal sources un­der­stand the dy­nam­ics of global well-be­ing as well as does a typ­i­cal re­searcher in the EA com­mu­nity, and rarely are they guided by com­pletely right eth­i­cal goals. Un­der­stand­ing the true im­pli­ca­tions and mean­ing of a source or ar­gu­ment is just as im­por­tant as judg­ing its sound­ness.

Fur­ther read­ing:

Lewis (2017). In Defence of Epistemic Modesty.

Yud­kowsky (2017). Inad­e­quate Equil­ibria: Why and How Civ­i­liza­tions Get Stuck.

Avoid sweep­ing worldviews

Judg­ments of poli­ti­cal ob­jec­tives should be di­rectly grounded in the eth­i­cal goals as con­ceived by the leader(s) of the pri­ori­ti­za­tion pro­cess, and anal­y­sis should fo­cus on an­swer­ing the ac­tual ques­tions at stake rather than satis­fy­ing some grander nar­ra­tive about so­ciety. The pri­ori­ti­za­tion pro­cess should not be driven by the sweep­ing as­sump­tion or re­jec­tion of en­tire poli­ti­cal ide­olo­gies and sci­en­tific method­olo­gies (in­so­far as ex­pert con­sen­sus on these things is lack­ing). In­stead, it should ex­tract the light­est ar­gu­ments and ideas that may be use­ful for an­swer­ing the ques­tion at hand, and treat them as in­de­pen­dent pieces of ev­i­dence just like any other.

Fur­ther read­ing:

Ace­moglu and Robin­son (2015). The Rise and De­cline of Gen­eral Laws of Cap­i­tal­ism.

Lake (2011). Why “isms” Are Evil: The­ory, Episte­mol­ogy, and Aca­demic Sects as Im­ped­i­ments to Un­der­stand­ing and Progress.

Lake (2013). The­ory is dead, long live the­ory: The end of the Great De­bates and the rise of eclec­ti­cism in In­ter­na­tional Re­la­tions.

Make the pri­ori­ti­za­tion pro­cess ro­bustly inclusive

To max­i­mize ac­cu­racy, the pri­ori­ti­za­tion pro­cess must be open to in­put from a wide range of ideas, sources and in­di­vi­d­u­als. They should be in­cluded based on their po­ten­tial to help an­swer the ques­tions at hand, never out­right ex­cluded for dog­matic or so­cial rea­sons.

Par­ti­ci­pants in the pri­ori­ti­za­tion pro­cess should be cog­nizant of psy­cholog­i­cal and so­cial blind spots which may lead them to ig­nore or mis­rep­re­sent cer­tain points of view, and should de­liber­ately seek out ev­i­dence and ar­gu­ments from peo­ple and sources which they may have marginal­ized. This is a con­tin­u­ous and cycli­cal pro­cess, driven by con­stant re-eval­u­a­tion of one’s cur­rent so­cial and ide­olog­i­cal po­si­tion. It should be grounded in the cog­ni­tion of the in­di­vi­d­ual and the char­ac­ter­is­tics of his/​her lo­cal en­vi­ron­ment. It should not be con­fused with the adop­tion of a fixed set of rules about which groups are un­fairly marginal­ized in so­ciety at a broader level.

Ele­ments of the Poli­ti­cal Pri­ori­ti­za­tion Process


Lead­er­ship of the poli­ti­cal pri­ori­ti­za­tion pro­cess starts with se­lect­ing its ba­sic goals, es­ti­mat­ing the available time and re­sources, and then im­ple­ment­ing an ap­pro­pri­ate pro­ject struc­ture. Con­trib­u­tors are gath­ered to provide rele­vant sources and ar­gu­ments, di­rect­ing them to fo­cus on things with high value-of-in­for­ma­tion. Pro­ject goals are stated clearly enough to fa­cil­i­tate effec­tive mod­el­ing. An over­all pic­ture is main­tained of the trade­offs and un­cer­tain­ties sur­round­ing the pro­ject, judg­ing when it is ap­pro­pri­ate to re­lease, re­vise or close it.


Con­tri­bu­tion to the poli­ti­cal pri­ori­ti­za­tion pro­cess is the iden­ti­fi­ca­tion and writ­ing of spe­cific ev­i­dence and ar­gu­ments to meet the pri­ori­ties as set by lead­er­ship. Con­tri­bu­tion in­volves judg­ment on spe­cific is­sues to provide the in­for­ma­tion rele­vant to mod­el­ing.


Model­ing is the ex­plicit in­te­gra­tion of judg­ments on spe­cific is­sues into an over­all pic­ture with use­ful con­clu­sions meet­ing the goals of the pro­ject. Model­ing can make use of quan­ti­ta­tive or qual­i­ta­tive tools.


Solic­i­ta­tion is the pro­cess of gath­er­ing ev­i­dence, ar­gu­ments and goals from the in­tended au­di­ence and from other out­siders in or­der to con­tribute to the pri­ori­ti­za­tion pro­cess. Solic­i­ta­tion should some­times—but not always—be pack­aged into a two-way com­mu­nica­tive en­gage­ment pro­cess that shows oth­ers the na­ture of the pro­ject, how their in­put is (or is not) be­ing in­cluded, and how their in­put could (or could not) change the con­clu­sions of the pro­ject.


Dissem­i­na­tion is the shar­ing of the pro­ject prod­ucts with the in­tended au­di­ence. Dissem­i­na­tion should always be fol­lowed by two-way com­mu­nica­tive en­gage­ment that an­swers the ques­tions and con­cerns of the au­di­ence and uses their in­put to in­form fu­ture di­rec­tions.

No comments.