Thoughts on electoral reform

Ex­ces­sive poli­ti­cal po­lari­sa­tion, es­pe­cially party po­lari­sa­tion in the US, makes it harder to reach con­sen­sus or a fair com­pro­mise, and un­der­mines trust in pub­lic in­sti­tu­tions. Efforts to avoid harm­ful long-term dy­nam­ics, and to strengthen demo­cratic gov­er­nance, are there­fore of in­ter­est to effec­tive al­tru­ists.

One con­crete lever is elec­toral re­form. By chang­ing to a bet­ter vot­ing sys­tem, we could (so the ar­gu­ment goes) elect offi­cials that bet­ter rep­re­sent the elec­torate, re­sult­ing in a more func­tional poli­ti­cal pro­cess.

Within effec­tive al­tru­ism, ap­proval vot­ing is the most promi­nent pro­posal for re­form (see e.g. this post). The Cen­ter for Elec­tion Science, which ad­vo­cates ap­proval vot­ing, ap­pears to be the only elec­toral re­form or­gani­sa­tion with sig­nifi­cant ties to the EA com­mu­nity—e.g. in the form of a grant from OpenPhil.

In this post, I will ques­tion the fo­cus on ap­proval vot­ing, and ar­gue that it might be bet­ter to sup­port other pro­posed vot­ing sys­tems that have a track record in com­pet­i­tive elec­tions. I’ll also offer some thoughts on how promis­ing elec­toral re­form is.

Which vot­ing sys­tem should we ad­vo­cate for?

There are many pos­si­ble meth­ods and many differ­ent crite­ria to eval­u­ate a vot­ing method, some of which are prov­ably in­com­pat­i­ble. So, like other meth­ods, ap­proval vot­ing satis­fies some crite­ria and fails oth­ers. (See here for an overview of sin­gle-win­ner vot­ing meth­ods and satis­fied/​failed crite­ria.) Given that no perfect method ex­ists, we should ar­guably look for a method which works well in prac­tice and has good chances of be­ing adopted.

Would ap­proval vot­ing work well in com­pet­i­tive elec­tions? I think there are good rea­sons to be scep­ti­cal:

  • Ap­proval vot­ing is vuln­er­a­ble to tac­ti­cal vot­ing. It fails the later-no-harm crite­rion: ap­prov­ing a sec­ond can­di­date can hurt your favourite. The av­er­age voter prob­a­bly isn’t that strate­gic, but in high-stakes elec­tions, savvy cam­paign lead­ers would surely at­tempt to get their sup­port­ers to vote tac­ti­cally. The win­ner, then, may not be the can­di­date with the most sup­port, but the one that’s best at ma­nipu­lat­ing the sys­tem. (See here for more de­tails on this.)

  • Ap­proval vot­ing rad­i­cally re-in­ter­prets the com­mon-sense no­tion of “hav­ing a ma­jor­ity”, lead­ing to re­sults that may be con­sid­ered coun­ter­in­tu­itive. This is re­flected in the vot­ing method crite­ria that ap­proval vot­ing fails. For in­stance, ap­proval vot­ing some­times se­lects a can­di­date even though a ma­jor­ity of vot­ers would, in a head-to-head con­test, pre­fer any other can­di­date. (This is the Con­dorcet loser crite­rion.)

  • Indi­cat­ing sup­port or op­po­si­tion for each can­di­date is more ex­pres­sive than just hav­ing a sin­gle vote, but it is still bi­nary and does not al­low vot­ers to ex­press more nu­anced prefer­ences be­tween differ­ent can­di­dates.

  • There is al­most no track record of ap­proval vot­ing be­ing suc­cess­fully used in com­pet­i­tive elec­tions. Where it was used, ap­proval vot­ing was of­ten re­pealed later on—e.g. in Dart­mouth alumni elec­tions and in in­ter­nal IEEE (In­sti­tute of Elec­tri­cal and Elec­tron­ics Eng­ineers) elec­tions (search for IEEE here).

Ad­vo­cates of ap­proval vot­ing have re­sponded to those crit­i­cisms. And of course, ap­proval vot­ing does also offer ad­van­tages: it avoids the spoiler effect and tends to favour mod­er­ate “com­pro­mise can­di­dates”. The lat­ter is of par­tic­u­lar in­ter­est if re­duc­ing po­lari­sa­tion is one of the main goals of elec­toral re­form. How­ever, the ten­dency to favour mod­er­ate can­di­dates could also be con­sid­ered a bias and is not uni­ver­sally viewed as a pos­i­tive fea­ture of a vot­ing sys­tem.

All things con­sid­ered, I’m not con­vinced that we should ad­vo­cate for ap­proval vot­ing rather than other meth­ods (e.g. in­stant runoff vot­ing or Con­dorcet meth­ods). It seems to me that effec­tive al­tru­ism has not ex­am­ined ap­proval vot­ing (or al­ter­na­tives) in suffi­cient de­tail.

In gen­eral, my im­pres­sion is that dis­cus­sions of vot­ing re­form suffer from the prob­lem that peo­ple tend to pick their favourite method and then cherry-pick one-sided ar­gu­ments in favour of it. In par­tic­u­lar, peo­ple overem­pha­sise crite­ria that favour their method does well while ig­nor­ing or and down­play­ing prob­lems. The Cen­ter for Elec­tion Science of­ten talks about no favourite be­trayal (which ap­proval vot­ing satis­fies) and not much about later-no-harm (which it fails). FairVote doesn’t talk much about no favourite be­trayal and talks a lot about later-no-harm—be­cause their favoured method (in­stant runoff vot­ing) satis­fies later-no-harm but fails no favourite be­trayal.

Given all this, what kind of vot­ing sys­tem should we ad­vo­cate for (if any)?

Since there is (some de­gree of) con­sen­sus that plu­ral­ity vot­ing is bad, but no con­sen­sus on which al­ter­na­tive is best, we should fo­cus on the re­form pro­pos­als that are most vi­able. That’s ar­guably in­stant runoff vot­ing (IRV, called ranked choice vot­ing /​ RCV in the US), which is cham­pi­oned by FairVote. Un­like ap­proval vot­ing, IRV has a track record in com­pet­i­tive elec­tions and is much more in line with con­ven­tional no­tions of “ma­jor­ity”. (My per­sonal favourite vot­ing sys­tem would be a Con­dorcet method such as Ranked Pairs, but there are no large or­gani­sa­tions ad­vo­cat­ing this, and it’s un­likely that Con­dorcet meth­ods will be adopted.)

IRV isn’t perfect ei­ther. It also fails im­por­tant crite­ria, and it isn’t clear whether IRV re­sults in less po­lari­sa­tion. Still, IRV seems clearly su­pe­rior to plu­ral­ity vot­ing and has stood the test of time, so I think efforts to im­ple­ment IRV are worth sup­port­ing. (Even the very sim­ple step of adding a runoff be­tween the top two can­di­dates would be a sig­nifi­cant im­prove­ment over plu­ral­ity vot­ing.)

Note that this dis­cus­sion is mostly about sin­gle-win­ner elec­tions such as US pres­i­den­tial elec­tions, rather than multi-win­ner elec­tions, such as elect­ing a par­li­a­ment (e.g. the House of Rep­re­sen­ta­tives). It seems not ob­vi­ous, over­all, whether we should fo­cus on chang­ing sin­gle-win­ner elec­tions or par­li­a­men­tary elec­tions.

For par­li­a­men­tary, I think it’s best to use a form of pro­por­tional rep­re­sen­ta­tion rather than (or in ad­di­tion to) first-past-the-post in sin­gle-seat con­stituen­cies. Pro­por­tional rep­re­sen­ta­tion tends to lead to multi-party sys­tems that re­quire cross-party col­lab­o­ra­tion and re­duce the team sport men­tal­ity that drives US po­lari­sa­tion.

How promis­ing is elec­toral re­form?

Clearly, work on elec­toral re­form is premised on the be­lief that the sta­tus quo of plu­ral­ity vot­ing (in the US/​UK) is a poor vot­ing method. I think this isn’t en­tirely ob­vi­ous. A steel­man of plu­ral­ity vot­ing is that it grants power to the largest co­her­ent poli­ti­cal coal­i­tion (co­her­ent in the sense of be­ing able to co­or­di­nate on a sin­gle can­di­date). That may be less than 50% of the vot­ers, but it’s not prima fa­cie un­rea­son­able to put the largest co­her­ent poli­ti­cal coal­i­tion in charge of things. The fact that plu­ral­ity vot­ing pro­vides tac­ti­cal in­cen­tives that limit the num­ber of (re­al­is­tic) op­tions (of­ten lead­ing to a two-party sys­tem) can be seen as a fea­ture, not a bug: it chan­nels demo­cratic de­ci­sion-mak­ing and pro­duces clear re­sults.

Still, I think the down­sides of plu­ral­ity vot­ing out­weigh its ad­van­tages, and there is some de­gree of con­sen­sus among ex­perts that plu­ral­ity vot­ing is not a good sys­tem.

Sup­pose, then, that we have set­tled on a vot­ing sys­tem that would be a sig­nifi­cant im­prove­ment over the sta­tus quo. That raises the ques­tion of how tractable it is to change the vot­ing sys­tem for high-stakes elec­tions such as the US pres­i­dent or the House of Rep­re­sen­ta­tives. Such efforts face sig­nifi­cant vested in­ter­ests of in­di­vi­d­u­als and par­ties that benefit from the cur­rent sys­tem. Also, coun­tries rarely make dra­matic changes to their vot­ing pro­ce­dures once es­tab­lished, though there are ex­cep­tions (e.g. New Zealand’s switch to pro­por­tional rep­re­sen­ta­tion).

Elec­toral re­form also doesn’t seem par­tic­u­larly ne­glected. There are sev­eral or­gani­sa­tions ad­vo­cat­ing elec­toral re­form in the US and UK. In terms of fund­ing, the Hewlett Foun­da­tion’s Madi­son Ini­ti­a­tive is a ma­jor player that’s in­ter­ested in elec­toral re­form (sup­port­ing FairVote’s ad­vo­cacy for ranked choice vot­ing) and other ways to strengthen US democ­racy.

All things con­sid­ered, I think elec­toral re­form, while prob­a­bly not a “top tier” in­ter­ven­tion, should be part of the longter­mist EA port­fo­lio.