Age-Weighted Voting

Epistemic sta­tus: I think this is an in­ter­est­ing idea that’s worth think­ing about, but it would need a lot more in­ves­ti­ga­tion be­fore I’d want to en­dorse it.

If we’re try­ing to pos­i­tively in­fluence the long-run fu­ture, we im­me­di­ately run into the prob­lem that pre­dict­ing the fu­ture is hard, and our best-guess plans to­day might turn out to be ir­rele­vant or even harm­ful de­pend­ing on how things turn out in the fu­ture. The nat­u­ral re­sponse to this is­sue is to in­stead try to change in­cen­tives — in par­tic­u­lar, poli­ti­cal in­cen­tives — such that peo­ple in the fu­ture take ac­tions that are bet­ter from the per­spec­tive of the long-run fu­ture. As a com­par­i­son: the best ac­tion for fem­i­nist men in the 19th cen­tury wasn’t to figure out how best to help women di­rectly (they prob­a­bly would have failed dis­mally, es­pe­cially if they were aiming at long-term benefits to women); it was to cam­paign to give women the vote, so that women could rep­re­sent their own in­ter­ests.

The trou­ble with the analo­gous rea­son­ing when if comes to fu­ture peo­ple is that, be­ing not-yet-ex­is­tent, fu­ture peo­ple can’t rep­re­sent their own in­ter­ests. So ‘give fu­ture peo­ple the vote’ isn’t a vi­able op­tion.

But there’s an al­ter­na­tive path. Gen­er­a­tions over­lap, and so by do­ing more to em­power younger peo­ple to­day, we give some­what more weight to the in­ter­ests of fu­ture peo­ple com­pared to the in­ter­ests of pre­sent peo­ple. This could be sig­nifi­cant. Cur­rently, the me­dian voter is 47.5 years old in the USA; the av­er­age age of sen­a­tors in the USA is 61.8 years. With an ag­ing pop­u­la­tion, these num­bers are very likely to get higher over time: in de­vel­oped coun­tries, the me­dian age is pro­ject to in­crease by 3 to 7 years by 2050 (and by as much as 15 years in South Korea). We live in some­thing close to a geron­toc­racy, and if vot­ers and poli­ti­ci­ans are act­ing in their self-in­ter­est, we should ex­pect that poli­tics as a whole has a shorter time hori­zon than if younger peo­ple were more em­pow­ered.

So one way of ex­tend­ing poli­ti­cal time hori­zons and in­creas­ing is to age-weight votes. The idea is that younger peo­ple would get more heav­ily weighted votes than older peo­ple, very roughly in pro­por­tion with life ex­pec­tancy. A nat­u­ral first pass sys­tem (though I think it could be im­proved upon) would be:

  • 18-27yr olds: 6x vot­ing weight

  • 28-37yr olds: 5x vot­ing weight

  • 38-47yr olds: 4x vot­ing weight

  • 48-57yr olds: 3x vot­ing weight

  • 58-67yr olds: 2x vot­ing weight

  • 68+yr olds: 1x vot­ing weight

Later edit: Note that, even with such heavy weights as these, the (effec­tive) me­dian voter age (in the US) would go from 55 to 40. (H/​T Zach Groff for these num­bers). As­sum­ing that the me­dian voter the­o­rem ap­prox­i­mately cap­tures poli­ti­cal dy­nam­ics of vot­ing, weight­ing by (ap­prox­i­mate) life-ex­pec­tancy would there­fore lengthen poli­ti­cal hori­zons some­what, but wouldn’t re­sult in young peo­ple hav­ing all the power.


As well as the po­ten­tial benefits from ex­tend­ing poli­ti­cal hori­zons, I think this pro­posal looks promis­ing on some other di­men­sions too:

It would be fair. In this sce­nario, all cit­i­zens get equal vot­ing power, it’s just that this vot­ing power is un­equally dis­tributed through­out some­one’s life.

In fact, there are ar­gu­ments that it would be fairer than the cur­rent sys­tem. First, it’s fairer in­so­far as there’s a closer as­so­ci­a­tion be­tween who has power over which poli­cies are en­acted and who has to bear the benefits and costs of those poli­cies. It avoids sce­nar­ios where some peo­ple can vote for short-ter­mist poli­cies that benefit them even though they don’t have to live with the long-run con­se­quences. Se­cond, the cur­rent sys­tem gives less vot­ing power to peo­ple who have the mis­for­tune of dy­ing young. The age-weighted sys­tem miti­gates this to an ex­tent. Fi­nally, if it does suc­ceed in en­courag­ing poli­cies with bet­ter long-term con­se­quences, it would be some­what fairer to fu­ture gen­er­a­tions, who are cur­rently com­pletely dis­en­fran­chised; though these gen­er­a­tions still wouldn’t be able to rep­re­sent them­selves, they would at least be benefit­ted to a greater de­gree.

It would miti­gate in­tertem­po­ral in­con­sis­tency. In the UK’s Euro­pean Union mem­ber­ship refer­en­dum, the vot­ing pat­tern was heav­ily cor­re­lated with age: older peo­ple were much more likely to vote to leave the EU than younger peo­ple. My cur­rent (poorly in­formed) un­der­stand­ing is that, in terms of the cor­re­la­tion be­tween age and con­ser­vatism, both ag­ing it­self and co­hort effects play a role. If the lat­ter is sig­nifi­cant in this case, this sug­gests that, in twenty years’ time, most of Bri­tain’s elec­torate will be in favour of be­ing part of the EU. If so, then a huge amount of time and effort will have been wasted in the tran­si­tion costs of leav­ing and re­join­ing.

There are, how­ever, a num­ber of open ques­tions re­gard­ing age-weight­ing of vot­ing, in­clud­ing:

  • Do younger peo­ple ac­tu­ally have more fu­ture-ori­ented views?

  • Does ex­tend­ing poli­ti­cal hori­zons by 20 years provide benefits from the per­spec­tive of much longer timescales?

  • Are younger peo­ple less well-in­formed, and so apt to make worse de­ci­sions?

  • Is this just a way of push­ing par­tic­u­lar poli­ti­cal views?

  • What would ac­tu­ally hap­pen if this were put in place, and how good or bad would those effects be?

  • What’s the best pre­cise mechanism for im­ple­ment­ing age-weight­ing vot­ing?

  • What would be the best plan for mak­ing age-weight­ing vot­ing hap­pen in the real world?

Some brief notes on these:

Age and fu­ture-ori­en­ta­tion: One could ar­gue that older peo­ple are more likely to con­sider the long run. They have less at stake in terms of per­sonal in­ter­est, so there­fore might weigh al­tru­is­tic con­cerns com­par­a­tively more highly than self-in­ter­ested con­cerns. (Imag­ine, at the limit, some­one who was vot­ing on their deathbed. They would only have moral con­cerns to guide their de­ci­sion. Thanks to Chris­tian Tarsney for this point). And, in gen­eral, vot­ing be­havi­our isn’t well-ex­plained by the ‘self-in­ter­ested voter’ model. How­ever, em­piri­cally there’s ev­i­dence that gen­er­a­tions tend to vote in their self-in­ter­est when it comes to is­sues that have differ­ent costs and benefits across time. Here’s Gabriel Ah­lfeldt sum­ming up some re­sults from a re­cent pa­per:

“[O]lder vot­ers are less likely to sup­port mea­sures that pro­tect the en­vi­ron­ment, pro­mote sus­tain­able use of en­ergy or im­prove trans­port. Older vot­ers are also less likely to sup­port ex­pen­di­tures on ed­u­ca­tion or welfare poli­cies, such as un­em­ploy­ment benefits, but they are more likely to sup­port ex­pen­di­tures on health sys­tems. The rea­sons for these ten­den­cies can be differ­ent in ev­ery cat­e­gory. But it is difficult to find a sin­gu­lar ex­pla­na­tion other than gen­er­a­tional self-in­ter­est, which would ex­plain why older vot­ers tend to be gen­er­ally less sup­port­ive of ex­pen­di­tures that benefit other gen­er­a­tions and pro­jects that have pos­i­tive ex­pected effects in the long run, but costs in the short run. It fits the bill that where it is harder to think of gen­er­a­tional-spe­cific in­ter­ests such as on ques­tions re­lated to an­i­mal pro­tec­tion, women’s rights or ur­ban de­vel­op­ment, there is also no ev­i­dence of a gen­er­a­tion gap.”

(Thee au­thors have a fol­low-up ar­ti­cle here.) How­ever, more work on this seems cru­cial.

Age-weighted vot­ing and the very long term. It’s hard to know to what ex­tent ex­tend­ing poli­ti­cal time hori­zons by a decade or two pro­vides benefits for the very long term. My ini­tial as­sump­tion would be that ex­tend­ing poli­ti­cal hori­zons is some­what benefi­cial for very long-term out­comes, though only weakly so. When I think through par­tic­u­lar is­sues — in par­tic­u­lar wor­ries about risks from tech­nolo­gies like ad­vanced AI and ad­vanced syn­thetic biol­ogy that will only be de­vel­oped in the com­ing decades — poli­tics hav­ing a longer time hori­zon tends to look pretty good. There is enor­mous will­ing­ness-to-pay to avoid ex­is­ten­tial catas­tro­phes (over a trillion dol­lars to miti­gate 0.1 per­centage points of risk, even just look­ing at US cit­i­zens’ will­ing­ness to re­duce chance of their own deaths [1]), so if we think tech­nolog­i­cal risks are cur­rently ne­glected, we need some de­bunk­ing ex­pla­na­tion of why this is so, and my­opic poli­ti­cal de­ci­sion-mak­ing seems plau­si­ble. But more work on this seems cru­cial, too.

Age and wis­dom: I sus­pect that this isn’t a ma­jor con­sid­er­a­tion in the choice be­tween these vot­ing sys­tems: if we wanted a more epis­to­cratic sys­tem, we would move quite far away from ei­ther of the cur­rent sys­tem or the age-weight­ing sys­tem.

But, if we are go­ing this route, there are at least some rea­sons for think­ing that younger vot­ers would make bet­ter de­ci­sions. Ed­u­ca­tion lev­els are ris­ing, so younger peo­ple are on av­er­age bet­ter ed­u­cated; they also have a more re­cent ed­u­ca­tion, so are there­fore more likely to be more up-to-date on con­tem­po­rary knowl­edge. The Flynn effect means that IQ scores are ris­ing, and this may be due in part to gen­uine in­creases in in­tel­li­gence (though the Flynn effect has stalled in the US in re­cent years). As a coun­ter­ar­gu­ment, crys­tallised in­tel­li­gence in­creases with age and, though fluid in­tel­li­gence de­creases with age, it seems to me that crys­tallised in­tel­li­gence is more im­por­tant than fluid in­tel­li­gence for in­formed vot­ing.

(Later Edit). Again, we should bear in mind that, even with the ap­prox­i­mate life-ex­pec­tancy weight­ing, the effec­tive me­dian voter age would move from 55 to 40. So, if we are think­ing through epis­to­cratic con­sid­er­a­tions, the key is­sue is whether 40 year olds make bet­ter de­ci­sions than 55 year olds, rather than whether 60 year olds make bet­ter de­ci­sions than 20 year olds.


Push­ing par­tic­u­lar poli­ti­cal views: One might worry that this pro­posal would have ma­jor par­ti­san con­se­quences — if so, then pro­po­nents of the idea might be bi­ased in favour of it if it is a way of sneak­ing in their favoured poli­ti­cal views, and it would de­crease poli­ti­cal fea­si­bil­ity. And cer­tainly, in the US at the mo­ment, age-weight­ing vot­ing would cause a one-time left­ward swing. But this isn’t true across all gen­er­a­tions. From a Pew Re­search re­port:

“As the Pew Re­search Cen­ter has of­ten noted, it is not always the case that younger gen­er­a­tions are more Demo­cratic. Two decades ago, the youngest adults – Gen­er­a­tion X – were the most Repub­li­can age co­hort on bal­ance, while the old­est – the Great­est Gen­er­a­tion– were the most Demo­cratic. In 1994, 47% of Gen Xers (then ages 18-29) iden­ti­fied with or leaned to­ward the Repub­li­can Party, while 42% iden­ti­fied as Democrats or leaned Demo­cratic. And mem­bers of the Great­est Gen­er­a­tion (then ages 67-81) — fa­vored the Demo­cratic Party over the GOP (49% to 42%)”

Other age-re­lated po­si­tions can be sur­pris­ing. Though younger peo­ple in the UK refer­en­dum were much more likely to vote in favour of re­main­ing in the EU, younger peo­ple in the Scot­tish in­de­pen­dence refer­en­dum were more likely to vote in favour of Scot­tish in­de­pen­dence.

Poli­ti­cal fea­si­bil­ity: It seems hard to be­lieve that some vot­ers would vol­un­tar­ily give up power. But it’s hap­pened be­fore via suffrage move­ments. And there are ways we could ta­per in the vot­ing weights such that no-one ever has less vot­ing power than they would have had oth­er­wise. Alter­na­tively, we could de­lay the im­ple­men­ta­tion of the age-weight­ing, ex­ploit­ing time bi­ases: if age-weight­ing only be­gins in twenty years’ time, then the older gen­er­a­tion have lit­tle to lose by vot­ing in its favour.


Thanks to Aron Val­lin­der, Zach Groff, Ben Grodeck, the other Global Pri­ori­ties Fel­lows and staff at the Global Pri­ori­ties In­sti­tute for helpful dis­cus­sion of this idea.


[1] The value of a statis­ti­cal life in the US is in the range of $3-$9 mil­lion dol­lars. Us­ing the low es­ti­mate, among 350 mil­lion cit­i­zens, the US as a whole should be will­ing to spend over $1 trillion to miti­gate an ex­tinc­tion risk by 0.1 per­centage points.