Age-Weighted Voting

post by William_MacAskill · 2019-07-12T15:21:31.538Z · EA · GW · 39 comments

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:

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:

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.

39 comments

Comments sorted by top scores.

comment by Paul_Christiano · 2019-07-12T16:37:38.710Z · EA(p) · GW(p)

I like the goal of politically empowering future people. Here's another policy with the same goal:

  • Run periodic surveys with retrospective evaluations of policy. For example, each year I can pick some policy decisions from {10, 20, 30} years ago and ask "Was this policy a mistake?", "Did we do too much, or too little?", and so on.
  • Subsidize liquid prediction markets about the results of these surveys in all future years. For example, we can bet about people in 2045's answers to "Did we do too much or too little about climate change in 2015-2025?"
  • We will get to see market odds on what people in 10, 20, or 30 years will say about our current policy decisions. For example, people arguing against a policy can cite facts like "The market expects that in 20 years we will consider this policy to have been a mistake."

This seems particularly politically feasible; a philanthropist can unilaterally set this up for a few million dollars of surveys and prediction market subsidies. You could start by running this kind of poll a few times; then opening a prediction market on next year's poll about policy decisions from a few decades ago; then lengthening the time horizon.

(I'd personally expect this to have a larger impact on future-orientation of policy, if we imagine it getting a fraction of the public buy-in that would be required for changing voting weights.)

comment by William_MacAskill · 2019-07-18T14:09:45.083Z · EA(p) · GW(p)

I really like this proposal! And agree it's radically more tractable than such a major change to voting systems.

comment by John_Maxwell (John_Maxwell_IV) · 2019-07-15T06:15:22.663Z · EA(p) · GW(p)

This is an exciting idea. My guess is that public buy-in would be easier than you might think; my impression is that the horse race aspect of betting markets appeals to the public and creates TV coverage etc. However, I think the surveys could be an issue. I suspect many people responding to surveys about events which happened 10-30 years ago would be doing so with the aim of influencing the betting markets which affect near future policy. There might end up being a meta-game regarding who will answer surveys 10-30 years down the line and what agenda they will have in mind.

comment by Paul_Christiano · 2019-07-15T15:45:14.972Z · EA(p) · GW(p)
I suspect many people responding to surveys about events which happened 10-30 years ago would be doing so with the aim of influencing the betting markets which affect near future policy.

It would be good to focus on questions for which that's not so bad, because our goal is to measure some kind of general sentiment in the future---if in the future people feel like "we should now do more/less of X" then that's pretty correlated with feeling like we did too little in the past (obviously not perfectly---we may have done too little 30 years ago but overcorrected 10 years ago---but if you are betting about public opinion in the US I don't think you should ever be thinking about that kind of distinction).

E.g. I think this would be OK for:

  • Did we do too much or too little about climate change?
  • Did we have too much or too little immigration of various kinds?
  • Were we too favorable or too unfavorable to unions?
  • Were taxes too high or too low?
  • Is compensating organ at market rates a good idea?

And so forth.

comment by Sanjay · 2019-07-15T22:20:37.624Z · EA(p) · GW(p)

I'm also very excited about this idea. The format of the ultimate judgement (i.e. the retrospective evaluation) seems important. A straightforward survey of the population suffers from the problem that teasing out an answer about the quality of a policy is hard, and most people won't have put the time or effort in (even assuming they don't have a hidden agenda, as John_Maxwell_IV highlights). But a survey of experts has its own problems too.

That said, I suspect these issues are surmountable, and would be keen to see this idea turn into action.

comment by Larks · 2019-07-16T03:46:16.379Z · EA(p) · GW(p)

I performed a very cursory literature review on the subject. Overall it seems the psychology research suggests that older people discount the future less than younger people, which might suggest giving their votes more weight.

Usually such a brief perusal of the literature would not give me a huge amount of confidence in the core claims; however in this case the conclusion should seem prima facie very plausible to anyone who has ever met a young boy.

In no particular order:

Age Differences in Temporal Discounting: The Role of Dispositional Affect and Anticipated Emotions:

Advanced age was associated with a lower tendency to discount the future, but this effect reached statistical significance only for the discounting of delayed gains.

Age-Related Changes in Decision Making:

Older adults are not always risk-averse, and their ability to postpone gratification tends to exceed that of younger adults.

Aging and altruism in intertemporal choice:

Research on life span changes in motivation suggests that altruistic motives become stronger with age, but no prior research has examined how altruism affects tolerance for temporal delays. Experiment 1 used a realistic financial decision making task involving choices for gains, losses, and donations. Each decision required an intertemporal choice between a smaller-immediate and a larger-later option. Participants more often chose the larger-later option in the context of donations than in the context of losses; thus, parting with more of their overall capital when the act of doing so benefited a charity. As predicted, the magnitude of this “altruism effect” was amplified in older relative to younger adults. -

Following Advice Because it's Been Paid For: Age, the Sunk-Cost Fallacy, and Loss Aversion:

[Y]ounger adults commit the sunk-cost fallacy more frequently, and make normatively correct decisions less frequently,than older adults when hypothetical sunk costs are at stake ... Younger adults made fewer initial investments of money on the Calorie Estimation Task than older adults. Younger adults demonstrated the sunk-cost fallacy more frequently, overinvested more after demonstrating the fallacy, and made the normatively correct decision less frequently than older adults on the hypothetical self-report measure of the sunk-cost fallacy. Younger adults indicated that they were more averse to hypothetical monetary losses on a Tradeoff Loss Aversion Task and more averse to hypothetical monetary losses on a Delay Discounting Loss Aversion Task. ... Older adults did not demonstrate the sunk-cost fallacy on the Calorie Estimation Task by following more expensive advice more closely than less expensive advice. -

Decision Making in Older Adults: A Comparison of Delay and Probability Discounting Across Ages:

[O]lder adults discounted delayed rewards less steeply than young adult

Acute stress and altruism in younger and older adults:

Even under acute psychosocial stress, older adults make more altruistic choices than younger adults.

Age-related differences in discounting future gains and losses:

Results indicated that impaired older adults discounted the future more than unimpaired older adults. Interestingly, middle-aged adults discounted future gains at a similar rate as impaired older adults, but discounted future losses less than impaired older adults (and similarly to unimpaired older adults).

Age‐related differences in delay discounting: Immediate reward, reward magnitude, and social influence:

Younger adults exhibited higher levels of delay discounting than older adults.

The only evidence I've found that older people might discount more was a Chinese study. I'm not sure why Chinese people would be different in this regard, though obviously their culture is different in many ways - or, the paper might just be wrong.

Age differences in delay discounting in Chinese adults:

The current study aimed to clarify this relationship using a relatively large sample of Chinese adults with a wide age range (viz., 18 to 86 years old). A total of1288 individuals completed the Monetary Choice Questionnaire. Results showed that the rate of delay discounting increased with age across adulthood, with younger participants (18–30 years) discounting lessthan both middle-aged participants (31–60 years) and older participants (over 60 years); and middle-aged participants discounting less than older participants. Furthermore, when the reward magnitude was large, participants were more likely to wait for delayed rewards.
comment by cole_haus · 2019-07-18T22:44:04.062Z · EA(p) · GW(p)

This paper has increased my general skepticism on the accuracy of any estimates of discount rates: Time Discounting and Time Preference: A Critical Review. It has a table listing studies that find discount rates ranging from -6% to ∞% .

comment by William_MacAskill · 2019-07-18T14:07:20.059Z · EA(p) · GW(p)

Hi, thanks so much for doing this! This is really interesting.

Something I think wasn't sufficiently clear from the post itself: even using the the weighting scheme I suggested in the post, that would move the median voter (in the US) from age 55 to age 40. (H/T Zach Groff for these numbers. Note this doesn’t account for incentive effects, of younger people being more likely to go out to vote, which could lower the median age to a little under 40.) And under reasonable assumptions (with the most controversial being single-peaked preferences), the median voter is decisive. So it’s not like 20 year olds are now deciding what happens. On the epistocratic question, then, we should be asking whether we think 40yr olds will make better decisions than 55 year olds; not whether 20 year olds make better decisions than 60 year olds. I'd need to dig into the studies a lot more to determine whether 40 year olds discount more steeply than 55 year olds.

And then, I've only done a quick scan of the studies you link to, but I don't think the discounting literature you're pointing to is actually all that relevant, because the timescales they are looking at are so short: 90 days in one case; up to 6 months in another. Whereas the time horizons for the impact of political decisions, especially the most important ones, are on the order of years or decades - over such timescales, discounting due to risk of death become a much bigger factor than discounting due to impulsiveness / impatience.

Usually such a brief perusal of the literature would not give me a huge amount of confidence in the core claims; however in this case the conclusion should seem prima facie very plausible to anyone who has ever met a young boy.

Again, I think this depends on what timescales we're talking about. Sure, it seems prima facie plausible that someone who is 21 is more likely to prefer $5 today to $10 in a month's time than a 60 year old is. But (on the assumption of self-interest) I'd strongly wager that a 21 year old is more likely to prefer $100 in 40 years' time over $10 in a month's time than a 60 year old is, because the 21 year old is so much more likely to be around and be able to enjoy the benefits.


The altruism and age discussion is interesting, and I agree that if it were borne out it could form part of an epistocratic argument for the age-weighting going the other way around.

comment by Dale · 2019-07-28T20:02:41.597Z · EA(p) · GW(p)
the the weighting scheme I suggested in the post, that would move the median voter (in the US) from age 55 to age 40. (H/T Zach Groff for these numbers. Note this doesn’t account for incentive effects, of younger people being more likely to go out to vote, which could lower the median age to a little under 40.) And under reasonable assumptions (with the most controversial being single-peaked preferences), the median voter is decisive. So it’s not like 20 year olds are now deciding what happens. On the epistocratic question, then, we should be asking whether we think 40yr olds will make better decisions than 55 year olds; not whether 20 year olds make better decisions than 60 year olds. I'd need to dig into the studies a lot more to determine whether 40 year olds discount more steeply than 55 year olds.

If you want to give extra influence to 40 year olds, it probably makes more sense just to give 40 year olds more votes. Otherwise you're putting a lot of faith in one model of how voters work, despite the median voter theorem having lost some of its academic appeal over time (multidimensional preferences, selectorate vs electorate, veto players, heresthetics).

Additionally, if we did give young people lots of extra votes, we'd probably get a Goodheart's Law type situation, where politicians would adopt special policies designed to exploit it - like promising student debt forgiveness, or to ban tuition fees (the latter of which seemed to have been quite successful at manipulating UK students to vote for the Democrat Party in 2015!)


comment by Sanjay · 2019-07-15T07:03:50.382Z · EA(p) · GW(p)

David Moss and I recently conducted a study with about 500 participants looking at the extent to which people place moral weight on the far future.

The study found that older people give much less moral weight to the future.

The study included the following questions:

  • Is it better to save (A) 1 person now or (B) 1/2/1,000/1,000,000 people 500 years from now? (This is 4 different questions, one after the other, with differing numbers of people stated in option (B))
  • How far do you disagree or agree (on a 7-point scale) that:
  • “Future generations of people, who have not been born yet, are equal in moral importance to people who are already alive”
  • “We should morally prioritise helping people who are in need now, relative to those who have yet to be born”
comment by cole_haus · 2019-07-18T22:39:52.265Z · EA(p) · GW(p)

I'm not sure how much you thought about this aspect, but I've recently become extra wary of surveys on this topic (beyond the ordinary skepticism I'd have for questions which are mostly about expressive preferences and not revealed preferences). Time Discounting and Time Preference: A Critical Review has a table listing studies that find discount rates ranging from -6% to ∞% . Even if that doesn't influence you as much as it did me, the paper has some good discussion of different methods of elicitation (which are especially likely to influence results given the difficulty of the domain).

comment by William_MacAskill · 2019-07-18T14:51:45.979Z · EA(p) · GW(p)

Cool! Could you send me a link to the study?

comment by Aaron Gertler (aarongertler) · 2019-08-13T04:29:29.591Z · EA(p) · GW(p)

Was this study ever published anywhere? I'd love to put it up on the Forum (or see it posted with a summary from the authors, if you'd be up for it!).

comment by Sanjay · 2019-08-14T05:57:59.125Z · EA(p) · GW(p)

Thanks for the nudge Aaron. I'm still working on it. Have had a bit more higher-quality data added in, so incorporating that. I'll add it to the EA Forum when I get round to it.

comment by weeatquince · 2019-08-01T22:31:41.005Z · EA(p) · GW(p)

It is an interesting suggestion and I had not come across the idea before and it is great to have people thinking of new innovative policy ideas. I agree that this idea is worth investigating.

I think my main point to add is just to set out the wider context. I think it is worth people who are interested in this being aware that there is already a vast array of tried and tested policy solutions that are known to encourage more long term thinking in governments. I would lean towards the view that almost all of the ideas I list below: have very strong evidence of working well, would be much easier to push for than age-weighted voting, and would have a bigger effect size than age-weighted voting.

Here's the list (example of evidence it helps in brackets)

* Longer election cycles (UK compared to Aus)
* A non-democratic second house (UK House of Lords)
* Having a permanent neutral civil service (as in UK)
* An explicit statement of policy intent setting out a consistent cross-government view that policy makers should think long-term.
* A formal guide to best practice on discounting or on how to make policy that balances the needs of present and future generations. (UK Treasury Green Book, but more long term focused)
* An independent Office for Future Generations, or similar, with a responsibility to ensure that Government is acting in a long term manner. (as in Wales)
* Independent government oversight bodies, (UKs National Audit Office, but more long term focused)
* Various other combinations of technocracy and democracy, where details are left to experts. (UK's Bank of England, Infrastructure Commission, etc, etc)
* A duty on Ministers to consider the long term. (as in Wales)
* Horizon scanning and foresight skills, support, tools and training brought into government (UK Gov Office for Science).
* Risk management skills, support, tools and training brought into government (this must happen somewhere right?).
* Good connections between academia and science and government. (UK Open Innovation Team)
* A government body that can support and facilitate others in government with long term planning. (UK Gov Office for Science, but ideally more long term focused).
* Transparency of long term thinking. Through publication of statistics, impact assessments, etc (Eg. UK Office for National Statistics)
* Additional democratic oversight of long term issues (UK parliamentary committees)
* Legislatively binding long term targets (UKs climate change laws)
* Rules forcing Ministers to stay in position longer (untested to my knowledge)
* Being a dictatorship (China, it does work although I don’t recommend)


I hope to find time to do more work to collate suggestions and the evidence for them and do a thorough literature review
(If anyone wants to volunteer to help then get in touch). Some links here. My notes are at: https://docs.google.com/document/d/1KGLc_6bKhi5ClZPGBeEQIDF1cC4Dy8mo/edit#heading=h.mefn6dbmnz2 See also:
http://researchbriefings.files.parliament.uk/documents/LLN-2019-0076/LLN-2019-0076.pdf


As an aside I have a personal little bugbear with people focusing on the voting system when they try to think about how to make policy work. It is a tiny tiny part of the system and one where evidence of how to do it better is often minimal and tractability to change is low. I have written about this here:
https://forum.effectivealtruism.org/posts/cpJRB7thJpESTquBK/introducing-gpi-s-new-research-agenda#Zy8kTJfGrY9z7HRYH

Also my top tip for anyone thinking about tractable policy options is to start with asking: do we already know how to make significant steps to solve this problem, from existing policy best practice. (I think in this case we do.)

comment by weeatquince · 2019-08-01T23:03:35.306Z · EA(p) · GW(p)

Also I do plan to write this up as a top level post soon

comment by PeterMcCluskey · 2019-07-12T19:25:35.245Z · EA(p) · GW(p)

War is more likely when the population has a higher fraction of young men (e.g. see Angry Young Men Are Making the World Less Stable ). That's doesn't quite say that young men vote more for war, but it's suggestive.

More war could easily overwhelm any benefits from weighted voting.

comment by Holly_Elmore · 2019-07-13T23:13:33.497Z · EA(p) · GW(p)

FWIW, I think the young lacking life experience and crystallized intelligence is pretty clutch. This argument rests on the young having not only a greater stake in future but being able to make sensible decisions about what to do with it. I would at least suggest that 18-25 yo voters not have a multiplier.

I do like reducing the influence of the old who know very well when voting that, for instance, climate change will not really affect them. But I think any vote weighting scheme has to take stakeholding and competence into account.

comment by John_Maxwell (John_Maxwell_IV) · 2019-07-15T05:44:10.988Z · EA(p) · GW(p)

I would at least suggest that 18-25 yo voters not have a multiplier.

Yes. As a reductio ad absurdum of Will's idea, why not give toddlers an extreme multiplier? Well, we know toddlers don't make good judgements. But it's not like your ability to make good judgments suddenly turns a corner on your 18th birthday. So as long as we're refactoring voting weights for different ages, we should also fix the 18th birthday step function issue, and create a scheme which gradually accounts for a person's increased wisdom as they age.

[Edit: A countervailing consideration is that if you make your scheme too wonky, it may not gather broad support.]

(I also think randomly selecting a small number of voters jury selection-style, to address the public goods problem inherent in becoming an informed & thoughtful voter, would probably be a higher-leverage improvement... but that's another discussion.)

comment by Jemma · 2019-07-15T10:04:33.535Z · EA(p) · GW(p)

You mean like sortition? https://www.sortitionfoundation.org/what_is_sortition

comment by David_Moss · 2019-07-15T16:21:19.587Z · EA(p) · GW(p)

There is already a proposal to use sortition to form a third legislative house, of citizens who would have responsibility for deliberating on whether legislation would harm future generations: Rupert Read's 'Guardians of the Future' (2012)

This seems more promising than re-weighting the value of votes of certain groups whose self-interest is presumed to lie more in the future given that i) voters tend not to vote much on the basis of self-interest, ii) to the extent that slightly younger generations have a greater interest in the future it is only in the very near future, which seems roughly equally compatible with disastrous policies iii) we have little, if any, reason to suppose that younger generations are epistemically capable of judging what policy would best serve their self-interest >50 years out.

The deliberative council idea has advantages over vote reweighting on all three counts: i) the citizens would be tasked explicitly with judging whether policies would aid or harm the future, rather than voting in whatever way in the hope that their vote proxies future interests, ii) they would be tasked with considering the long run future not just their self-interest (which extends maybe 50 years into the future, but which, due to time preference, might on average be a lot shorter), iii) such a deliberative council would have ample time and access to expertise (deliberative fora tend to give participants access to a variety of experts to help inform their deliberations) and be explicitly and implicitly (e.g. by the setup) to deliberate about what would produce the best interests- these kinds of setups have been widely used participants seem to tend to deliberate pretty well and reach relatively informed judgments (at least compared to the typical voter): see some case studies.

That said I think there may still be grounds to reject even this proposal, primarily that one may still be concerned about (iii) the epistemic question, even in these comparatively ideal circumstances.

comment by William_MacAskill · 2019-07-18T14:16:18.442Z · EA(p) · GW(p)

I mentioned this in response to Larks too, but one thing to bear in mind is that even using the the weighting scheme I suggested in the post - which seemingly strongly favors young people - that would move the median voter (in the US) from age 55 to age 40. So, at least assuming the median voter theorem is approximately accurate in this context, the key epistocratic question is about 40yr olds vs 55yr olds.


And if I had to choose now, I would also prefer a tapering system, where vote-weight starts off lower, then increases, and then decreases again. A benefit of that system is that you could make the 'voting age' a gradual progression rather than an immediate jump. Perhaps 12yr olds get a very weak vote, which scales up until 25, then scales down after 35.

comment by Holly_Elmore · 2019-08-02T19:35:58.849Z · EA(p) · GW(p)
one thing to bear in mind is that even using the the weighting scheme I suggested in the post - which seemingly strongly favors young people - that would move the median voter (in the US) from age 55 to age 40.

How do you get this result? Are you just saying with these multipliers applied to the current age distribution of voters, the median US vote would be cast by a 40 yo? Or if this anticipating the response to the multipliers? Like, for example, does this take into account that young people would probably vote more if their votes counted 6x more?


I'm not knocking the overall idea, but I am skeptical that young people will be that much better at resisting short-term political temptations than old people. If young people got huge vote multipliers, politicians would only pander to their weaknesses more. I guess like most people commenting here I have the most faith in middle-aged people. I like the idea of a more gradual tapering up and down of the vote multiplier, but a system that complicated is probably doomed.

Maybe parents should get huge vote multipliers. Seems to me they usually care about the future a lot more than the young people who are on track to outlive them.

comment by cjcorliss · 2019-07-23T21:59:04.736Z · EA(p) · GW(p)

Bryan Caplan's book "The Myth of the Rational Voter" explains that voters being merely ignorant or irrational is not a big issue. The uniformed voters will make random mistakes in voting that cancel each other out, and elections are still decided by the median informed voter. If that is true, younger voters' greater ignorance (/higher intelligence) will cause them to contribute less (/more) to the pool of informed voters.

What we should really care about are biases, where people are consistently making mistakes in one direction, that are common across the population (or the age group in this case). Age might be a factor. Caplan proposes four biases: Anti-market bias, Anti-foreign bias, Make-work bias, Pessimistic bias.

https://en.wikipedia.org/wiki/The_Myth_of_the_Rational_Voter

comment by Paul_Christiano · 2019-07-12T16:16:14.019Z · EA(p) · GW(p)
It would mitigate intertemporal inconsistency

If different generations have different views, then it seems like we'll have an same inconsistency when we shift power from one generation to the next regardless of when we do it. Under your proposal the change happens when the next generation turns 18-37, but doesn't seem to be lessened. For example, the brexit inconsistency would have been between 20 years ago and today rather than between today and 20 years from now, but it would have been just as large.

In fact I'd expect age-weighting to have more temporal inconsistency overall: in the status quo you average out idiosyncratic variation over multiple generations and swap out 1/3 of people every 20 years, while in your proposal you concentrate most power in a single generation which you completely change every 20 years.

Age and wisdom: [...] 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. 

Another counterargument: older people have also seen firsthand the long-run consequences of one generation's policies and have more time to update about what sources of evidence are reliable. It's not clear to me whether this is a larger or smaller impact than "expect to live through the consequences of policies." I think folk wisdom often involves deference to elders specifically on questions about long-term consequences.

(I personally think that I'm better at picking policies at 30 than 20, and expect to be better still at 40.)

comment by Larks · 2019-07-12T16:42:51.153Z · EA(p) · GW(p)

Yes - hence the standard pair of arguments:

The young people who will have to live with the consequences of Brexit voted to Remain.

vs

The older people with actual experience of living outside the EU voted to Leave.
comment by William_MacAskill · 2019-07-18T14:30:03.547Z · EA(p) · GW(p)
(I personally think that I'm better at picking policies at 30 than 20, and expect to be better still at 40.)

Again, see comments to Holly and Larks about where the median voting age ends up. I'm going to add that point as an edit into the main post.


Under your proposal the change happens when the next generation turns 18-37, but doesn't seem to be lessened. For example, the brexit inconsistency would have been between 20 years ago and today rather than between today and 20 years from now, but it would have been just as large.

This is a good point, and my post overstates the case on this. There is still an important difference, though, which is that if there's a difference between the views of 60 year olds and 30 year olds, we can foresee there will be an intertemporal inconsistency and can choose to avoid it. Whereas if there's a difference between the views of 30 year olds and 0 year olds we (presumably) don't know about it and can't do anything about it.


There's another intertemporal inconsistency consideration: If we assume rational self-interest and risk-aversion (just in the sense of consumption having diminishing utility), we should expect that earlier on in life, people will prefer more redistributive policies (e.g. progressive tax and redistribution, social safety net for disabilities, weighing costs to prisoners of harsh penalties against benefits of lower crime rate). This is because they have uncertainty about how much they are going to earn, whether they are going to end up disabled, whether they'll commit a crime. Whereas older people know how things have turned out for them, and face much less risk: those who are wealthier will no longer support redistributive policies; those who know they aren't going to jail will prefer harsh on crime policies. The early age-weighting is therefore one way to hold people to the decisions they'd make ex ante. I think it's up for debate how much that matters, but it's appealing to me - I'm generally attracted to veil of ignorance arguments, and this makes political decision-making slightly more veil-of-ignorance-y.

comment by Larks · 2019-07-15T03:28:05.453Z · EA(p) · GW(p)
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.

First of all I think using live political examples like this is not a great idea [LW · GW].

It also seems possible to me (though I am less certain) that the 'older people are rightfully more sceptical of official cost estimates' alternative theory I described in this comment [EA(p) · GW(p)] might apply here. One of the key arguments against Brexit was that it would lead to a high degree of short-term disruption, due to uncertainty, disruptions to supply chains, and a loss of export markets on the continent. In contrast, many of the benefits, like protection from further institutional decline in the EU, were longer-term.

Yet after the fact it is worth noting that many of the anti-Brexit forecasts for immediate large negative consequences have been proved wrong, at least so far. For example the Treasury forecast that unemployment would rise by 520,000 after a Leave vote, whereas in actual fact the UK labour market has improved significantly since then.

There's also a big difference between wanting to Remain and wanting to Rejoin later on:

  • Once the UK has left, the transition issues will be sunk costs - indeed transition costs will become a reason to not rejoin!
  • Status quo bias will switch to supporting independence.
  • The outside view also suggests the UK would want to remain independent, as the other independent western european countries like Norway, Iceland and Switzerland have become if anything less likely to join now than they were in the past (see for example here and here).
  • This also seems to be the case in the UK - for example, the idea of joining the Euro, despite being supported by the Conservative Party in 1990 and Tony Blair later, has been totally discredited even among Remain supporters.

So overall I don't think Brexit-related temporal inconsistency is a great reason to support increasing the weight on younger voters.

comment by William_MacAskill · 2019-07-18T14:40:34.563Z · EA(p) · GW(p)
First of all I think using live political examples like this is not a great idea [LW · GW].

I don't think that a blanket ban on live political examples is a great norm. There are definite risks from tribalism from doing so, but we also just have a lot more information with which to test our views (compared to, say, how age-weighted voting would have affected the French Revolution). If we're worried about tribalism, we should just call tribalism out directly, rather than ban certain topics.

In this particular case, I found that thinking about Brexit and the Scottish Independence Referendum helpful to test my starting intuitions. In particular, it somewhat weakened my adherence to my starting assumption of rational self-interest of voters' political positions - I don't really see the age-related discrepancies in people's votes on Brexit and Scottish Independence as being well explained by whether the position involves short-term benefits for long-term harms. (Rather than, say, by how much weight one puts on national sovereignty, which is a political view that might just go in and out of fashion.)

comment by HedoniumShockwave · 2019-07-14T16:25:15.653Z · EA(p) · GW(p)
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.

This post's argument seems to rely pretty heavily on the assumption that people vote in their self-interest. Yet, as you note, "in general, voting behaviour isn’t well-explained by the ‘self-interested voter’ model." For other readers interested in this debate I'd suggest Caplan ([2007](https://en.wikipedia.org/wiki/The_Myth_of_the_Rational_Voter)),

The evidence cited that older generations do vote in their self-interest (looking specifically at cases of direct democracy, which is arguably a somewhat peripheral part of the political systems in questions) seems exceedingly weak compared to the evidence against self-interested voting in general.

For example, it relies on subjectively coding rather broad categories of decision as consistent with narrow self-interest by the elderly, neutral or opposed to narrow self-interest. In most cases (see table 2), this just seems to come down elderly voters being opposed to more spending (except for health). It seems rather speculative to chalk this up to generational self-interest, especially in the face of a lot of evidence that voters are generally not self-interested.

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.

It's also worth noting explicitly that the dynamics of politicians (potentially) voting in their self-interest should be quite different from ordinary voters (potentially) voting in their self-interest. Certainly a 61.8 year old senator might slightly personally benefit from voting for, say, higher health spending, and thereby be more motivated to do so than a 47.5 year old senator, but it seems like in most cases we would expect these effects to be swamped by other incentives politicians have e.g. maintaining political power.

Imagine, at the limit, someone who was voting on their deathbed. They would only have moral concerns to guide their decision.

I'm not sure if this is of substantive importance to your argument, but this doesn't seem true on either the conception of self-interest typically employed in these debates or the folk concept of self-interest. Even "narrow self-interest" is typically taken to include the interests of one's immediate family/household. It seems like we would typically say that someone who, on their deathbed, acts to ensure that their money to go to their family rather than someone else's family, is acting self-interestedly.

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.

I would expect this to have some effects which may be considered deleterious.

For one, it seems like this is not just changing the weighting of the vote, but it's also changing the incentives to vote (and to be politically engaged more broadly) i.e. presumably older generations would be less likely to vote and younger generations more. One might view that as a feature rather than a bug, working in line with the direct effect of the vote weighting (especially since it's the opposite of current trends). However, it would likely also decrease political (and perhaps more broadly, civic) engagement in older generations, including their incentives to engage in political deliberation and debate. This would plausibly have negative effects, both by making the older generations less informed and deliberative (and so plausibly reducing the quality of their vote), but also reducing their input to the deliberative system as a whole.

A further problem is that some evidence suggests that, weighting votes differently (on the presumption that votes reflect individual interests might itself help to create self-interested behaviour that there wasn't before (http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.597.7508&rep=rep1&type=pdf)

comment by Jon_Behar · 2019-07-15T20:32:45.599Z · EA(p) · GW(p)
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.

I think you’re being pretty cavalier about the “intelligence vs. wisdom” issue. Paul Christiano’s comment [EA(p) · GW(p)] “I personally think that I'm better at picking policies at 30 than 20, and expect to be better still at 40” rings very true to me (I’m 40), and I’m pretty sure my friends (mostly in 30s and 40s) would emphatically feel the same way. I’m curious about the age of the oldest person you got feedback from, and what they thought about this idea.

Re: epistocracy, it’s true there are policies that could increase the average intelligence and/or wisdom of the electorate. But those are typically the same policies that have historically (and/or currently) been used to disenfranchise marginalized people. As one example of the baggage attached to these policies, here’s how Wikipedia describes how literacy tests have historically been used in the US:

From the 1890s to the 1960s, many state governments in the United States administered literacy tests to prospective voters purportedly to test their literacy in order to vote. In practice, these tests were intended to disenfranchise racial minorities. Southern state legislatures employed literacy tests as part of the voter registration process starting in the late 19th century. Literacy tests, along with poll taxes, residency and property restrictions and extra-legal activities (violence, intimidation)[2] were all used to deny suffrage to African Americans. The first formal voter literacy tests were introduced in 1890. At first, whites were generally exempted from the literacy test if they could meet alternate requirements that in practice excluded blacks, such as a grandfather clause or a finding of "good moral character." (emphasis added)

Personally, to give future people more representation I’d favor legislation like limits on budget deficits, overall debt levels, and programs with upfront benefits but big deferred costs (e.g. it’s easy to offer a generous pension if all the costs come down the road). One of the easiest and most common ways to “steal” resources from future generations is to run up big debts they’ll need to pay off.

comment by Adam_Gibbons · 2019-07-16T01:35:23.350Z · EA(p) · GW(p)

Interesting post! Some comments:

(1) "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."

I think this ignores models of rational voter ignorance. Levels of political ignorance have been consistently high since empirical research into the phenomenon first began, even while education levels have increased. Why? Well, if you take seriously models of rational ignorance, voters in large electoral democracies are simply not incentivized to acquire whatever information helps them to vote competently - the costs of acquisition are too high, while the benefits are too low. Would age-weighted voting ameliorate this problem? I'm not confident that it would. Similarly, if political (epistemic) rationality is not incentivized (or actively disincentivized) in modern democracies, the political preferences of younger people may not be appropriately responsive to the needs of future generations. Ensuring an interest in future generations seems insufficient; regarding such interest in the right way is what we need.

(2) Recent research in political science seems to show that policymakers and legislators are mostly free to enact policy as they see fit, with minimal responsiveness to the political preferences of the electorate. Voter preferences do exert some pressure, but most policymaking is done out of the public eye. If that's right, we might get more bang for our buck if we focus on efforts to improve the long-term decision-making abilities of legislators, bureaucrats, policymakers, and the like. Of course, others (like Jess Whittlestone) have already advocated for improved institutional decision-making as a priority. That line of research is deeply important. However, we might also consider institutional, structural reforms that foster long-term decision-making. I'm thinking of proposals like Bruce Tonn's Futures Congress, Berggruen and Gardel's work on intelligent governance, Alex Guerrero's lottocracy, and more. It's a truism that incentives matter, but what incentives we possess is in part a function of the broader structures we inhabit; if we change the institutional structure, we change the incentives. If feasible, then, we should consider changing institutional structure.


comment by Halstead · 2019-07-17T08:34:20.497Z · EA(p) · GW(p)

You might also want to look at Brighouse and Fleurbaey's Democracy and Proportionality where they argue that people should get power in proportion to their stake in a decision.

comment by William_MacAskill · 2019-07-18T14:53:35.648Z · EA(p) · GW(p)

Thanks! I hadn't seen that before!

comment by Larks · 2019-07-15T03:02:39.452Z · EA(p) · GW(p)
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.”

I don't think this is a very fair summary. Taking the environmental example, for example, older people are more supportive of nuclear power:

From the whole model in Table2, we know that five sociodemographic variables show a significant impact on the acceptance of nuclear power. Age and education have a positive impact on acceptance, whereas gender, social class, and residence have a negative impact on it. As people become older, they reveal more positive attitude toward nuclear power. Such an older-age effect confirmed the research findings by Slovic et al. [8]. Enlightenment from education increases support for nuclear power. According to Kim et al. [16], higher education brings out more acceptance. - Comparative Analysis of Public Attitudes toward Nuclear Power Energy across 27 European Countries by Applying the Multilevel Model

Yet being pro-nuclear power seems like the more long-termist solution. Nuclear power stations are very expensive, and cost billion of dollars to create, but once built they can generate electricity very cheaply and reliably for a very long time - and with essentially no carbon emissions. The relative opposition of young people suggests that either older people are not voting in a more short-sighted fashion, or that the inexperience of young people outweighs this effect.

Similarly, lets look at the core case study that Ahlfeldt et al. discuss: Stuttgart's new train station. They find that older people tended to vote against the transit investment, and suggest this is because old people wouldn't be alive for as long to reap the benefits.

However, they omit to mention that, like many large government projects, the project ended up running dramatically over-budget and behind schedule:

According to newly released official calculations, the total costs for the Stuttgart 21 railway station could reach EUR 8.2 billion, twice as high as the estimate of EUR 4.5 billion presented at the start of the project. The increase was driven by higher building costs due to a booming German construction industry and tight domestic labor market, problems concerning the building site, and a desire by the management of Deutsche Bahn to create a financial buffer against the risk of future price increases. The new figures released were based on an assessment by the accountancy firm Price Waterhouse Coopers (PwC) and the engineering company Emch+Berger.
Additionally, the completion date of the project was pushed back for a fifth time to 2025, four years later than originally anticipated. (source)

It seems reasonable to think that older voters would have developed a sense of suspicion over time about upfront estimates - an intuitive sense of planning fallacy, and problems with the political process. With the benefit of these more accurate views, it would be rational for them to be more likely to reject the project; the problem lies in the naïvety of the young voters.

The authors seem to have considered a similar idea... but strangely assume that experience could only make one more in favour of large projects:

Experience with similar projects in the past combined with a sense of morale could theoretically imply that the likelihood of support could increase with age,if the project is perceived as socially desirable.

What's more, that age brings wisdom should have been an obvious hypothesis, as if you look at the controls in their tables, you'll see that higher levels of education also lead to opposition to the project (assuming I read the table correctly; the authors conveniently did not describe this result in the text). So basically they found that both more educated and more experienced people opposed the project, a project that did indeed fall short of expectations, and yet attributed this to selfishness instead!

comment by Larks · 2019-07-14T20:03:15.767Z · EA(p) · GW(p)

Thanks for writing this! I found it very interesting, and certainly far above most EA posts about politics.

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

One effect of this is to create a big discontinuity at 18 years, out of all proportion to the marginal increase in knowledge and wisdom. One option would be to have the peak later, as a rough sum of the knowledge curve and the incentive curve, e.g.:

18-27yr olds: 1x voting weight
28-37yr olds: 2x voting weight
38-47yr olds: 3x voting weight
48-57yr olds: 3x voting weight
58-67yr olds: 2x voting weight
68+yr olds: 1x voting weight

An alternative fix would be to give parents of minor children extra votes. This would aim to exploit the fact that parents typically exhibit really high levels of altruism towards their children, so arguably actually have longer-term horizons than childless 18yr olds despite lower life expectancy. We see this behaviour in voting through parents apparent obsession with good schools and safety for their children. So we might have something like:

0-7yr olds: parents +4 votes each
8-17yr olds: parents +3 votes each
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
comment by William_MacAskill · 2019-07-18T14:54:40.458Z · EA(p) · GW(p)

Thanks! I also do favour a tapering-in system, if I had to guess now. And I think that surrogacy voting is pretty interesting, too.

comment by SoerenMind · 2019-07-12T21:26:05.907Z · EA(p) · GW(p)

Many policies are later revoked and aren't about trading off present vs future resources (e. g. income redistribution). So those who are still alive when a policy's effects stop got more than their fair share of voting power under this proposal if I understand correctly. E. g. if I'm 80 when a policy against redistribution comes into effect, and it's revoked when I die at 84, my 1x vote weighting seems unfair because everyone else was also just affected for 4 years.

comment by Denkenberger · 2019-08-01T02:01:35.778Z · EA(p) · GW(p)

Interesting - have you considering age-weighted voting in the context of quadratic voting?