Electoral Savings Accounts: One Person, One Vote (but you can save it for later)

The conversation around “high information” vs “low information” voters hypothesizes a world where voters lie somewhere on a spectrum of “well-informed” to “uninformed”:

Those who fret about “low-information voters” dislike that low-information and high-information voters all count the same at the ballot box, feeling this dilutes the opinions of those who are well-informed about the issues.

Segmenting voters into “high” and “low” information buckets oversimplifies, however, by dropping a dimension — time.  A particular voter’s informedness and enthusiasm (we’ll treat them the same, as a first-order approximation) vary widely over time and life circumstances:

The graph is different for everyone, but in this example, our voter partied through college, became more politically engaged after getting their first job, disengaged when their life got busy, and re-engaged after their kids went to college.

(Of course, we could break this down further into individual issues a voter cares about.  Interest in taxation, education, and environmentalism etc wax and wane with personal circumstances)

How does enthusiasm translate into voting patterns?  Our voter’s voting history may look something like this:

There’s clearly a correlation with enthusiasm and votes cast; in years where the voter is completely disengaged, he/she will likely not bother to vote at all. But in all but the most-disengaged years, the voter will cast the legal maximum of 1 vote.

But if we could design an optimized democracy, is this how we would structure representation?  Probably not.  In a theoretically optimized democracy, voters would cast a number of votes corresponding to how informed and enthusiastic they were about the issues at hand:

Is that possible?

Saving votes for later

We can’t just ask voters how informed they are (poll tests have a sordid history) or how enthusiastic they are (there’s no advantage in being honest).  And any system of buying & selling votes, to rebalance between people, is prone to corruption and disenfranchisement.

But we could let voters cast multiple votes by saving votes, and borrowing future votes, from themselves:

• When a voter doesn’t choose to vote, the vote is “saved”, and they are free to use it in the future
• A voter can “borrow” votes from up to 10 years in their future
• When casting a vote, a voter can spend as many votes as they have available for a race — either dipping into their bank, or borrowing from their future.

This allows our example voter tocast votes that match their enthusiasm curve, with a bit of saving and borrowing:

• In college, our voter was busy partying, and didn’t really care about the issues.  So they didn’t feel any pressure to vote —but there’s no disenfranchisement, because the votes are saved for later.

• Once they sobered up and got a job, they really cared about the issues.  Maybe about taxes, or climate change, or both.  Not only did they spend the votes they’d banked from college, but they borrowed from elections into their 30’s.

• In middle age, they again disengaged, but again not to any permanent disenfranchisement, because once their kids went to college, they were free to spend off their banked votes (or, just save them up).

Electoral Savings Accounts

The details of borrowing and saving votes sounds complicated, but the implementation is actually pretty simple:

• When a voter turns 18, they start with an Electoral Savings Account (ESA) of 10 votes — ie, allowing them to “borrow” ten years into the future 
• Each election, the voter’s ESA increases by one.
• Each election, each race, voters may spend anywhere between none and all of the votes in their ESA balance.
• Voters hold separate ESAs for each elected position; a voter can vote for school board candidates while abstaining from the presidency, or vice versa.

The last point, when coupled with the fact that people move to new jurisdictions, requires some inter-jurisdictional coordination, to categorize Seattle’s city council election in the same “bucket” as New York’s city council election.  But for the most part, the important races have direct analogues in other cities across the country, and it would not be challenging to build a national mapping from one elected position into a known bucket.

Why are ESAs good for democracy?

The headline reason for ESAs is that they align votes with the times voters are most interested, but there are other reasons they would incentivize a healthy democracy:

ESAs reward honesty by political parties (and punish dishonesty)

It is common for political parties, in their public stances and media advertisements, to frame every race in every election as a highest-priority, life-or-death issue (“the most important election of your lifetime”).  Currently, there is no downside to doing so, because angry voters are good for fundraising.

But using ESAs, a party which hypes up the importance of a non-critical election risks misleading their voters into wasting their entire ESA on an unimportant race.  On the other hand, a party which rightly acknowledges that their opponent is a boring centrist, can save up a war-chest of ESAs their voters can spend on a later, more important, election.

“Voting against everyone” isn’t self-disenfranchisement

Currently, when party primaries produce two terrible candidates, centrist voters are left with two unappealing options:

• Vote for the slightly lesser evil
• Voting for nobody

The second option — spoiling a ballot, or just not showing up — is usually unappealing because it amounts to self-disenfranchisement, and sends no clear message to the candidates.

But if an abstained vote goes directly into your ESA, there’s a great reason to skip an election to punish a slate of bad candidates — you can spend the vote later, on a candidate you actually like. 

Check on tyranny-by-majority

Even districts which consistently vote with 45% – 55% splits in a First Past The Post (FPTP) system are considered non-competitive districts, because the party with 55% of the vote almost always wins.  This is a bad deal for the 45% of the population who, despite having 45% of the population, receive 0% of the representation.

ESAs increase the representation of the minority by allowing them to use their votes “when it matters”.  Instead of constantly throwing their votes away on 45-55% elections, the minority party can save up votes to flip the election when an especially viable candidate is on the ballot:

(here, abstaining from voting, in the years with light green, and double-voting in the years with light red)

Disincentivize very polarizing candidates

When a candidate is running against voters who have a substantial ESA pool available, it does not pay to be antagonistic or polarizing. In a FPTP system — what we have now — winning 51% of the vote is Good Enough, and it is often good tactics to make the remaining 49% hate you.  But this is bad for society overall.

The problem is that in a FPTP system, “49% of the population who hate your guts” is electorally indistinguishable from “49% of the population that mildly dislikes you”.  Being very, very angry doesn’t matter.  But if the very angry minority has an ESA balance to spend, they can punish specifically infuriating candidates with electoral upsets.

What’s more, in an electorally efficient system — where candidates and voters behave rationally — this threat of upset votes via ESA spending is enough to motivate inclusivity (or at least a lack of outright antagonism).  And then when the ESAs aren’t actually spent, they’ll continue to motivate inclusivity in the next election, and so forth.

It’s healthy to not care about politics for a few years

High-intensity interest in politics is not good for mental health.  Allowing voters to check out of politics for a few years without sacrificing their representation relative to those who stay involved, allows voters to optimize for their own wellbeing.

Summary

ESAs are a simple mechanism, but would fundamentally change the dynamics between voters and elections for the better.  Instead of making votes a use-it-or-lose it opportunity — which cuts out voters who don’t have the time or energy  to research and vote for a whole slate of candidates — it trusts voters with a resource that they can spend when and where they please.

Given that the fundamental premise of democracy is that we do trust the people, it seems likely we could make democracy even more robust by trusting voters to cast ballots not just if, but when they see fit.

Appendix: Variations on ESAs

ESAs as proposed above are “as simple as possible”, but there may be opportunities for refinement at the margins, at the cost of higher complexity:

Expire banked votes

ESA votes as described compound neither positively or negatively over time; 1 vote saved in 1980 can be spent as 1 vote in 2025.

An (IMO unlikely) but possible scenario is if vote-hoarding becomes a destabilizing problem, because voters regularly procrastinate instead of casting votes.  A gentle way to nudge voters into voting sooner rather than later would be to cap the number of years a vote can be banked — for example, a vote not spent within 10 years would expire.

Age-cap borrowed votes

By giving voters a 10-year window of future votes to borrow (that is, by initializing their ESA with 10 votes), we’ve likely inflated the total vote supply.  This is because at the end of a person’s life, they are likely to have spent down their pool of votes, borrowing votes from years they are not alive.

This isn’t catastrophic, but if we wanted to re-normalize the total vote count, ESAs could stop accumulating votes for a corresponding decade, for example between the ages of 60-70 (since the majority of voters will make it to age 70).

Spend votes fungibly across races

If we are allowing voters to spend their votes fungibly across time — because their enthusiasm waxes and wanes over time — a natural extension is to allow voters to spend their votes fungibly across elected positions, in alignment with their enthusiasm.

There are complications in this system (should a vote for a municipal sewage administrator be equivalent to a vote for a senator?), but this is actually the same as quadratic voting, a system fully compatible with ESAs.