Make. All. Votes. Count. (Part I: PR)

A couple of days ago, in Cooper v. Harris, the Supreme Court ruled that North Carolina’s racial gerrymander was unconstitutional. In one of the two districts in question, district 1, the legislature’s unconstitutional action was to redraw a district that was already typically won by a Black congressperson — currently, George “GK” Butterfield—to include more Black voters.

In other words, it’s unconstitutional to have too many people in the district who are happy with GK as their congressperson. If there aren’t enough people there who dislike him, then all the extra ones who do like him are basically wasting their votes; packing them into the district is just a way to decrease the chances a Democrat will win in a neighboring district.

Given America’s “first past the post” (FPTP) voting mechanism, the only way to be fair on a state level is to make sure that there are enough losers wasting their votes on the district level.

That’s insane.

I’m not criticizing the Supreme Court’s decision here; I think they got it right. What’s broken is a system which is built for wasted votes. In every single district, at least half the votes are wasted—they don’t help swing the result. That includes any votes above what it takes to win, as well as any votes for any losing candidate.

The bad effects don’t end there. FPTP is also what makes gerrymandering possible, meaning incumbent politicians can cook the books in their favor. Partisan primaries become more important than general elections, leading to radicalized parties unable to find compromise. And the whole system basically blows up in everybody’s face as soon as there are more than two parties with serious support. (Bonus shrapnel: everybody blames the other party that’s most similar in ideology, and forgets about fighting their real enemy.)

Most of the world doesn’t have this problem. Over 2/3 of democratic countries use some form of proportional representation (PR); that is, voting mechanisms where only a small fraction of the votes are wasted, where gerrymandering is impossible, and where the political spectrum doesn’t always begin and end with the same two parties.

Sounds good, right? So, why can’t we do that here? Well, there are some people who like the status quo. Political consultants are experts in divide and conquer; big donors like the fact that a political duopoly gives the voters less power and their donations more. And so whenever a reform is proposed, they’ll focus on its small flaws, in order to cover up the fact that it would fix a massive problem.

OK, then. All we have to do is find a PR mechanism without even the smallest of downsides. Should be a piece of cake, am I right?

Of course, it’s not actually easy. Other countries’ PR systems have a lot of good aspects, and any one of them would on the whole be a step up for the USA. But there are flaws. Party list systems can make some incumbents entrenched and unaccountable; not because of their real popularity, but just because of their luck in where they’re from or how good they are at insider wheeling and dealing. Mixed-member systems create two classes of legislators, and weakens the link between representatives and constituents. Existing STV (single transferable vote) systems mean complicated ballots that can be a chore to vote with. And some people argue that all three of these would break politics down into numerous tiny, radicalized, one-issue parties, making gridlock actually worse.

But we can do even better. “Mechanism design” is a whole academic field these days. We have enough of a theoretical foundation to pick and choose aspects of different systems, and design one that delivers the huge benefits of PR without giving up on any of the lesser benefits of FPTP.

So here it is, the voting mechanism with no downsides:¹ GOLD voting.² It’s a proportional mechanism that uses the same single-winner districts as FPTP.

  • Ballots list the candidates in the local district, but also have room where voters can write in a candidate from somewhere else.
  • Voters pick one candidate and one of two “transfer methods” which govern where their vote goes if their candidate is eliminated.
  • ——— The first transfer method is “same-party voters”; transferred votes are split among other candidates in the same party, in proportion to the number of votes each got originally. When there are no more candidates from that party, the ballot is exhausted.
  • ——— The second transfer method is “candidate”; transferred votes go in order of a pre-declared list that the original candidate prepared. The first three names on that list are listed on the ballot and the rest of the list for each candidate is available at the polling station. Tied preferences are allowed, and are handled by temporarily splitting up the vote as explained above.
  • Initially, all but the top two (or in rare cases,³ three) candidates in each district are eliminated. This prevents a local nonentity from winning purely based on transferred votes. It also means that usually around 3–5 parties will win seats; neither the eternal two-way monopoly of FPTP, nor a fragmented world of a dozen parties as some fear from PR.
  • Then, candidates are eliminated and votes are transferred using an STV process, which ends up finding one winner per district and allocating the ballots so that each winner has the same number of ballots, fewer than that number of ballots are wasted, and each ballot goes to a candidate (or is split among candidates) that the voter liked as much as possible.

That all sounds a bit complex, so let’s look at an example. Imagine this had been used to elect congresspeople from Tennessee in 2016. Here are the totals for the two main parties:

╔══════════╦═════╦═════╗
║ District ║ R ║ D ║
╠══════════╬═════╬═════╣
║ 1 ║ 78% ║ 15% ║
║ 2 ║ 76% ║ 24% ║
║ 3 ║ 66% ║ 29% ║
║ 4 ║ 65% ║ 35% ║
║ 5 ║ 38% ║ 63%
║ 6 ║ 71% ║ 22% ║
║ 7 ║ 72% ║ 24% ║
║ 8 ║ 69% ║ 25% ║
║ 9 ║ 19% ║ 79%
║ ║ ║ ║
║ total ║ 62% ║ 35% ║
╚══════════╩═════╩═════╝

Under FPTP, that resulted in 7 Republicans and 2 Democrats winning:

But the proportional outcome when there are 9 seats is for each party to get 1 seat for every 10% of the statewide vote they control. Under GOLD, that would happen because the Democrats with the lowest numbers of votes would be eliminated and their votes would transfer, until one more Democrat — probably the candidate in the 4th district, who got 35%—won a seat. In the end, almost all the votes would end up with a winning Republican or Democrat, and fewer than 10% would be wasted.

(Whether the extra Democratic winner was in the 4th district or some other one might depend on how the 4th district candidates, both Democratic and Republican, got write-in and/or transferred votes. So if that Democrat was particularly unpopular, or that Republican particularly popular, on a statewide basis, then it would happen somewhere else. In fact, even the candidates in the most heavily partisan districts wouldn’t be guaranteed a seat, if their party’s voters preferred to write-in somebody else rather than voting for them.)

One way to think about this is to first find the winners using the rules above, but then, just to make drawing maps easier, re-allocate the transferred votes so that each district’s votes only transfer to the nearest same-party winner. Then, you’d see the following:

Each Republican winner needs votes from 1 or 2 districts
Each Democratic winner needs votes from 2–4 districts.

Of course, if you were really using GOLD voting in Tennessee, it would change how people vote. Since third party votes wouldn’t be wasted, probably more people would vote third-party. Perhaps a Libertarian or Green would make the top two in one of the districts, and, with the votes transferred from other districts, win a seat. Perhaps not; but even then, the Libertarian and Green candidates could declare their “candidate” transfer orders in order to benefit the Republican or Democrat they found most congenial; so they’d still have a fair, proportional voice in politics.

So GOLD voting has the advantages of PR: it makes gerrymandering obsolete, and takes wasted votes from being over half to being a small fraction. And it keeps the all the advantages anyone could ask for from FPTP: simple ballots (so simple, they can be tallied with the voting machines we already have); accountability, with no entrenched incumbents; incentives for broad parties, not ungovernable fragmentation; and geographic balance.

About those last two “advantages”… yes, they’re debatable. Some people might prefer a system without those aspects. But even for those people, GOLD is still a step up because of its other proportional aspects; while for the people who do like those advantages, it’s not a step down. Sometimes, it’s better to go for a moderate change that keeps everybody happy, than to hold out for the “perfect” system that upsets the people currently in power.

And GOLD has another reason to keep the one-winner-per-district rule. There’s actually a 1967 federal law that mandates that. Changing a federal law just takes a majority in Congress, but with GOLD you wouldn’t even need that; the whole reform is possible on a state-by-state level. The most important place to start is with “purple” swing states with badly-gerrymandered districts — places like Michigan, Pennsylvania, and North Carolina (as seen in this recent Brennan center report).

Let’s get to work!

PR kills the evil Gerrymander

In future installments of this series, I’ll show how to fix single-winner elections (with Approval voting or 3–2–1 voting) and nonpartisan city council elections (with a GOLD-like proportional system called 3RD).

¹ Well, GOLD has no downsides as compared to FPTP, at least. I’m not claiming it has no flaws at all; the ghost of Kenneth Arrow would haunt me for that. (He’s missed.)

² The letters GOLD stand for: Geographic (because it’s district-based); Open List (because if everyone chose the “voter” transfer method, it would work just like an open list system); Delegated (because the “candidate” transfer method is a form of delegation).

³ For the purposes of elimination, the top two candidates in a district are based on local votes (from that district) alone. A third candidate from a district may survive without elimination if they have more total votes than any candidate has local votes.

--

--

Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store
Jameson Quinn

Jameson Quinn

Opinion, info, and research on improved voting systems and democracy. Building website to use these voting systems securely for private elections.