The room where it happens

Bad voting methods put you outside. PLACE lets you in.

When laws are made, there are usually winners and losers. Inside the legislature, there’s a vote; one side wins, the other loses.

But there’s another set of losers: the people who aren’t even represented in the legislature. In most English-speaking countries, we’re used to that. Democracy means winners of losers, so a two-level, representative democracy must mean two levels of losers, right?

Wrong. That’s only because we use first-past-the-post voting—you know, where you can only vote for one candidate, and only in your local district. That means that a substantial number of voters in each district—sometimes more than half, if there were more than two candidates—are left out at the first step. But other voting methods exist which leave only a tiny fraction of unrepresented voters.

(Some political scientists call this problem “wasted votes”. But that suggests that any losing battle is wasted effort, so I’ll say “unrepresented voters”.)

As the US gears up for a new round of redistricting in 2020, while lawsuits from the last round in 2010 remain unresolved, this problem has been in the headlines. Consider this graphic, from the WUNC article “How Gerrymandered Districts Helped GOP Keep Veto-Proof Majority”:

If Democrats got 46% of the vote, but just 36% of the seats, then the missing 10% are locked out of the room where it happens. In this case, the difference favored Republican politicians, and didn’t switch the majority from one side to the other; but there are other cases where one or both of these isn’t true.

However, the 10% visible in the donut chart above is only the tip of the iceberg. US elections routinely don’t represent over 30% of voters, and sometimes over 50%. Gerrymandering turns some of this misrepresentation into a partisan weapon that shows up in graphs like the above, but FPTP creates the problem; one that would still exist even if the graph above were perfectly aligned.

This means that fixing misrepresentation wouldn’t just fix deliberate partisan gerrymandering. It would also fix:

• Geographic clustering, where the one party’s strongholds are naturally more saturated than the other party’s, leading to a distortion like the one above.
• Safe seats, where most elections are not really competitive, so politicians can grow complacent and even corrupt.
• Certain kinds of voters systematically don’t matter. If you’re a Democrat from Kansas or a Republican from Massachusetts, you’re not just in the minority locally, you’re ignored by your own party.
• A narrowing of debate, where voters and the media are incentivized to discuss the issues only insofar as the two parties differ. Raising an issue that might split your own party coalition can backfire and cause the other side to win.

Drawing better districts would be nice, but truly fixes none of these problems. What does fix them is proportional representation (#PropRep for short).

There are various different voting methods that give proportional representation. All of them are designed to make the colors line up in graphs like the one above—to make sure that 46% of the voters get about 46% of the seats. But at an even deeper level, perfect proportionality is just a consequence of a system which represents every voter. In the real world, you can never reach that perfection, but good voting methods can get more than 90% of the way there.

How? Well, it’s impossible for every voter to get their first choice of representative (unless you made legislatures hundreds of times bigger and allowed different representatives to have different voting power). So in order to make sure (almost) every voter is represented, you need some way to figure out voters’ other choices—second, third, etc.

There are three basic ways to do this, which can be used alone or in combination:

• “Ranking”: Ask each voter for their full preference order. This is the system used in Single Transferable Voting (also sometimes called “Ranked Choice Voting”). It seems reasonable, but unfortunately there are downsides. It leads to complex ballots and centralized, hard-to-audit vote-counting procedures. That usually makes it unworkable for more than a few seats at a time, so voters are usually drawn into districts with 3–7 seats each; that restricts voters’ freedom to vote for their true first choice.
• “Delegation”: If your favorite doesn’t win, that candidate’s choice decides where your vote goes next. To prevent corruption, the transfer orders can be required to be publicly predeclared, so you know when you vote where your vote might end up.
• “Pooling”: Infer one voters’ preference from those of other similar voters. This is basically the idea of “open party list” systems; if your choice doesn’t win, you are represented by somebody else from the same party, and who that is is decided by the other voters from your party.

For over a century, most voting reformers in the US have been focused on promoting STV, a ranking-based prop-rep method (as well as IRV or ranked Bucklin, related single-winner methods). And they’ve had some victories in that time—over two dozen cities have moved to using STV (and a similar number for IRV or Bucklin). But of those, only one city in the US (Cambridge, MA) is still governed using STV today. (Soon to be joined by Amherst, MA.)

I’d argue that the inherent flaws of ranking are part of why it hasn’t caught on. The very idea of representative democracy is that you can’t expect every voter to be an expert on the details of every law, so they should choose representatives to do that for them. This doesn’t just save time; it promotes leadership and increases bargaining power, so that, ideally, mutually-beneficial solutions can be reached.

Ranked ballots, on the other hand, expect each voter to have studied the full platform of all the candidates so as to decide whether they should be the voter’s 6th choice or 13th choice or whatever. But then, in the end, that ballot will be counted for only one of those choices. This is the kind of illusion of power that political junkies love, but less-engaged voters don’t care for.

And more seriously, the power to submit your individual preference rankings actually undermines two other kinds of power that are more important. The first is breadth of choice. Because STV ballots would be unworkably complex with too many candidates, voters are usually limited to voting in districts of 5-or-so members, even if their true first choice candidate is outside that district. The second is unified bargaining power. A community that’s unified around a single first-choice candidate will almost always lose some of that unity — and thus some of their power—when it comes to second- or third- choices.

That’s why I believe that the solution to gerrymandering, and all the other problems of misrepresentation listed above, lies in a new voting method: PLACE voting. This is primarily based on delegation, so unlike STV it gives maximum breadth of choice and bargaining power. And to ensure that delegated power is not used corruptly, general delegated ratings must be predeclared, and then combined with “pooling” to be turned into a full preference order. You can read the details here, and look at an example here.

While it fixes the huge problems of FPTP, PLACE also keeps all of the lesser benefits we have today:

• Simple ballots
• Geographic representativeness—one winner per district
• Single chain of accountability—you know who your representative is, and they’re actually from your favorite party (among those parties that won seats)
• Non-disruptive—doesn’t throw out legitimately-popular incumbents under the current system just because districts change.
• Favors big-tent parties with a full platform, not just single-issue splinter parties. We should have more than 2 real options, but we don’t want to go to something like Israel where parties are so splintered that the average voter’s party has under 1/6 support. PLACE voting strikes a good compromise in this regard.