Minneapolis RCV Report: IRV Is No Magic Bullet

Dr. David Schultz (not be confused with Mark Schulze, of Schulze method fame) has posted his report for the Minneapolis Elections Department on the use of ranked choice voting [PDF] (more commonly known as instant runoff voting or IRV), as it was used in that city's recent elections. Schultz was (and apparently still is) a strong supporter of the use of IRV, having previously served as a board member for FairVote Minnesota, a fact he plainly states in his paper's Conflict of Interest heading. And this refreshing honesty continues throughout the document... which probably isn't good news for FairVote's or Dr. Schultz's objectives.

First on the chopping block is the claim that IRV increases turnout. Turnout for the 2009 IRV election was down 21% form the 2005 election. Now, there are many reasons for turnout to vary between elections--for instance, the report mentions that there weren't any particularly heated contests on the ballot--so this alone isn't certainly damning for IRV; but it surely doesn't help.

We also find that the well-advertised claim of decreased election costs turned out to not be true. We already knew this, based on a previous report, but it's worth reiterating: even when one-time costs, including voter-education costs, are ignored, the 2009 IRV election cost 20% more than the 2005 election. This is primarily a labor-cost issue, since Minneapolis was not able to find any voting machines capable of adjudicating an IRV election which meet the necessary security and accountability requirements set by law; so the vote had to be counted by hand. Schultz hopes that, perhaps, other towns in Minnesota will express interest in IRV, which will encourage the approval of the cost-saving IRV-capable voting machines. Considering the continuing problems of Diebold Premier Dominion Voting, I wouldn't hold my breath.

I do have one serious complaint about this report, and that's Schultz's claim that first past the post (FPTP, AKA plurality voting) is, like IRV, non-monotonic. This statement is unequivocally false, and the blatant attempt to whitewash this shortcoming of IRV stands out, painfully, against the otherwise honest assessments given throughout.

But my "favorite" part of the report is the section on Spoiled Ballots and Voter Error. Schultz begins by assuring us that "[T]he worst fears were not realized." By which he means IRV only quadrupled the ballot-spoilage rate, from 1.0% to 4.1%. It's true, previous IRV elections would suggest a six- or seven-fold increase in spoilage rates would be expected. Although it should be pointed out that, in addition to the 4.1% of ballots that had to be thrown out, an additional 6.4% had "errors" which the hand-counting procedure was able to "ascertain the intent of," for a total error rate of 10.5%. It's not clear how these ballots would have been handled by an automated system, had one been available.

Finally: how much of an impact did IRV have on the election? In all 20 single-winner contests, the candidate with the most initial 1st-place votes won the election. In 17 of these 20, the winner surpassed the 50% threshold immediately, and in the other 3 cases they did so after the first round of eliminations. But I don't want to hold this too strongly against IRV: as mentioned earlier, there weren't any hotly-contested races in this election. That usually means low-turnout and a lot of blow-out elections. Realistically, any electoral system would probably have come to the same conclusions in these races as IRV did. The true measure of an electoral system's quality is how well it can handle highly-contested and close-to-call elections. The predominant examples suggests that IRV would handle such elections poorly; but the Minneapolis data provides no real information for or against that proposition.

I hope that Dr. Schultz will carefully consider the empirical data that his report has highlighted, and that it will help to temper the misleading rhetoric that FairVote Minnesota has used, and which the national FairVote organization continues to use, to push for IRV's adoption around the country.

Instant Runoff (AKA Ranked Choice) Voting Has Spoilers

Another new YouTube video from SJVoter, showing how, despite claims to the contrary, instant runoff voting has spoilers.

Hmm... that example looks really familiar...

Linky Linky

Ran across a great summary of voting theory issues and why score voting is worth taking a look at. It covers a lot of the same points I often make, so if you've been following along here, you won't find anything new (well, I don't think I've talked about the bees thing yet...), but if you're just starting out, you'll find it to be an enlightening read.

Three Is More Than Two (But Less Than Infinity)

Why are approval voting and score voting such good voting methods? This certainly wasn't the expected result when Dr. Smith's voting simulation was first run, and a lot of work has gone into trying to make sense of the result.

I've Got a Theorem...

When anyone first starts to dig into the vagaries of electoral systems, they will quickly hit upon the Nobel-winning work of Kenneth Arrow and his impossibility theorem. But there was another very important--perhaps more important--theorem that came along just a few years later. It goes by the mouthful-of-a-name of The Gibbard-Satterwaite theorem. You do have to admit "Arrow" is a much catchier, and certainly shorter, name; so we'll call this "G-S" for short.

What G-S proves is that, under any electoral system using rank-order ballots, if there are at least three candidates, there will always be situations where a voter who knew how every other voter was voting, would be better off by voting strategically. Always. Even if you allow for equal-rank preferences (which some Condorcet methods can handle) you still run into this problem. So the conjecture is that no single-winner, rank-order-based method could ever possibly support three strong parties, since any candidate would eventually run headfirst into this problem and the perfectly-informed voter will have to choose between honesty and strategy.

But approval voting and score voting don't have this problem with a third candidate. It is still always in your best interest to rate your true-favorite highest, and your true-hated lowest, and doing so will never cause the election outcome to be worse than any other outcome you could achieve. In short, there's no incentive to vote strategically.

Preemptive Counter-Counter-Arguments

Now, there are a couple counter-arguments that people will bring up at this point. One of them is that, since G-S is about a "perfectly-informed" voter, they claim that this means approval (and score) only work this well if there is perfect polling for an election. Which is clearly impossible, so we should obviously use INSERT_FAVORITE_METHOD instead. The logic here is entirely unsound. First, even if G-S doesn't guarantee the effectiveness of ratings-based methods, it certainly does guarantee the ineffectiveness of all ranking-based methods; to stump for a known-bad over an unknown-but-potentially-good, seems blindingly counter-productive. Secondly, the fact that even the best-informed voters may have to strategize to avoid a bad outcome will tend to cause less-than-perfectly-omniscient voters to hedge their bets, and strategically go with the lesser of two evils out of fear of the greater evil.

But a more bizarre (or perhaps just more brazen) argument is that, since approval and score can't pass G-S with four or more candidates, then they are clearly insufficient. Which is an absolutely mindboggling argument, since even school children know that three is still more than two, regardless of the fact that three is less than infinity. Are these methods perfect? No, they aren't. Are they able to deliver an outcome that it is impossible for any ranking-based method to deliver? Yes, they are.

Too Infinity!

The fact that these methods have no perverse incentives in three-candidate elections is probably a large contributor to their improved performance in three-candidate elections, even if they still aren't perfect. And we know that all methods do better when everyone is honest, so having nothing to gain from being dishonest probably accounts for something. And perhaps this improved performance with three candidates somehow carries over and provide better results with four-or-more candidate races too, since performance drops at a noticeably slower rate than the rank-order methods do.

Perhaps knowing why something works isn't as important as knowing that it works; but being able to explain why may help convince some people who refuse to believe that.

I Am Full of Links Today

Another one for you, this time from The New Yorker book review; but it does a mighty job of summarizing 200 years of the intersection of math and politics in that rambling, New Yorker style, with a shout-out to my favorite book on the subject, "Gaming the Vote" (Amazon link on your right). An enlightening read, even if it does unfairly complicated approval voting in the opener (selecting the voters, no matter how complicated, isn't a part of the voting system!)

Must-Read Pro-Approval/Pro-Range Piece

Frequent LoAE commenter Broken Ladder post this excellent piece at As It Ought To Be, about the superiority of score and approval voting. The piece is by Andrew Jennings, Clay Shentrup (who I believe has also commented here), and Dr. Warren Smith (whose work I regularly quote).

Note to self: write up a post on the "complexity" argument.

Promising Trends

I want you to take a look at this Google Trend, particularly the breakdown by cities (bottom, center).

Minneapolis and San Francisco, which have both passed instant runoff laws in the last few years, have practically zero people who have looked into approval voting. But on the plus side, in Los Angeles, which is slowly examining IRV, even though IRV is more-often searched for, approval is still getting a noticeable number of hits. And in New York City, which has also begun considering IRV, approval voting has actually caught more people's interest than IRV has.

It's hard work, but it seems that the word is getting out. Please, do your part, especially if you're in LA or NYC. Unfortunately, score voting/range voting have no noticeable activity in any city; but keep talking about approval voting, because people are starting to listen!

The United Kingdom and the Alternative Vote

If you follow international news, you may be aware that, as part of the agreement for a new government, the Liberal Democrats have argued for a national referendum on the alternative vote, or what we would call in the United States (and what I will call in this post) instant runoff voting. Now, I'm no fan of IRV. But I think that, if the UK went this way, they might actually come out a bit better for it.

At first, I will be ignoring any non-outcome-based effects (i.e., costs) and holding everything else equal. And under these conditions, IRV would provide slightly better outcomes than plurality, given the assumption that a significant number of voters vote honestly rather than tactically. And, examining the election data for the UK, it seems that more voters are honest. There are two reasons that this may be the case.

One, the UK is a parliamentary system rather than a presidential one, so there is an extra layer of dis-connectedness between voters and the executive branch. This means there's less incentive to rally around or against the prime minister, and less direct control over them, meaning there is (or at least there is perceived to be) less to gain by voting tactically (this is, I think, a large reason for the LibDems success in the last several years).

Two, compared to the US, there is practically no polling done in the UK, except on the national level. And knowing that, say, 23% of people plan to vote for the LibDems, tells you practically nothing about what percentage of people will be voting for them in your local election, fancy new attempts to model such things not withstanding. Without that information, it's difficult to know how to vote tactically. Election results bear this out: there are a great many constituencies where tactical voters could have easily swayed the outcome in their benefit. So, given that there is and will continue to be significant honesty, the IRV results could be slightly more satisfying, based on Smith's simulation data.

But only slightly. And the elections will be more costly to count (perhaps only a little, perhaps 20% more). And this election marked a strong shift towards rallying around the party's would-be prime ministers (featuring the first-ever debate among them). And there's also a rising interest in local polling, partially caused by interest in IRV's possible benefits. Taking everything into account, there will likely be less election-satisfaction in the future no matter what, and if IRV is passed, it will probably be blamed on it, as a great deal of money will have been spent with little to show for it; which surely won't endear voters to the politicians who fought for— and over-sold the benefits of—it.

I wish someone with Nick Clegg's ear would tell him about approval and score voting. It would not be appreciably more costly, and would lead to markedly better outcomes (and probably more LibDems) regardless of voter honesty.

Always Learning

I picked the wrong day to be out sick, so since I missed out on the latter half of the conversation, I wanted to give a big thank you to DLW and Broken Ladder for an engaging debate in the comments. In the near future, expect to see some post from me based on those discussions. When intelligent people disagree with you, you should take the opportunity to come up with better ways to make your arguments, and that's what I'll be trying to do.