While I appreciate the legacy of pro-democratic reform from the League of Women Voters, their analysis and advocacy of Instant Runoff Voting, including its 3-ranking-limit form called Ranked Choice Voting, has exhibited a concerning amount of mathematical inaccuracies and oversimplifications. 

1. IRV "ensures majority rule", and
2. IRV "eliminates problems of spoiler candidates"

As Smith demonstrates, both claims are simply false.

With IRV, candidate X can win even in cases where a majority of voters preferred Y to X, and Y got more first-place votes than X. In fact, in the last IRV mayoral race in Burlington, Vermont, a majority of voters preferred the Democrat over the Progressive, yet the Progressive still won.

Secondly, there was a bloc of voters who preferred Republican over Democrat over Progressive, who could have gotten the Democrat instead of the Progressive, if only they had insincerely ranked the Democrat ahead of those other two. For them, the Republican was a spoiler, and by ranking the Republican ahead of the Democrat, they "wasted their vote".

Moving on to the 2010 San Leandro mayoral race in particular, we see yet another way that IRV can fail to pick a true majority winner. It's called "ballot exhaustion". In this case, Cassidy won with 10277 votes from the 23494 people who voted in the mayoral race. That means he won with 43.7% of the vote, not a majority. This is especially common with RCV, since it forces voters to rank no more than three candidates.

This latest piece by the League gives the general impression that IRV is easy and straightforward, citing the 99.8% valid ballots, and assuring us that "the process is not complex and apparently requires a fourth grade arithmetic level to understand how it works". Both claims are highly misleading.

First, IRV typically causes ballot spoilage rates to increase by about seven times. To use an example from nearby San Francisco, their 2004 district RCV races featured a combined valid ballot ratio of about 99.4%. That seems similarly as impressive. Until you consider that the simultaneous non-RCV races had a valid ballot ratio of 99.918%. That's 7.3 times as many spoiled ballots with RCV. And presumably the figures would be similar if we were to pull up some historical non-RCV data from prior San Leandro elections.

And while it may only require a fourth grade arithmetic level to understand IRV, that does not mean that most voters actually do understand it. I am a professional software engineer in San Francisco's financial district, and through random polling of my peers, I have found that even these exceptionally analytical and mathematically gifted people, who all vote, do not understand how IRV works. To really make this tangible, I just opened up a chat window with one of my co-workers, and had the following discussion.

I challenge readers to do the same experiment. You'll find that all those voter information pamphlets and online videos don't actually get much notice and/or retention. Voters rely more on intuition. That common voter intuition might explain why, when I once called the Australian Green Party (Australia uses IRV in their House of Representatives), a representative explained to me that one of the most common questions he's asked by supporters calling on the phone is, "why should I waste my first place vote on the Green Party candidate?"

Next we're told about the virtues of turnout increases. But I do not think this can be attributed to the use of RCV. Consider that after adopting IRV, Minneapolis had the lowest turnout since 1902. And let's look at another example from nearby San Francisco. In the 2003 mayoral runoff election between Gonzalez and Newsom, there were 253,872 ballots cast (54.46% of registered voters), compared to 208,028 in the general election. By contrast, the 2007 RCV election featured significantly fewer registered voters, and yet it still had a lower turnout of only 149,424 (35.61%).

Notice also that turnout in the Gonzalez-Newsom runoff election went up. The common pro-RCV talking point is that it avoids the alleged lower turnout in delayed runoff elections. But what that race shows, along with several other similar examples, is that turnout in delayed runoffs tends to go up in elections of greater consequence. True, it tends to go down in less consequential races, like "comptroller".

The notion that IRV/RCV gives voters "more choices" or helps to "break the lock of the two-party system" is similarly flawed. IRV has maintained two-party domination in every country where it has seen long-term widespread use. For instance, Australia has used IRV in its House of Representatives since 1918, and is two-party dominated. There was actually a single exception in the most recent 2010 elections, in which a Green Party candidate in liberal Melbourne won a seat, resulting in 1 out of 564 seats being occupied by a third party.

The mathematical reality is that with IRV, supporting a minor party candidate is statistically more likely to harm a voter than to help a voter, so voters are incentivized to insincerely/tactically cast their favorite major party candidate in first place, regardless of who their sincere overall favorite is. Most IRV proponents are utterly unaware of that.

But it actually turns out that this tactical incentive is of little practical importance (aside from demonstrating that pro-RCV groups like the League disseminate a lot of misinformation about RCV), since most of the ranked voting users who cast tactical votes apparently do so purely based on intuition, as evinced by the phone calls received by my Australian Green Party friend above, as well as by election analysis.

The ultimate irony here is that most of the 27 or so countries which use a delayed (not "instant") Top-Two Runoff (aka TTR) actually do have three or more successful parties, a fact pointed out in Duverger's Law, one of the most famous observations in the history of voting theory. And TTR is precisely the system that IRV has replaced in every case I know of in the USA, including San Leandro.

Next, Syeda repeats the already debunked claim that RCV "provides a method to arrive at a majority candidate", and does so "without having the costs associated with runoff elections". This "IRV saves money" meme is specious. IRV has not been correlated with a significant cost savings over TTR. This makes sense considering that IRV is more expensive than a Plurality election, but less expensive than Plurality followed by a runoff, which is often not necessary. Here are some official budget numbers from San Francisco. Can you tell by looking at them which year RCV was adopted?

Lastly, there are at least two voting methods which are far superior to IRV/RCV and Plurality. They are called Score Voting (aka Range Voting) and Approval Voting. You can read about their benefits here and here.

Clay Shentrup

San Francisco, CA

The Center for Election Science - Secretary, Director