We Can Do Better than Instant Runoff Voting [Approval/Range Voting]

Mr. Bouricius, thanks for your comments and the link to fairvote.org; i will check it out.

I am just learning about the options of Approval/Range and Instant Runoff voting, and am not sure what to think yet. I have invited Mr. Unger and Mr. Shentrup to come over here to NowPublic and comment as well.

In the meantime, for those who are checking this comment thread, see the discussion in the thread at opednews.com

I agree with Mr. Bouricius that a proportional representation system would be better than our winner-take-all system. It would not only, as he says, be more conducive to the election of representatives of more parties, representing a broader range of views, but it would also make the pesky redistricting issue less critical. I also agree that no voting method is perfect.

But his statement, "because 'approving' a second choice may help defeat the voter's first choice, ... it is likely to devolve to typical vote-for-one plurality voting", misses a key point. This condition, which would occur only if the third party were a contender, causes a genuine dilemma as to whether to approve the SECOND choice. It does NOT, as in plurality voting incite voters to abandon their FIRST choices. There are NO situations under approval voting (AV) that do this. However, as in the example I gave, under IRV there ARE realistic situations in which voters are motivated to lower the rankings of their first choices. (This is just one example of bad behavior of IRV. For other examples and references to still more examples, see http://www1.cs.columbia.edu/~unger/articles/irv.html) Deciding whether reducing the risk that the worst candidate will win is worth increasing the risk that the best candidate will lose, is a basic dilemma that I believe is endemic to all systems for single-winner elections.

Mr. Bouricius does not respond at all about to my point that IRV requires central tabulation of votes, a very serious fault.

His arguments for IRV and against AV, are largely confined to citing the views of "experts", and to mentioning places where IRV is used. The latter is not very persuasive, since we agree that the most widely used method, plurality voting, is very unsatisfactory. Citing experts is a game anyone can play--I prefer to present my case directly, with arguments and examples that readers can evaluate for themselves. I would be interested in how Mr. Bouricius would respond specifically to the points made in my article.

Terrill Bouricius is a prominent IRV advocate and FairVote activist, who has frequently repeated misleading and even false pro-IRV talking points, like the ones discussed on this page. To address his points one-by-one (but out-of-order, to expose one of his more heinous lies first):

However, because "approving" a second choice may help defeat the voter's first choice, most experts agree that it is likely to devolve to typical vote-for-one pluarlity voting.

This is simply a lie, that no "experts" believe, because it's simply absurd. To demonstrate that, just consider the approximately 90% of Nader supporters who voted for Gore back in 2000. If they could have used Approval Voting, they obviously would have wanted to vote for Nader too.

Another example: say that a Democrat voting in the primary had these preferences, and voted for Obama:
Richardson > Kucinich > Dodd > Gravel > Obama > Biden > Edwards > Clinton

Had the primary used Approval Voting, that voter would have wanted to also vote for Gravel, Dodd, Kucinich, and Richardson - since that could help but not hurt.

These simple examples devastate Terrill's talking point; yet he still repeats it. He's an IRV shill, plain and simple.

Range voting is essentially the same method [as Approval Voting] but with a layer of complexity that doesn't actually accomplish anything

Your claim that it "doesn't actually accomplish anything" depends on an ignorant assumption that all voters using Score Voting will vote "approval-style", when in fact it's reasonable to conclude that 50% or more voters will use intermediate values. And what that accomplishes is a huge increase in average voter satisfaction.

(a smart voter who understood Range voting would simply give candidates a zero or top score to avoid weakening her voting power with middling scores.)

You're clumbsily contradicting yourself. When you criticize Approval Voting, you say it will degrade to plurality voting (which is not strategic). Yet when you criticize Score Voting, you say it will degrade into Approval Voting due to strategic voting. Be a little more careful with your lie.

Approval has its own problems

Approval Voting is robustly superior to IRV. It produces substantially lower Bayesian regret than IRV (i.e. produces election outcomes with greater average voter satisfaction). It is much simpler than IRV, and reduces spoiled ballots (wheras IRV increases them by a factor of 7), and can be centrally counted (unlike IRV), and can be counted on all standard voting machines with no upgrades (unlike IRV), and makes it always safe for a voter to support his favorite candidate. There is simply no comparison here. Approval is better, and Score Voting (aka Range Voting) is better still.

Unfortunately there is no such thing as a perfect voting method

That's a red herring. No one's saying Score Voting or Approval Voting is perfect - just that they are massively superior to IRV.

-- instant runoff voting (IRV) and approval voting.. are certainly better than our current plurality voting method

It's not clear that IRV is better than plurality, because it only reduces Bayesian regret moderately, but increases spoiled ballots and susceptibility to fraud.

Approval voting is certainly a simple reform. It might be worth giving it a try to see how it would work in real-world, high-stakes elections

You can't measure Bayesian regret with "real-world, high-stakes elections", since no technology exists to read voters' minds. Extensive computerized election simulations, which can do this, have been performed with all mixture of strategic-vs.-honest voters, from 100% honest to 100% strategic. The result was that Approval Voting demolished IRV among that whole range. We don't have to know which of those values most closely matches reality - we tested along the whole range. In fact, even when we use a model specifically intended to be favorable to IRV and unfavorable to Score or Approval Voting (giving you the benefit of the doubt, despite your contempt for science), IRV still lags far behind.

(it [Approval Voting] has a limited track record, with the former Soviet Union being the only country that has ever tried using it)

Lie.

In 1990, Oregon used AV in a statewide advisory referendum on school financing, which presented voters with five different options and allowed them to vote for as many as they wished. Prof. Jack H. Nagel pointed out that the first four US presidential elections (1788-1800) were done using a form of approval voting (albeit with the undesirable twist that the second-place finisher became the vice president). In the late 1980s, AV was used in some competitive elections in countries in Eastern Europe...
- http://scorevoting.net/Approval.html

Terrill continues...

Analyses by election methods experts (such as Nicolaus Tideman and James Green-Armytage) have found that Approval voting and Range voting are more vulnerable to manipulation through strategic voting than just about any other method, so it makes sense to test them out some in non-governmental organizations first

1) Tideman's analysis was deeply flawed, because it picked an arbitrary set of criteria, with arbitrary weighting - whereas Bayesian regret figures, which measure the effects of all criteria (even ones that haven't been invented), at their actual weight yield dramatically different results, which favor Score Voting and Approval Voting.

2) Again, real world testing can't employ mind-reading, and so cannot measure Bayesian regret, and so cannot measure the negative impact of strategic voting.

3) Actual Bayesian regret figures show that Approval Voting is about as good with 100% strategic voters as IRV is with 100% expressive ("sincere") voters. So even if we generously assume, for the sake of argument, that your testing could be done, and revealed that 90% of Approval Voting users were strategic, and only 10% of IRV users were strategic -- then still Score/Approval Voting outperform IRV.

(some [organizations] that adopted approval voting previously were dissatisfied and have since repealed it).

Well, your narrative says that the IEEE, and Dartmouth University, abandoned Approval Voting because it had problems. But none of their stated explanations show any evidence of these alleged problems. The appearance is that this happened, in much the same way as proportional representation was ushered out of cities like New York City (decades ago) - because it did work, and that was not good for the establishment. Regardless of which theory is correct, there is definitely no evidence that this reflects any strategic voting problems.

Instant runoff voting, on the other hand, has a successful track record in Australia, Ireland, and now in various U.S. cities, allowing third parties to run active campaigns without being labeled "spoilers."

Lie. IRV does not solve the spoiler problem. That link proves it.

You could just as well say that plurality voting has "a successful track record in the USA". Moreover, IRV's history in those countries substantiates our biggest criticisms of it. It doesn't escape two-party domination. It substantially increases the number of spoiled ballots. It offers very little benefit, and introduces some negatives.

However, it is certainly not a panacea for third parties, as any winner-take-all method (including approval voting) has a tendency to encourage two dominant parties.

Lie. There is no evidence that Approval Voting encourages two dominant parties. The biggest reason to doubt that is that it satisfies the "favorite betrayal criterion", which IRV does not.

Only proportional representation, such as with the multi-seat variant of IRV known as "choice voting" or "single transferable voting" can genuinely open the door for third parties.

Lie.

1) You have no evidence that Score Voting and/or Approval Voting won't lead to three or more viable parties.

2) There is evidence that in the USA will not be able to transition to proportional representation without getting Score Voting first.

3) Proportional (Reweighted) Score Voting and Asset Voting are superior to STV

4) There are numerous elected offices, like Senate seats, that are single-winner, and so proportional representation won't help them; they require a better single-winner voting method. IRV is not good enough. Score Voting and Approval Voting are excellent.

There is a mass of information on proportional representation, IRV and other voting methods at www.FairVote.org

That site has been a source for lies and misleading talking points like the ones you've spouted here.

I come from the school that is skeptical about the benefits of score voting, simply because it was developed from rat'l choice principles about voters and quite frankly we don't know if folks will strategically vote or not. 

As I just explained above in extreme detail, this is false.

The Bayesian regret figures I cited include the full range of strategy, from 0% strategic voters to 100% strategic voters - and levels of ignorance which simulate all types of irrational behavior. As we point out over and over and over again, the advantage of score voting over alternatives actually increases the more strategic voters you have.

So I focus on the incorp of PR into state legislative elections in what I call Project Democratic Renewal.

Anyone who talks of proportional representation should first read this link.