Saturday, December 01, 2007

Is Range Right?

A few months ago, I commented that our voting system is flawed and that proportional representation or instant runoff voting may be superior. I believe now that I was incorrect, thanks to some help from Clay Shentrup and the information over at
My main concern with the winner-takes-all style of voting currently dominant in the US is its tendency to establish unbeatable incumbent politicians and parties. Government is only effective if it is legitimate, and its legitimacy is based on responsiveness to voters. The security blanket offered to incumbents by the winner-takes-all system eliminates that responsiveness in many cases and thus the legitimacy and effectiveness of government itself.
That seems like a bad idea.
I studied runoff voting as one potential alternative. This remedy was pushed hard by the Green Party a few years ago as a way to eliminate the "wasted vote" concept. The idea is that you prioritize each candidate running, giving each a number. If your number one choice gets the fewest votes in the first round, that candidate is eliminated and your vote turns to your number two.
As it turns out though, in Ireland and Australia, where it's used, Instant Runoff Voting produces the same two party dominance as winner-takes-all.
And fundamentally, the problem isn't how many important parties there are, it's how responsive government is to the needs of the governed, and IRV doesn't offer solutions here and can actually increase problems, such as situations where a candidate is selected where a majority of voters prefer another because of arbitrary cutoffs.
I thought perhaps Proportional Representation might solve this problem. In this system, voters vote for a party, and then however many votes that party gets is how many seats they get in a government body. The party selects who goes in those seats.
The evidence is unclear that this system is better, except that there is some evidence that countries with more PR like Ireland and Switzerland have a superior quality of life than per capita GNP would suggest. I don't like that this system prevents voting directly for candidates. I think that people are more important than parties.
This other option, range voting seems more promising. In range voting, voters rank each candidate from 0 to 100 based on their personal preference. Then the numbers are all added and whoever has the highers numbers wins.
Mathematically, this system comes closest to a state where God comes down and personally selects the candidate that best represents the people, which everyone immediately agrees with. In one analysis, the switch from a plurality winner-takes all system to a range voting system produced a superior improvement in democracy than the difference between the plurality system and picking candidate's names randomly out of a hat. This suggests that a global switch from plurality to range voting might usher in a Renaissance of good government and associated economic boom as better decisions lead to less waste and greater opportunity.

Can it happen in Charlottesville?


Anonymous said...

I appreciate your post, but a couple of corrections.

1.IRV and Range Voting are single-winner ("winner take all") election methods.

2. Proportional Representation does not require voting for a party. STV (used in the Australian Senate), as well as Reweighted Range Voting and Asset Voting, are proportional representation systems where you vote for candidates, not parties .

The evidence is unclear that this system [proportional representation] is better, except that there is some evidence that countries with more PR like Ireland and Switzerland have a superior quality of life than per capita GNP would suggest.

Well said. And even those who like proportional representation should be wary of trying to implement IRV for American elections, as an effort to transition to STV, because IRV also producesleads to two-party domination, and American impediments to such a transition are far greater than those faced in Australia when they made such a transition, decades ago.

I don't like that this system prevents voting directly for candidates.

Maybe you'll be happier with P.R. now that you know that's not actually the case.

In range voting, voters rank each candidate from 0 to 100 based on their personal preference.

Or we could use 0-10, or 0-1 (i.e. Approval Voting) for simplicity. Still almost as good, and arguably worth the added simplicity.

Mathematically, this system comes closest to a state where God comes down and personally selects the candidate that best represents the people, which everyone immediately agrees with.

That's quite an inventive way of conveying the concept of Bayesian Regret.

Can it happen in Charlottesville?

Approval Voting is probably your best bet. It's radically simpler than IRV, and more utilitarian, and resistant to the effects of strategic voting. We just change the current "vote for one" rule to "vote for one, or more" (which is identical to scoring the candidates on a 0-1 scale).

From there it might be possible to eventually upgrade to a larger range, like 0-10, or 0-99.

Anyway, thanks for your passion about this issue. I believe that upgrading our voting method is our number one priority.

tbouricius said...

As Clay S. (Broken ladder) said, Range Voting and IRV are both "winner-take-all" methods. Either of these would probably be more democratic than current plurality elections, in that they help avoid the spoiler problem. However, Proportional Representation, such as the "Choice Voting" or STV method used in Ireland and Australia (which is basically the multi-seat PR version of IRV) which allows ranking candidates in order of choice, is necessary to really have a multi-party democracy.

It is clear that IRV is the significant reform with the best prospects in the U.S. (Range Voting has never been used for any government elections anywhere in the world), and Australia's adoption of IRV led to the subsequent adoption of proportional representation (STV-style), for their Federal Senate.

People can learn more about both IRV and Proprotional Representation at

Terry Bouricius

Anonymous said...


IRV reduces the spoiler problem, whereas Range Voting completely eliminates it. It also eliminates a host of other problems that IRV doesn't, which is why it is significantly more democratic than IRV, per the world's most extensive Bayesian regret calculations, by Princeton math Ph.D. Warren D. Smith. When you imply that either of these methods would "probably" be more democratic, you give the false impression that they are comparable in quality, when in fact IRV is one of the worst voting methods, and Range Voting is, for practical purposes, the best.

You again make the false claim that Range Voting has seen no government use, despite the fact that Approval Voting, the simplest form of Range Voting, has seen use not only in several political processes, but in several professional organizations with tens of thousands of members. This has been explained to you before, but in typical FairVote fashion, you pretend you didn't know better. And either way, this is a hypocritical argument for you to make, since you could have said the same thing to the original proponents of IRV; and since the decades of IRV use in political elections has confirmed exactly the problems I cite.

It is also misleading to imply that we cannot judge a voting method aside from testing it in the real world. The untintuitive reality is that computer calculations are a superior way to guage the effectivess of a voting method. That's because we don't have a way to read voters' minds, and objectively measure their satisfaction with the election outcome. Computer simulations change that, and allow us to look at billions of elections, and model all sorts of different voter behaviors.

I don't know what you mean when you say IRV has the "best prospects". IRV typically increases the rate of spoiled ballots, and can markedly increase election costs, and has in some cases spurred the adoption of fraud-conducive electronic voting machines. It also cannot be counted in precincts, meaning it is more susceptible to a central fraud conspiracy. Just look at the message you get when you go to the San Francisco government web site, and try to check the results of the our Nov. 6th "ranked choice voting" elections:

Due to the requirement that all ballots must be centrally tallied in City Hall and not at the polling places, the Department of Elections has not set a date for releasing any preliminary results using the ranked-choice voting method.

So it's not even entirely clear that IRV would be a net gain or a net loss. And Approval Voting is cheaper and simpler to implement, and can be done on all standard plurality voting machines, and can be counted in precincts, and experimentally reduces the rate of spoiled ballots. So you're just wrong about IRV being somehow more adoptable. The only reason that IRV has seen some recent success, is that FairVote has promoted it with the same deceptive arguments that you're making here.

Finally, with regard to proportional representation, you ignore some crucial facts:

* Proportional representation is not required to have a multi-party system. For instance, most of the 27 countries that use a genuine (not "instant") runoff system, have escaped two-party duopoly.

* Range/Approval Voting would plausibly have an even stronger multi-party effect than runoffs, since they make it safe to always vote for your favorite candidates (unlike IRV or runoffs).

* The legal obstacles facing proportional representation in the U.S. are bigger than those in Australia when they transitioned from IRV to STV; P.R. is difficult to achieve at the federal level in the USA without a massive rewriting of the Constitution, and risks being stopped cold by the (2-party dominated) Supreme Court -- so it may be impossible to implement P.R. here without first breaking the duopoly, which Approval Voting plausibly would accomplish, but which IRV would not accomplish.

* Many important elections will always remain single-winner non-proportional, e.g. mayor, governor, senator, president - so why would you want to sacrifice all of them to a horrible voting method like IRV, just to get proportional representation? Especially when...

* It is not even clear that proportional representation is superior to good single-winner election methods, and...

* Range/Approval Voting are a natural intermediate/tranitionary step to Asset Voting and Reweighted Range Voting, two forms of proportional representation that are simpler and better than the ancient STV system you'd like to employ?

What Terry and his FairVote associates have got to realize is that the facts are just massively against them, and if they keep stubbornly ignoring those facts, their long-term accomplishments are going to be severely limited. Failing to deal with reality will stifle your efforts Mr. Bouricius.

Clay Shentrup
San Francisco, CA