Wednesday, October 10, 2007

New election or funding for the fire department?

The recent election for mayor of Fairbanks has resulted in neither of the top two contenders (both women, I'm happy to see) gaining the required 40% plus. This means Fairbanks will have to have a new election, an expensive business. Given that Fairbanks is having a few fiscal worries these days, it seems reasonable to me that the city should adopt Instant Runoff Voting. Saves a lot of money in a situation like this.

And that money could have gone to pay for raises to keep up with inflation, or firefighter health benefits, or other cuts they took back in 2003.



On the contrary, there is evidence that IRV can actually increase costs.

It is also conducive to the use of fraud-conducive electronic voting machines, says election integrity expert Rebecca Mercuri.

A hugely simpler and better voting method is Approval Voting. It works with the same ballots and voting machines we're all familiar with, except we change the "vote for one" rule to "vote for one or more". It's as simple as this: stop discarding "over-votes" - Count All the Votes!

Approval Voting is the simplest form of Range Voting (like Olympic scoring, but effectively a 0-1 "range" in the case of Approval). Here's an open letter to the IRV community from the Princeton math Ph.D. who co-founded the Center for Range Voting:

IRV is one of the worst voting methods out there. We can do better, and with less mess, by simply using Approval Voting.

Deirdre Helfferich said...

There are problems with range voting, as FairVote points out on its blog. Also, any voting method is vulnerable to fraud when insecure computer systems or software are used. Ranked choice voting (IRV) has been used with paper ballots, and used to be used in the US.


FairVote's Range Voting piece is deceptive and unscientific. It is rigorously refuted here by a Princeton math Ph.D. Note that the piece you link to is written by a political operative who has made numerous false and misleading claims in defense of IRV.

You raise an excellent point about the dangers of electronic voting machines. IRV has been shown to incentivize the adoption of fraud-prone electronic voting machines, as it cannot be performed on ordinary machines. Range Voting and especially its simplified form, Approval Voting, can be performed on all standard machines that we already have. And it's easier/cheaper to manually count.

Deirdre Helfferich said...

I'll look the piece over, but IRV has been used for a long time. It can indeed be used on voting machines, since it has been used in pre-computer days. Range voting is still not without flaws (like any system).

The basic issue to me is really that plurality voting is not that good a system. So far as I can tell, any of the voting methods involving ranking (which includes range voting) are better than plurality voting.



Do you think that plurality is better than IRV because it has been used longer? And what would you say about the fact that IRV is 137 years old, whereas Approval Voting was used for over 500 years in Renaissance Venice (until Napoleon came along and ended that)?

And no, IRV cannot be used on standard plurality machines, like Range and Approval can be. It was done on paper in pre-computer days. And no matter what technology you use, IRV still cannot be sub-totaled in precincts, which makes it more susceptible to centralized fraud.

You are correct that no voting method is perfect, but Range Voting is a huge improvement over plurality and IRV.

The basic issue to me is really that plurality voting is not that good a system.

And when you consider that IRV only produces a different outcome than plurality in about 5% of elections, and multiplies the number of spoiled ballots by a factor of 7, and also is substantially more's not even clear that it is an improvement.

So far as I can tell, any of the voting methods involving ranking (which includes range voting) are better than plurality voting.

Range Voting (Approval Voting being the simplest form) uses ratings, not rankings. IIt is a "cardinal" voting method, not an "ordinal" voting method.