Comment from Ratings:   more people started waking up the reality behind the IRV propaganda. Good job Steve.

by Clay Shentrup (0 articles, 0 quicklinks, 0 diaries, 39 comments) on Wednesday, Aug 20, 2008 at 1:30:14 AM

  (0+)

Unfortunately there is no such thing as a perfect voting method -- instant runoff voting (IRV) and approval voting included -- though both are certainly better than our current plurality voting method.

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

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out some more in non-governmental organizations first (some that adopted it previously were dissatisfied and have since repealed it).

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." 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. 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.

by Terry Bouricius (0 articles, 0 quicklinks, 0 diaries, 5 comments) on Wednesday, Aug 20, 2008 at 7:27:21 PM

  (0+)

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

It has been extensively studied via computer simulation, which places it far above IRV in its ability to pick satisfying candidates.
=> http://rangevoting.org/BayRegDum.html

Here's why those simulations are superior in many ways to anything that could be learned from "real-world, high-stakes elections".
=>

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.

Simply not true. Most mathematically credentialed voting methods experts (e.g. Warren Smith and Steve Brams) consider Approval Voting to be extremely resistant to strategic voting - in fact that is considered to be one of its strong points, along with simplicity. The idea that Approval Voting would degrade to plurality is refuted by the observation that around 90% of voters who supported Nader in 2000 actually voted for someone else.

To a rational observer, this reason behind this is obvious. Most of them were concerned that Bush might beat Gore. But if your argument were true, they would have been hesistant to vote for Gore out of concern that such a move might help Gore to defeat Nader.

Do I have to explain why this argument makes you sound like such a crack pot?

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

Tideman's "strategy resistance" measure is severely flawed, as explained here.
=> http://rangevoting.org/Tideman

Simply put, Tideman uses arbitrary measures of strategy resistance, rather than assessing the objective Bayesian regret of various voting systems as the "strategic-ness knob" is turned from 0% to 100%.

As for organizations which you say have gotten rid of Approval Voting, there is no evidence to suggest that this had anything to do with the plurality-style voting problem that you mention. In fact the two examples I'm aware of, the IEEE and Dartmouth University, both reverted back to plurality voting. Faced with similar charges by Terrill's partner in crime, Rob Richie of FairVote, Robert Norman (a Dartmouth emeritus professor of mathematics, and Approval Voting advocate) says:

Robert Z. Norman, a Dartmouth professor emeritus of mathematics and an Approval Voting advocate, does not believe this had anything to do with problems with Approval Voting.  He says:

"The claims about bullet voting in the Dartmouth Alumni election remind me that with a per voter average of  voting for 1.8 candidates, the proportion of bullet votes has to be fairly small. The alternative, as I mentioned in a communcation a day or two ago, is that nearly everyone voted for one or three candidates but not two. Unlikely as that might be, it would suggest that most of those who voted followed a strategy of either voting for the petition candidate or voting for all nominated candidates, in which case Richie's claim that the opposition was disorganized falls apart, as does the claim by some of the Alumni Council people that in a 1 on 1 situation the petition candidate would been defeated."

There's a moral here. Richie, Bouricius, and many other IRV supporters will say absolutely anything, without regard to truth.

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."

IRV reduces the probability that a third party will be a spoiler (though it does not completely prevent it, and does not stop two-party duopoly). If that is your standard of "success", then you have awfully low standards. Score Voting and Approval Voting are much better and simpler.

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.

False. There is absolutely no aspect of Approval Voting that makes it conducive to a two-party dominated government, because it never punishes a voter for supporting all candidates he likes more than his most strategic pick. It passes the "Favorite Betrayal Criterion", unlike IRV.

And might I add that you are demonstrably wrong considering that most of the 27 countries with a genuine (not "instant") runoff have escaped duopoly.
=> http://rangevoting.org/TTRvIRVstats.html

These are things I've pointed out to you many many times, so I see these arguments from you as very disingenuous.

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.

That's simply not true, for reasons I've outlined above.

Also, if you want to have proportional representation in America, you most likely will have to first get Score Voting or Approval Voting in order to break up the duopoly and get some candidates into office who will work to get rid of laws and other impediments to proportional representation. IRV can't do that.

Now let the world acknowledge the deception and outright lies of Terrill Bouricius.

Clay Shentrup
San Francisco, CA
clay@brokenladder.com
206.801.0484

by Clay Shentrup (0 articles, 0 quicklinks, 0 diaries, 39 comments) on Friday, Aug 22, 2008 at 12:19:14 AM

  (0+)

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.) 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 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.

by Stephen Unger (6 articles, 0 quicklinks, 0 diaries, 10 comments [1 recommended, 0 rejected]) on Thursday, Aug 21, 2008 at 5:22:36 PM

  (0+)

I would agree the central tabulation is a minor security concern, yet indeed a concern nonetheless. However, that is not that fault of the voting system, as IRV could be counted by hand, and also as voting machine vendors get on board with more updates to equipment, this will work itself out in the not too distant future. Are we really going to let private corporations dictate what election reforms we can have, or allow Stephen to paint the voting system as responsible for their not very adaptable products? That seems like a very unfair assessment.

by Anthony Lorenzo (0 articles, 0 quicklinks, 0 diaries, 3 comments) on Thursday, Aug 21, 2008 at 5:56:19 PM

  (0+)

Granted, Instant Runoff Voting (IRV) is my most-preferred SINGLE-MEMBER election system, the reality is that single-member election systems can't possibly produce fully representative (of the voters) outcomes. Not only is approval voting inferior, as it leads to the same plurality outcome of non-majority winners, it is also a single-member election system. Single-member systems lead to a two party system and serious lack of diversity for the voters.

I agree with Terrill that approval is not really taken seriously by anyone other than a few mathematicians. In fact, I find it much less appealing than Condorcet and Instant Runoff Voting, as it introduces some real problems to the system such as failing on the majority criterion in many situations with 3 or more candidates.

I would accept at least range voting is better than approval voting, not that relative to IRV I find range a better system, however IRV is much better in my opinion than approval. If approval voting were such a great reform, why doesn't any government use it? Also, who has endorsed it? What parties?

by Anthony Lorenzo (0 articles, 0 quicklinks, 0 diaries, 3 comments) on Thursday, Aug 21, 2008 at 5:52:18 PM

  (0+)

Granted, Instant Runoff Voting (IRV) is my most-preferred SINGLE-MEMBER election system, the reality is that single-member election systems can't possibly produce fully representative (of the voters) outcomes.  Not only is approval voting inferior, as it leads to the same plurality outcome of non-majority winners, it is also a single-member election system.  Single-member systems lead to a two party system and serious lack of diversity for the voters. 

 I agree with Terrill that approval is not really taken seriously by anyone other than a few mathematicians.  In fact, I find it much less appealing than Condorcet and Instant Runoff Voting, as it introduces some real problems to the system such as failing on the majority criterion in many situations with 3 or more candidates.  

 I would accept at least range voting is better than approval voting, not that relative to IRV I find range a better system, however IRV is much better in my opinion than approval.  If approval voting were such a great reform, why doesn't any government in the United States use it? Also, who has endorsed it? What parties?

by Anthony Lorenzo (0 articles, 0 quicklinks, 0 diaries, 3 comments) on Thursday, Aug 21, 2008 at 5:52:39 PM

  (0+)

Anthony,

Can you explain what makes you think that Approval Voting would maintain two-party domination? It passes the Favorite Betrayal Criterion, so I don't see why that would happen. Consider that about 24 of the 27 countries that use a genuine (not "instant") runoff have escaped two-party duopoly in their single-winner elections.
=> http://rangevoting.org/TTRvIRVstats.html

Also, can you explain why you believe Approval Voting would behave like plurality voting in practice? All the Approval Voting data I've seen shows that not to be true. For instance, say I was a voter in the Democratic primary, who preferred

Gravel > Kucinich > Dodd > Richardson > Biden > Obama > Edwards > Clinton

and I actually voted for Obama in the primary (because I was being strategic and didn't want to throw away my vote on a candidate I liked more than Obama, but who I knew couldn't win).

If I could go back in time with Approval Voting and continue to vote for more candidates, I'd then vote for Gravel, Kucinich, Dodd, Richardson, and Biden - not just Obama.

Clay Shentrup

by Clay Shentrup (0 articles, 0 quicklinks, 0 diaries, 39 comments) on Friday, Aug 22, 2008 at 12:27:56 AM

  (0+)

Mr. Lorenzo has apparently misunderstood my point. Central tabulation is a major problem, not because of difficulties with current e-voting systems, but rather because it does not allow us to partition the tabulation problem into a set of simple tasks that can be handled locally with the results easily merged in a public manner, as can be done with all competing voting systems. Hand-counting is a manageable problem for a few thousand votes (most US precincts don't exceed about a thousand votes.) There is no way to verify the veracity of machine tabulation of hundreds of thousands much less millions of IRV votes. This is in addition to opportunities for fraud in the transmission of results from precincts to the tabulation center.

Using IRV would, on the one hand make it almost mandatory to use machines for tabulation, and on the other hand make it virtually impossible to check on the machine outputs via hand-counting. So the e-machine vendors that Mr. Lorenzo is concerned  about (properly in  my view), would essentially control our elections. I should perhaps note that, regardless of the voting system used, I consider e-voting (both DRE and Optical Scan) as dangerous threats to our democracy, and present a case for hand-counted paper ballots in "Forward to the Past: Junk the Machines, Count Votes Manually".

by Stephen Unger (6 articles, 0 quicklinks, 0 diaries, 10 comments [1 recommended, 0 rejected]) on Thursday, Aug 21, 2008 at 7:36:37 PM

  (0+)

I fully agree with Mr. Lorenzo that a proportional representation system is what we really need. But until we get there, we are stuck with single-winner elections and should try to get the best possible voting system for this.

Mr. Lorenzo proclaims that this is IRV and that Approval Voting (AV) is inferior. He supports this position by citing an authority (who, he says, dismisses AV supporters as "mathematicians") and making various assertions with no foundation. He makes no effort to respond to the concrete arguments and examples that I presented demonstrating serious problems with IRV and benefits of AV.

Finally, it is disappointing that one who purports to be a reformer is reduced to attacking a serious proposal by arguing that it has not yet been implemented.

by Stephen Unger (6 articles, 0 quicklinks, 0 diaries, 10 comments [1 recommended, 0 rejected]) on Thursday, Aug 21, 2008 at 8:05:04 PM

  (0+)

I already agree with your conclusions due to having read about both methods already. However, I don't think your examples were obvious at first sight.

The normal way to express ranked choices is using the > sign, so perhaps a better way to express the first example would be:

35 R>D>G
5 R>G>D
25 D>R>G
10 D>G>R
20 G>D>R
5 G>R>D

Totals

R: 40%

D: 35%

G: 25%

Ofc, that might be just that I am used to seeing them expressed like that. The second count for IRV is also normally explained as a transfer rather than a recount after excluding the eliminated candidate.

The example could be simplified by only having 3 types of voter.

In the first election, the votes might be:

40: R>D>G

35: D>R>G

25: G>D>R

The Green candidates loses the first round and is eliminated. His 25 votes are then passed to the voters' second choice (D).

The Democrat candidates wins the 2nd round (60-40)

In the next election, the Green candidate has a really good campaign and manages to increase his support at the expense of the Democrat, and Republicans are unchanged.

40: R>D>G

29: D>R>G

31: G>D>R

Now the Democrats lose the first round (barely). Since the Republicans are the 2nd choice of those voters, these votes transfer to the Republicans. Thus, the Republican candidates win 59 - 41.

By *increasing* their support, the Green party have managed to not only lose, but cause a worse outcome (the GP supporters preferred the Democrat to the Republican).

If the GP supporters were smart, they would have voted D>G>R (i.e. pretended to be Democrat supporters). This would mean that the Democrat candidate would win 60% of the votes in the first round and thus win the seat.

However, that means they are being forced to be dishonest about what party they actually support. But, wasn't that the whole point of bring in IRV, to allow people to vote honestly.

IRV eliminates the spoiler problem in some instances, but once the 3rd party candidate gets popular (so he has a chance of making it to the top-2, suddenly the spoiler problem reappears).

by Raph Frank (0 articles, 0 quicklinks, 0 diaries, 1 comments) on Friday, Aug 22, 2008 at 5:22:48 PM

  (0+)