Subscribe
Subscribe
MY AMERICAN SCIENTIST
LOG IN! REGISTER!
SEARCH
 
Logo IMG
HOME > ON THE BOOKSHELF > COMMENTS

Electoral Games People Play


Comments


Steven Hill, a FairVote activist and proponent of Instant Runoff Voting, is mistaken in his criticism of score voting (aka range voting), and its simplified form, approval voting. He says:

"[Score voting] ..works fine when voters don't care greatly about the outcome. But if range voting is used for public elections, once again smart candidates will urge their supporters to vote strategically by not rating other candidates—that is, to bullet vote. So range voting also would tend to regress to plurality voting."

This claim is simply false, as demonstrated by the 2000 U.S. Presidential election, in which NES polling data shows that 90% of the voters who preferred Nader voted for someone else, even though he passionately urged them to vote for their sincere favorite. This is an observation that Hill himself has made in numerous pro-IRV articles, e.g.:

"[With IRV] voters are liberated to vote for the candidates they really like without worrying about "spoilers." You can rank your favorite candidate first, knowing if she or he can't win, you haven't wasted your vote because it will go to your second choice."
- http://www.truthout.org/article/steven-hill-instant-runoff-voting-is-catching-on

So Hill is making two contradictory claims, depending on what happens to support his argument in a given context. This behavior is typical of FairVote activists, including Rob Richie and Terrill Bouricius. (You can see a similar response to this one, but addressed to Terrill Bouricius, here: http://groups.google.com/group/scorevoting/web/degrade-plurality )

It might surprise readers to know that Poundstone covered all this extensively in his book, after talking with mathematics and politics experts, like Warren Smith, the Princeton math Ph.D. who started the score voting movement. And various score voting advocates, including myself, have corrected Hill on this issue in the past. But FairVote has an agenda to implement proportional representation, and sees IRV as a stepping stone to that goal. And so they have absolutely no concern for what the math/politics experts say about better alternatives to IRV. They will simply ignore those facts and keep repeating these talking points.

Hill continues: "In short, range and approval voting sound good in theory but have serious shortcomings that become apparent when one takes into account human psychology and the blood sport of politics, with their disincentives to honest voting."

This is more deceptive propaganda.

IRV is _also_ susceptible to strategic voting.
http://scorevoting.net/TarrIrv.html

And score voting yields more satisfying/representative election outcomes than IRV, even if we assume for the sake of argument that IRV results in less strategic voting.
http://scorevoting.net/StratHonMix.html

I am somewhat shocked that an anti-science partisan like Steven Hill was offered the chance to write this review, given his complete lack of objectiveness on this issue.

Clay Shentrup
San Francisco, CA
206.801.0484
posted by Clay Shentrup
October 12, 2008 @ 6:33 AM


Steven Hill, a FairVote activist and proponent of Instant Runoff Voting, is mistaken in his criticism of score voting (aka range voting), and its simplified form, approval voting. He says:

"[Score voting] ..works fine when voters don't care greatly about the outcome. But if range voting is used for public elections, once again smart candidates will urge their supporters to vote strategically by not rating other candidates—that is, to bullet vote. So range voting also would tend to regress to plurality voting."

This claim is simply false, as demonstrated by the 2000 U.S. Presidential election, in which NES polling data shows that 90% of the voters who preferred Nader voted for someone else, even though he passionately urged them to vote for their sincere favorite. This is an observation that Hill himself has made in numerous pro-IRV articles, e.g.:

"[With IRV] voters are liberated to vote for the candidates they really like without worrying about "spoilers." You can rank your favorite candidate first, knowing if she or he can't win, you haven't wasted your vote because it will go to your second choice."
- http://www.truthout.org/article/steven-hill-instant-runoff-voting-is-catching-on

So Hill is making two contradictory claims, depending on what happens to support his argument in a given context. This behavior is typical of FairVote activists, including Rob Richie and Terrill Bouricius. (You can see a similar response to this one, but addressed to Terrill Bouricius, here: http://groups.google.com/group/scorevoting/web/degrade-plurality )

It might surprise readers to know that Poundstone covered all this extensively in his book, after talking with mathematics and politics experts, like Warren Smith, the Princeton math Ph.D. who started the score voting movement. And various score voting advocates, including myself, have corrected Hill on this issue in the past. But FairVote has an agenda to implement proportional representation, and sees IRV as a stepping stone to that goal. And so they have absolutely no concern for what the math/politics experts say about better alternatives to IRV. They will simply ignore those facts and keep repeating these talking points.

Hill continues: "In short, range and approval voting sound good in theory but have serious shortcomings that become apparent when one takes into account human psychology and the blood sport of politics, with their disincentives to honest voting."

This is more deceptive propaganda.

IRV is _also_ susceptible to strategic voting.
http://scorevoting.net/TarrIrv.html

And score voting yields more satisfying/representative election outcomes than IRV, even if we assume for the sake of argument that IRV results in less strategic voting.
http://scorevoting.net/StratHonMix.html

I am somewhat shocked that an anti-science partisan like Steven Hill was offered the chance to write this review, given his complete lack of objectiveness on this issue.

Clay Shentrup
San Francisco, CA
206.801.0484
posted by Clay Shentrup
October 12, 2008 @ 6:34 AM


Great review by Steven Hill!

I completely agree that we should talk more about the underlying values that we want our democracy to support and foster, and should seek to enact electoral methods that would best live up to those values. Doug Amy's book is excellent, and is an ideal starting point for a discussion about electoral reform and/or a move to improved methods.

I also agree with the comments about approval and range voting systems. There is absolutely no question that candidates and campaigns would game those systems miserably to push their supporters towards bullet voting (or the range equivalent of that by suggesting supporters score opponents as low as possible). It's now wonder that those systems are not used in any real municipal elections anywhere.
posted by Rob Dickinson
October 15, 2008 @ 2:47 AM


Every voting method has failings. Indeed some fairness criteria used by election method experts are mutually exclusive, so it is impossible for any system to be "perfect." But of the single-winner alternative systems that are vastly superior to current U.S. plurality elections, IRV has by far the best prospects for widespread adoption (having already been adopted by many cities, and used by millions of voters around the world).
However, I want to stress a point that Hill makes late in his review...All of the voting methods discussed in Poundstone's book are irrelevant to the main problem with voting in the U.S. That is, they are all winner-take-all methods that, at their best, still fail to provide full representation and reflect the diversity of the electorate. Most modern democracies use some form of proportional representation to elect legislative bodies, with the U.S. and Candada being two of the few that remain mired in 18th century winner-take-all plurality elections. Unfortunately, by ignoring this fundamental problem, Poundstone has provided a faulty road map for would-be reformers.
posted by Terrill Bouricius
October 15, 2008 @ 10:17 AM


Poundstone is an entertaining science writer (previous books include a famous one on Carl Sagan), and this book is a good read. But he is a science writer, not a mathematician, economist or political scientist. He ended up accepting pretty much at face value the opinions of advocates for range voting (as it's called in the book) or score voting (as the advocates call it now).

In addition, as Hill and one of the previous commenters point out, he doesn't understand the fundamental difference between executive branch and legislative elections and the fundamental importance of the choice between majoritarian and proportional methods of electing legislatures.

In spite of it's biases, however, Poundstone's book is quite valuable simply because there isn't much else for the non-technical reader (Donald Saari's "Chaotic Elections", for example, is clearly written but quite a bit more technical), and because he makes the subject pretty entertaining.

I think Steven Hill strikes a good balance in this review, both in tone and substance. If anything, he is a little too fair to Poundstone. I would not say, as Hill does, that all of the alternatives to plurality voting are better than plurality. The Borda Count, for example, is completely unworkable for public elections in spite of the purely formal properties that mathematician Saari finds so impressive.

Poundstone enjoys the intellectual puzzles of mathematics and related disciplines. Well and good. I'm entertained by them too. But that's not all there is to choosing a voting method.

posted by Bob Richard
October 15, 2008 @ 12:18 PM


Steven Hill got it right in pointing out that range voting, approval voting and Borda count, though all better than plain old plurality voting, have a
tendency to degenerate into plurality voting in elections where voters care about the outcome.

Clay Shentrup's comment gets it wrong in suggesting that this is somehow contradicted by the fact that many voters in 2000 who preferred Ralph Nader voted for someone else despite his passionate pleas for there sincere votes.

While most voters can't articulate the differences among electoral systems, that doesn't mean that they aren't aware at some level of features of the electoral system used for their governments. The United States generally uses plurality winner elections, and consequently voters are generally
aware of the possibility of spoiler effects. Voters don't automatically vote for the candidate they most prefer, nor do they blindly follow his or
her recommendations. They balance their desire to make a statement of support for their most preferred candidate with their desire to affect the outcome of the election.

It's to tip that balance that "major candidates" exaggerate the flaws of their "major" opponents, to try to win over voters who might otherwise vote for a "minor candidate" or stay home, as this emphasizes the importance of the election's outcome (and the consequences of any "spoiler
effect"). Similarly, "minor candidates" downplay the differences among the "major candidates" in order to deemphasize the significance of the election outcome, so voters who have a preference among the "major candidates" will feel less pressure to vote for the "lesser evil".

In 2000, the Democratic Party campaign's appeal to the spoiler effect was successful enough that Gore won a plurality of the popular vote, with more than half a million votes more than Bush. However, it wasn't successful
enough to win Florida by more than the margin of error and fraud in the vote counting process, so Bush was named President by the Supreme Court.

Similarly, in range voting, approval voting and Borda count elections, the politics of particular contests would significantly influence the extent
to which they degenerate via bullet voting into plurality elections, but the tendency is there.

posted by Dave Kadlecek
October 15, 2008 @ 1:37 PM


I think Steven Hills review is more than fair to all voting systems.

He misses a few arguments against Range Voting that should be mentioned. First, Range Voting advocates always compare results against the standard of Condorcet results. Yet they never seem to justify Condorcet as the “best” standard to use to judge the various systems. I fail to see why Range Voting Advocates don't just advocate Condorcet as the best voting system. If that is the standard that is used to compare all other systems, why not just advocate it. No other system could be better than the standard.

As Steve Hill points out, there are many other objectives that Condorcet or Range Voting don't address. Douglass Amy reviews all of these and recommends that one first decide upon the objectives of the voting system before picking a method. That makes more sense to me than to assume that one system is the “best”. We all know that “best” is in the eye of the beholder and depends on the objectives trying to be achieved. As the objectives change, then “best” also changes.

Sometimes the logic of Range Voting advocates escapes me. Just because 90% of Nader supporters voted for Gore when the common believe was that, “a vote for Nader is a vote for Bush” doesn't prove Steve's statement wrong. People saw through Nader's pleas. However, with Range Voting, if a candidate says don't rank my opponents because that could cause me to loose the election, it seems reasonable that most voters would follow the advice unless it is a Nader situation or it is clearly wrong advice. Under a Range Voting system, if Nader said vote for me first or vote for me only because votes for others helps them, then people would again vote for Gore, not Nader because voting for Nader would still help Bush win.

Range Voter advocates are fond of claiming that with IRV insincere votes can help a voters candidate to win – well not actually win but will help another candidate win who may not be as egregious (from the voters point of view) as other choices. These are always very carefully constructed examples that are not too likely to happen. Change a few numbers just a little bit and the results change. Most examples require advance knowledge of the vote in order to vote strategically. In some examples, if too many or not enough voters change their votes, the results could be completely different. I have also seen examples of two Democrats running against one Republican. I have yet to see a real world election like this. Again the examples always compare the results against the Condorcet system without any justification for using this as the standard of comparison. Perhaps most voters prefer polarized candidates to the middle of the road. They may accept the middle of the road as a second choice but would rather take the risk that their first choice win than to vote for their second insincerely. If the middle of the road candidate is what the voters really want, they would give that candidate their first choice. In the Range Voting examples, the advocates seem to assume that voters are too unsophisticated to figure this out. But as the Nader example makes clear, the voters understand the effects of their vote very clearly.

I thank Steve Hill for this thoughtful and balanced review.

Chuck O'Neil

posted by Charles O'Neil
October 15, 2008 @ 6:28 PM


I'm still scratching my head over Mr Shentrup's comment. With a voting system that satisfies later-no-harm (as does IRV), a voter has no strategic motivation to bullet-vote. I fail to see the contradiction, and I'd be grateful to Mr Shentrup if he'd explain himself further.

Like any acceptable voting system, IRV is subject to strategic manipulation under some circumstances, but later-no-hard makes successful manipulation rather difficult, without close to perfect information about the intentions of the other voters.
posted by Jonathan Lundell
October 15, 2008 @ 7:34 PM


Dave Kadlecek,

No I did not "get it wrong" in citing the 2000 election election, in which 90% of voters who supported Nader betrayed him. This and numerous other examples throughout history amply demonstrate that the vast majority of voters care more about maximizing their happiness with election outcomes than about getting their favorite candidate elected at all costs. I even cite examples where Hill and other FairVote shills (e.g. Terrill Bouricius) make the very same argument IN FAVOR of IRV.

It is a simple fact that plurality-style voting is NOT a good strategy with score voting or approval voting. It is a simple fact that examples of these voting methods used in REAL elections do not exhibit this behavior. It is a simple fact that voter behavior with other voting methods demonstrates that the vast majority of voters do not employ the kind of strategy that underlies this "degrades to plurality" argument. This is just a simple deception on the part of Hill, you, and other IRV advocates. There is simply no evidence supporting these claims.

Moreover, Hill writes as if to imply that Poundstone and other score voting advocates had not even thought of this issue - when in fact we've been studying this extensively for years, and have actually seen evidence that plurality-style voting does not occur in cases where approval voting is used in consequential elections (e.g. in universities such as Dartmouth and San Francisco State University).

My experience is that IRV advocates like Steven Hill will say absolutely anything to support IRV, and they have stated outright falsehoods on numerous occasions, e.g. in Science magazine:
http://scorevoting.net/Irvtalk.html
posted by Clay Shentrup
October 16, 2008 @ 4:19 AM


Jonathon wrote:

"I'm still scratching my head over Mr Shentrup's comment. With a voting system that satisfies later-no-harm (as does IRV), a voter has no strategic motivation to bullet-vote."

1) ALL deterministic rank-order voting methods exhibit some susceptibility to strategic voting, as described in the Gibbard-Satterthwaite theorem:
http://scorevoting.net/GibbSat.html
2) Here is specifically why IRV is susceptible to strategic voting:
http://scorevoting.net/TarrIrv.html

3) Even if we assume for the sake of argument that IRV is free from ANY strategic voting, and that score voting has a large number of strategic voters, score voting still better represents the will of voters by a wide margin. http://scorevoting.net/StratHonMix.html

Moreover, satisfying later-no-harm is a BAD thing for a voting method to do, as it decreases the ability of that voting method to utilize revealed preference, and essentially requires the voting method to be ordinal rather than cardinal (and thus susceptible to a myriad of problems outlined by Arrow's theorem and the Gibbard-Satterthwaite theorem).
posted by Clay Shentrup
October 16, 2008 @ 4:24 AM


Total Records : 17


 

Connect With Us:

Facebook Icon Sm Twitter Icon Google+ Icon Pinterest Icon RSS Feed

Sigma Xi/Amazon Smile (SciNight)


Latest Multimedia

Bishop with beehives

The disappearance of honeybees continues to make headlines in the news and science journals, but are their numbers still dwindling, and if so, what are the causes?

Dr. Jack Bishop, a researcher at the National Institute of Environmental Health Sciences (NIEHS) and a hobby beekeeper, discusses the external influences that are linked to bee population decline, as well as ways to help honeybees thrive.

Click the Title to view all multimedia content!


Subscribe to Free eNewsletters!

  • Sigma Xi SmartBrief:

    A free daily summary of the latest news in scientific research. Each story is summarized concisely and linked directly to the original source for further reading.

  • American Scientist Update

  • An early peek at each new issue, with descriptions of feature articles, columns, Science Observers and more. Every other issue contains links to everything in the latest issue's table of contents.

  • Scientists' Nightstand

  • News of book reviews published in American Scientist and around the web, as well as other noteworthy happenings in the world of science books.

    To sign up for automatic emails of the American Scientist Update and Scientists' Nightstand issues, create an online profile, then sign up in the My AmSci area.


Subscribe to American Scientist