electopia | Indy Week

Member since Oct 15, 2008

San Francisco, CA



  • No friends yet.
Become My Friend Find friends »

Recent Comments

Re: “Q&A with Donald Saari about election math

The posts by Tbouricius and progressnerd demonstrate the typical misleading spin that many IRV advocates put on this issue. Tbouricius asserts "both the Borda method and current plurality voting are generally viewed by experts as among the absolute worst."

This claim is false by any reasonable standard. The vast majority of credentialed math experts who study voting methods consider Instant Runoff Voting to be the worst of the five commonly discussed alternatives to plurality (e.g. Steven Brams, Warren Smith). The objective measure of a voting method's quality is Bayesian regret - an "economic" indicator of representativeness; and Borda surpasses IRV by a significant margin. Here's a page describing this: http://scorevoting.net/BayRegDum.html (a lower Bayesian regret is better, like golf score)

Bouricius commonly references the "strategy resistance" assessment of Nicolaus Tideman. But Tideman's figures are severely flawed, because they use an arbitrary set of arbitrarily significant criteria for measuring strategy resistance. Warren D. Smith, the Princeton math Ph.D. whose Bayesian regret figures I cited above, has an thorough review of Tideman's book here: http://scorevoting.net/TidemanRev.html

We have pointed out these flaws to Bouricius, yet he repeats this talking point with no concern for that. I specifically authored this short piece to address another of his misleading talking points, so you can see for yourself how blatant this is: http://groups.google.com/group/scorevoting/web/degrade-plurality

Bouricius continues: "Instant runoff voting (IRV) as recently used in Cary, NC, is far superior to all of these options". This is simply ludicrous. IRV has a higher Bayesian regret than Borda, approval voting, score voting, or Condorcet. It is effectively the worst alternative voting method that has been proposed. It also cannot be sub-totaled in precincts, so it must be centrally tabulated. It increases the rate of ballot spoilage by a factor of seven. It may incentivize the adoption of fraud-prone (electronic) voting machines. It does not satisfy the favorite betrayal criterion. It historically has not ended two-party domination. Here is a point-by-point description of the massive superiority of score voting over IRV: http://scorevoting.net/CFERlet.html

Bouricius also misleads about IRV. IT does not accomplish what two-round runoffs do. It is mathematically different. It does not overcome spoilers. It does not guarantee a "majority winner" (nor should it): http://scorevoting.net/CoreSupp.html

And there is no guarantee that IRV saves money: http://scorevoting.net/VermontIrvCost.html

progressnerd says: The Borda count is highly susceptible to strategic voting.

But progressnerd supports Instant Runoff Voting, which is even more susceptible to strategic voting: http://scorevoting.net/TarrIrv.html

he continues: "The best single-winner system for elections in Instant Runoff Voting.

Well, that's just simply not true. It is the worst of the 5 commonly proposed alternative voting methods, according to Bayesian regret, ballot spoilage rates, summability, auditability, you name it.

"Saari concocts a theoretical example involving milk, wine, and beer to demonstrate a flaw with IRV, but in elections involving real political candidates, not beverages, these supposedly flaws don't occur with any discernible frequency."

You don't know that! Out of all the IRV elections that have been held, there is relatively little published, complete ballot data to analyze. And more importantly, you cannot read human minds to know what the real preferences of the voters were. Anonymous polling data can be a respectable substitute, but you don't even provide that. You simply repeat that "IRV succeeds in practice", without showing substantive evidence of this. Rigorous election simulations, involving real world problems like strategic voters and voter ignorance, have shown IRV to be a very poor voting method. If you contend that it somehow behaves well in practice, you have a huge burden of proof to reach. You have to show actual data, not just assertion. You cannot do that. If you could, I think you would have done it by now.

Posted by electopia on 10/18/2008 at 4:56 PM

Re: “Q&A with Donald Saari about election math

Donald Saari's criticism of Instant Runoff Voting is basically accurate, but his advocacy for the Borda voting method puts him in fringe territory, at odds with virtually every reputable election methods expert on the planet. Most notable is Warren D. Smith, the Princeton math Ph.D. whose computerized election simulations and mathematical analysis has turned much of election theory on its head. He provides a rigorous refutation of essentially everything Saari believes, here: http://scorevoting.net/DonSaari.html Smith is a proponent of score voting (aka range voting), in which we score the candidates on a scale like 0-10 or 1-5, in much the same way as some Olympic events are judged, or how we rate products on web sites like Amazon.com. To counter Saari's assertion that "the system that most accurately reflects the views of the voters is the Borda count", we simply cite Smith's Bayesian regret calculations, which show score voting to handily outperform Borda. http://scorevoting.net/UniqBest.html To break it down qualitatively, Smith explains: "Borda is worse than score voting in every way I can think of: less expressive, worse for honest voters, worse for strategic voters, extremely sensitive to voter-strategy, vulnerable to candidate cloning, more complicated both for the voter and for voting machines, severely fails irrelevant alternatives, and incentivizes 'favorite-betrayal.'" Saari's assertions about the niceness of Borda voting cite qualities like "symmetry" and "geometry", failing to observe that score voting has the same properties. More notably, Saari's proof that Borda is the "only" procedures which satisfies reversal symmetry, starts by defining score voting not to be a voting method. And Borda's susceptibility to strategic voting prompted Jean-Charles de Borda himself to cry in frustration that his voting system was "designed for honest men." Saari is utterly unqualified to speak as an "expert" about voting methods. Warren Smith says it best: "Saari is essentially the only voting expert in the world who doesn't think Borda is a very bad voting method. But Saari is a rather prolific and publicity-hungry one, and thus, unfortunately, has an outsized impact on the popular mind. "

Posted by electopia on 10/15/2008 at 11:46 PM

Extra Extra!

Make sure you're signed up so we can inbox you the latest.

  • Weekly Newsletter (Wednesday) - The stories in this week's issue
  • Weekly Events Newsletter (Wednesday) - Our picks for your weekend and beyond

Login to choose
your subscriptions!

Favorite Places

  • None.
Find places »

Saved Events

  • Nada.
Find events »

Saved Stories

  • Nope.
Find stories »

Custom Lists

  • Zip.

© 2017 Indy Week • 320 E. Chapel Hill St., Suite 200, Durham, NC 27701 • phone 919-286-1972 • fax 919-286-4274
RSS Feeds | Powered by Foundation