2009-11-03

Afghan election largely a matter of a bad voting system

2009 elections demonstrate flaws in voting methods

Press Release – the Center for Range Voting – 2 November 2009

Of the three most-examined elections in the USA plus the Afghan Presidential election – i.e. the four most-important-for-the-US elections in 2009 – three clearly exhibited major flaws. The problems in the two US elections are traceable to the flawed voting method, "plurality," employed. In contrast, score voting (also called "range voting") and approval voting both would have behaved reasonably. But Instant Runoff Voting (IRV; lately called by its US advocates "ranked choice voting") still would have yielded clearly-undemocratic behavior.

The Afghan election was a massive fraud. If the Afghans had foolishly employed IRV, they would have been even more vulnerable to that fraud with even less chance of correcting and detecting it. For Afghanistan we recommend approval and asset voting.
What went wrong:

* New Jersey Governor race: All three candidates had approximately equal quality in the collective opinion of the voters – with Daggett having the most approval and greatest average score in score-voting and approval-style polls. It also is probable Daggett would have defeated every rival in head-to-head races. This truth ws massively distorted by plurality voting ("must not 'waste' vote" strategic imperative amplified by media and cash) which caused Daggett to finish far behind in last place. Second, a Corzine victory (if occurs) quite likely will be due to Daggett's "spoiler" effect. IRV would have cured the second problem but not the first.
* New York district-23 congress special election: The pre-election polls before Scozzafava dropped out (after, they oscillated wildly) seemed fairly clearly to indicate that all three candidates again had approximately equal quality in the collective opinion of the voters and make it probable that Scozzafava would have defeated every rival in head-to-head races. But the massive built-in undemocratic distortion of plurality voting ("must not 'waste' vote" strategic imperative amplified by media and cash) caused the outfunded Scozzafava to fall far behind her two rivals in plurality-style polls, to the extent where she actually felt it best (as she was urged) to drop out of the race to avoid possible "spoiler" and "vote-splitting" problems. With approval and score voting, spoilers and vote-splits do not exist – and the nonsense that a perfectly good candidate dropping out can be regarded as an "improvement" of democracy, is abolished. Our data based on approval and score-style polling [as well as subsequent developments, such as Scozzafava endorsing Owens(D)] suggests the whole fear of a Republican-Conservative "vote split" actually may have been misguided. With IRV, this whole pathology still would have occurred.
* Re Afghanistan, it is worth noting that IRV has inherent properties making fraud harder to stop and detect, and also harming ballot privacy (e.g. any attempt to publish IRV-style ballots in Afghanistan's 41-candidate race, would have instantly identified voters enabling vote-buying and coercion).

Take-home lesson:

* Score voting and approval voting: good (simple, behave well).
* Plurality and instant runoff: bad (plurality behaves badly; IRV has problems less often, but as this makes clear, still too-often, and also IRV is more complicated and fraud-prone)
* "Asset voting," originally invented by famed children's writer Lewis Carroll, looks like an excellent choice for multiwinner elections in a place like Afghanistan where proportionality, simplicity, and fraud-proofness all are paramount.

More details:

* Detailed look at the USA 2009 elections.
* Afghanistan recommendations.
* Survey of IRV pathologies, the Burlington VT 2009 mayor election which demonstrated many of them, and graphical illustrations of how IRV favors "extremists" while preventing "centrists" from winning.
* About score voting's use in Ancient Sparta and Renaissance Venice.

Contacts:

Dr. Warren D. Smith warren.wds AT gmail.com (prefer email) Phone: 631-675-6128 co-founder, Center for Range Voting

Dr. Rob LeGrand rob AT approvalvoting.org Assistant Prof. Dept. of Mathematics and Computer Science, Bridgewater College. E-mail: rlegrand AT bridgewater.edu, office phone: 540-828-5754

Prof. Steven J. Brams NYU politics dept. steven.brams AT nyu.edu Phone: (212) 998-8510 FAX: (212) 995-4184

No comments: