[EM] Why I think IRV isn't a serious alternative

[Abd ul-Rahman Lomax]

At 04:29 PM 11/25/2008, Jonathan Lundell wrote:
>On Nov 25, 2008, at 1:19 PM, Markus Schulze wrote:
>
>>Or are only IRV supporters allowed to use polling data
>>to show the greatness of IRV, while advocates of other
>>methods have to use complete ballot data?
>
>I think we must be careful about using polling data when we're
>comparing election methods in which voters have different strategic
>motivations, and that taking sufficient care may preclude drawing firm
>conclusions.

Absolutely; however, the difference in "strategic motivation" between 
three-rank IRV (as is used in the U.S.) and top-two runoff is 
practically nil. The vast majority of voters simply vote sincerely, 
I'm sure, in both methods. They will vote in the first round of 
top-two runoff for their favorite, almost certainly. Strategic voting 
(such as turkey-raising) can backfire, you know. Make a mistake, you 
might end up with a turkey. Top two runoff allows the voter to 
postpone the decision of who to rank "second."

I just saw the video from the San Francisco Department of Elections, 
2006, that said it elects winners with a "majority." No qualifier. 
Not majority after excluding exhausted ballots. A lot of people, 
including officials and others who should know better, *including 
opponents of IRV*, were hornswoggled. If IRV were to actually be a 
majority method, as claimed, i.e., if it continued to require a true 
majority, it would be a much better method. I.e., top-two runoff IRV. 
Instead, San Francisco gets Plurality results for a far higher cost 
than Plurality. Basically, SF could have gotten the same result by 
eliminating the majority requirement. There is not one election that 
that has turned out differenty.

Would Plurality voters have voted the same as in IRV. Most of them 
would have, I'm pretty sure. The difference would be fewer votes for 
minor candidates, and the rations between the frontrunners would 
generally have remained the same. Voting systems analysts have 
generally thought of factional elections, where supporters of a minor 
candidate will, almost entirely, vote for only one of the two 
frontrunners, where vote-splitting only affects one side. In 
elections with a lot of candidates, there are likely to be 
vote-splits that cuts both ways. The Nader effect in Florida, for 
example, would have been countered to some degree by the effect from 
Libertarian and other candidates. We really don't know what would 
have happened had Florida been IRV or another method allowing more 
than one vote.

>Personally, I don't think that any available single-winner method, IRV
>not excepted, is particularly "great", though I prefer ranked-ordinal
>methods to FPTP or TTR.

It's almost certainly true that TTR has generally better results than 
IRV. Essentially, when needed, two ballots are better than one. Three 
would be better than two! Democratic process skips all this crap and 
iterates binary decisions, with a majority requirement to make any 
decision. It continues to iterate until a majority is found, or a 
majority decides to adjourn....

None of the Above is always on the ballot with true democratic 
elections, and doesn't have to be a named candidate. With Approval, 
for example, just write it in! Lizard People would have been fine.

>  My mild preference for IRV over Condorcet
>methods (and stronger preference over approval and range) has to do
>with wanting to keep strategic voter considerations to a minimum. That
>ends up being a somewhat subjective and intuitive conclusion; at least
>that's how I see it.

Yeah. Unfortunately, "intuition" sucks when its been misled by 
centuries of diffuse propaganda and simple habitual assumptions, plus 
a boatload of very targeted spin-doctrine actively promoted recently. 
You want to minimize strategic voting, why not use methods designed 
to do exactly that, to the point where it is debatable whether what 
remains is "strategic voting" or "harmful" at all?

Why *prohibit* voters from equal ranking? Why do you imagine that you 
get better results by confining the voters without clear cause? IRV 
with equal ranking allowed: much better! It would allow voters to 
vote Approval style or Ranked style, whichever they prefer. Power to 
the Voters! Count All the Votes!