[EM] More non-altruistic attacks on IRV usage.

[Juho Laatu]

On 27.11.2011, at 8.05, matt welland wrote:

> On Sat, 2011-11-26 at 22:31 -0500, robert bristow-johnson wrote:
>> On 11/26/11 6:58 PM, matt welland wrote:
> 
>>> Also, do folks generally see approval as better than or worse than IRV?
>> they don't know anything about Approval (or Score or Borda or Bucklin or 
>> Condorcet) despite some effort by me to illustrate it regarding the 
>> state senate race in our county.
> 
> I wasn't clear. I want to hear opinions from the list: Is approval
> better or worse than IRV and why?

Unlike others, I think Approval might be worse.

Lets assume that there are two wings, left and right. Left has slight majority this time. Left consists of multiple candidates or multiple parties. Right has one candidate.

One basic problem of Approval is that all left supporters have to approve all plausible winners of the left wing in order to guarantee that left will win. That makes Approval quite numb to the opinions of the voters. If (almost) all approve all, the choice among left wing candidates will be random. Some voters might be tempted to approve only their favourites, and make them win this way. They may well succeed. But if number of strategic voters grows, then right wins. This kind of close competitions are not rare in politics. And in such situations one can not tell which candidate is the strongest among the left wing candidates (and a natural choice that all left supporters should approve). All candidates present themselves as likely winners, and their supporters tend to think that their favourite candidate is the strongest one.

Approval is nice because the ballots are simple. It works fine with two major parties and some new third parties. But when the third parties grow, the problems arise. There are no good solutions and no good guidance to the left voters in the situation where left wing has two or more plausible winners.

If one of the left wing candidates is a Condorcet winner (closer to the centre than the competing candidate), then that candidate may propose that all supporters of the other candidate should approve him although his supporters need not approve that other candidate. Maybe there are some voters that would even rank the right candidate second. But often there is no such clear order. And the other left candidate might be slightly ahead in first preferences.

IRV has its problems too. The reason why it might be better than Approval is that voters still have some sensible strategies, like ability to compromise. In the environment above left wing IRV voters will anyway rank all left wing candidates first. One of them will win, although the best of them might be eliminated too early. If there are two equally strong left candidates, the number of first preferences will decide which one of the left candidates will win. That is not as bad as the problems of Approval in this situation.

In IRV minor parties are a bigger problem than in Approval. In this example they may steal first preference votes from the second favourite of their supporters, and thereby make some worse left wing candidate win. In this situation the voters may compromise. If their own candidate has no chances to win, they might be ok with ranking the stronger second favourite above him. Not good, but at least the voters can do something. And even if they will do nothing, they would still get a left winner.

If there is a clear Condorcet winner (like in Burlington), IRV will have problems. So will Approval. But this mail is already too long, so I'll stop here.

My basic argument against Approval is that although IRV may make wrong decisions, it does not lead to as terrible situations as Approval does (with more than two plausible winners). In Approval the idea of all left wing voters approving all the left wing candidates sounds quite impossible. Therefore it is likely to violate the opinion of the majority. And the voters do not have any good strategies to fix the problem. Approving all the left wing candidates and letting a random one of them win, or to allow others (maybe the few strategic voters) to decide, does not sound like a system that voters would like to keep. In IRV people are (as we have seen) quite ignorant and don't understand that someone else than the ("fair") IRV winner should have won. The results are a bit random, but often people just think "better luck next time". So, "impossible situations" vs. "randomish elimination process".

Juho