[EM] Jameson: Conditional methods and DH3

[MIKE OSSIPOFF]

Jameson:

That DH3 sounds like another way of saying "turkey-raising", which itself is just an instance of burial strategy, a word that you yourself used in
your description of DH3.

Good point. Thanks for pointing out that the conditional methods aren't entirely free of burial or turkey-raising. Like Beatpath, they have burial
strategy, and, like Beatpath, there is a risk that deters that strategy. In Beatpath, the Democrats might strategically vote Hitler or Gingrich over Romney, 
possibly resulting in the election of someone worse than Romney, even from the Democrat standpoint.

One difference is that, in the conditional methods, it takes several factions (sets of 1st-choice voters for different candidates) to elect the turkey, whereas,
with Beatpath, one faction of turkey-raisers is enough to bring turkey-victory.

I'd previously said that the conditional methods seem almost too good to be true. Well, _complete_ freedom from other methods' strategy problems
was too good to be true. 

But note a few other differences between the conditional methods and Beatpath: The conditional methods meet FBC and don't have the
co-operation/defection problem.

And "turkey-raising" and "burial" don't mean quite the same thing in the conditional methods. Nothing is the same in the conditional methods.

In the conditional methods, the turkey-raisers only benefit if the turkey's supporters want to conditionally give a vote to the turkey-raisers' candidate. They
probably just want to help a compromise (and maybe get coalition votes in return). And, if small candidate supporters know that the big-candidates' voters are unlikely to
risk helping a small candidate, because of the risk you describe, then the small candidates' supporters can make their support for a bigger compromise
candidate unconditional. That will probably be the usual practice. Conditional votes are specifically for situations like the ABE.

The conditional methods' strategy "problem" amounts to fishing for coalition vote-sharing--a rather innocent thing.

No method is entirely free of strategy. With the conditional methods, be careful how you fish, if you must use that strategy.

Say the method is AOC (that's optionally conditional Approval), and that there are several progressive factions, variously
favoring candidates P1, P2, P3...etc. The P1 voters (and the other progressives too)  consider anything worse than a progressive to be unaccepable, and they
consider P2 to be an acceptable compromise, to avoid a non-progressive win. But they don't like P2 or hir supporters as much as the other progressives. 
For that reason, and because all of the progressives, when voting sincerely, would vote eachother's candidates over the nonprogressives,
the P1 voters give to P2 a conditional approval. Also, in that way, they aren't giving to P2 something that hir voters can use
against P1 (as they would be doing if they gave P2 an unconditional approval).

That's an example of what the conditional votes are for, to avoid co-operation/defection.

Now, the P1 voters could gain votes, via coalition, by their conditional P2 approval. This could advance them with respect to the other
progressive factions. The other factions could respond by, themselves, giving conditional votes to P2. If they do, it's with the knowledge that
lots of factions making coalition with P2 could elect p2. But maybe they, too, consider P2 to be an acceptable compromise, against the
nonprogressives.

On the other hand, maybe they aren't as close to P2 as the P1 voters are. Maybe they'd prefer to just let the P1 faction help P2, to avoid
multiplying P2's coalition-support. It's their choice.

So, the coalition-benefit strategy, though there, is mild.

And the progressives should all be supporting eachother's candidates, and that would greatly reduce or eliminate the danger of P2 getting
an undue share of the coalition support. In fact, if, compared to P1,  P2 is less liked by P3, P4... progressives, then s/he'll get fewer approvals from them.

In other words, of course, as you said, you were only talking about competitive burial strategy. Without that burial use of conditional voting,
the problem you describe doesn't happen.

Of course if the nonprogressives (maybe correctly) perceive themselves having less votes than the progressives, then they, too, might
help P2, who is closer to them than are the other progressive candidates. Those approvals are likely to be unconditional, since, if there's a
perception of a progressive majority, then nonprogressive candidates can't expect votes from progressives.

By the way, one thing that I like about Approval is that, if it incorrectly appears that the progressives have a majority, then the nonprogressives
should, strategically, give a vote to a progressive. So that progressive-majority misconception, held by all, because everyone has the same polling data, won't hurt
the progressives.

Of course it works the other way too: If it looks as if nonprogressives have a majority, that could make them win. 

...except that, in both cases, principled, noncompromising voters, people who are basing their vote on acceptability vs unacceptability, aren't going to
cross the progressive/nonprogressive line with their approvals. In other words, Approval favors candidates preferred by principled uncompromising voters,
people who are sure about what is acceptable or unacceptable.

How many Democrat voters do you think vote for the Democrat because of uncompromising principle, rather than lesser-of-2-evils, hold-your-nose, voting?
>From what they themselves say, not too many.

Anyway,  the problem you speak of only occurs when voters are burying. When they're using conditional votes as intended, the problem doesn't happen.

Mike Ossipoff





 		 	   		  
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.electorama.com/pipermail/election-methods-electorama.com/attachments/20120216/097e608f/attachment-0003.htm>