[EM] FW: Approval vs RV and rank methods

[Dave Ketchum]

Your subject implies a promise to compare among three methods.

Your content goes into great detail as to Approval vs RV, but very little 
as to rank.

Seeing rank as far superior to either, I read looking to see why you might 
disagree.

More below.

On Sun, 06 Nov 2005 05:46:59 +0000 MIKE OSSIPOFF wrote:

> Aside from RV's obvious advantage in familiarity, which is likely to be more 
> important than Approval's advantage of greater simplicity, there are 
> situations in which RV gives better results, or is fairer to the voter, than 
> Approval:
> 
> 1. By not requiring all-or-nothing ratings, RV mitigates the effects of poor 
> strategy and voter misjudement about the candidates. Better that the 
> misinformed voter not vote his misinformation full-strength, as he must do 
> in Approval.
> 
> 2. Sometimes someone doesn't know whether or not to vote for someone in 
> Approval. Approval then forces him/her to vote a decisiveness that he/she 
> doesn't feel. RV lets the person vote his/her indecisiveness. If you don't 
> know, why should you have to decisively vote all or nothing?
> 
> Those two advantages are related, of course. But I've been glad to have RV 
> in a #2 situation, while, in #1, I'm referring to voters other than me. It's 
> a question of not knowing and not feeling like expressing an all-or-nothing 
> rating, vs misinformation and strategic misjudement. Different situations, 
> but somewhat related.
> 
> The above are the reasons why I prefer RV to Approval as a public proposal.
> 
> But, other than those situations, other than public elections, Approval is 
> probably usually better.
> 
> In Approval, as in good rank methods (to a large extent), in public 
> elections, a 0-info strategic vote is the same as a sincere vote. In 
> Approval, for this purpose, I'm defining a sincere ballot as one that best 
> reflects your pairwise preferences. Either way, you should vote for the 
> above-mean candidates. I told why on EM a week or so ago.
> 
> Well, ok, one other RV advantage for committees and organizations is that, 
> if everyone is guaranteed to rate sincerely, then RV more reliably maximizes 
> social utility when there are few voters. With more voters, Approval's SU 
> roughness averages out.
> 
> But strategists can take advantage of sincere voters. Not in Approval, 
> where, if it's 0-info, and if we disregard the small-committee effects, 
> difficult to apply anyway, there's no difference between sincere and 
> strategic.
> 
> And Approval is easier to vote (except for when you can't decide whether to 
> vote for something or someone, and RV would let you split the difference). 
> Vote in a poll with lots of candidates, and then tell me which is easier. 
> Vote in James's EM wiki polls, between 50 voting systems, one by RV and one 
> by Approval, and then tell me which is easier.
> 
> Approval is easier to count. Sure, the counting-ease difference between 
> Approval and RV isn't a decisive consideration, but it's one more advantage 
> on top of the others.
> 
> Approval has an elegant simplicity unmatched by any other method. And the 
> more you examine Approval, the better it is.
> 
> None of this contradicts my advocacy of RV over Approval for public 
> elections.
> 
> And, as much as I like Approval, it's true that we're better off without 
> incompetent voters being forced to vote their incompetency full-strength, as 
> they'd have to do with Approval.


Time for comment:

Approval does not let competent voters vote their decisions full strength. 
  Easy enough to vote for A and against B.  Then there is no way in 
Approval to say for C that C is preferred over B BUT that this preference 
must not interfere with voting full-strength preference for A over C.

RV permits fine grained expression of such differences but, after giving 
voters headaches as to how to express them in more detail than rank 
permits, gives the counters headaches trying to decipher exactly what the 
voters meant.

For all of which reasons I prefer rank, especially for public elections:
      Voters can indicate order of their preferences.
      Counters can read these preferences in the same language as the 
voters used for writing.

But time to debate, another day, as to which variation of rank is best.
DWK

> 
> As I said, RV somewhat mitigates that problem. Good FBC-complying rank 
> methods, such as MDDA, MDDB, MDD,ER-Bucklin(whole), and MAMPO do a much 
> better job of mitigating that problem.
> 
> Just a few more comments:
> 
> I've said that, for public elections, I prefer the best FBC-complying rank 
> methods to Approval. But that's based on practical considerations that apply 
> to that particular voting situation.
> 
> Approval's beautiful simplicity, and the advantages that you get along with 
> that simplicity, make Approval look starkly, basically, naturally, general, 
> obvious and right--and makes all the other methods, such as even the best 
> rank methods, look contrived and arbitrary in comparison.
> 
> One gets the feeling that no one devised or invented Approval. Someone 
> merely discovered the obvious. That isn't true of any other method. In that 
> way, Approval makes all the other methods look like contrivances, 
> contraptions.
> 
> I'm not criticizing contaptions. They can be useful, and sometimes even 
> necessary. For instance, our voters in public elections might need one.
> 
> The voters haven't been tried with Approval, so of course no one can say 
> whether they'll be able to give up their lesser-evil when their favorite 
> outpolls their greater-evil.  But their behavior so far is not promising in 
> that regard. So one can't be very confident when they're forced to vote 
> their judgement full-strength, all or nothing. Still, if Approval were 
> especially easy to enact, it would certainly be worth a try.
> 
> The RV contrivance, as I said, mitigates bad judgement by allowing less than 
> full-strength voted preferences. MDDA, MDDB, MDD,ER-Bucklin(whole) and MAMPO 
> do even better by making it safe to rank one's lesser-evil below one's 
> favorite, thereby casting a full-strength for Favorite over Lesser-Evil, 
> while casting a full-strength vote for Lesser-Evil over Greater-Evil.
> 
> So we may well need contrivances or contraptions because of a lack of voter 
> competence.
> 
> Someone might object that Approval doesn't receive all the information. No, 
> it receives and counts the important information.
> 
> But with our voting situation and electorate, there may be a 
> pathologically-caused need for more information.
> 
> Mike Ossipoff

-- 
  davek at clarityconnect.com    people.clarityconnect.com/webpages3/davek
  Dave Ketchum   108 Halstead Ave, Owego, NY  13827-1708   607-687-5026
            Do to no one what you would not want done to you.
                  If you want peace, work for justice.