Talk:Method evaluation poll 2005

From Electowiki
Jump to navigation Jump to search

Condorcet//Approval with FBC patch[edit source]

I put Condorcet//Approval with FBC patch under "ranked" methods rather than "cutoff" methods since I don't advise that one be allowed to rank among disapproved candidates in this method. That way, there is no burial incentive. Kevin Venzke

Good to know. James Green-Armytage 15:20, 16 Jun 2005 (PDT)


Rating methods[edit source]

Also, I was a bit lazy in giving all the CR methods a 4. I think CR is a pointless complication. Kevin Venzke 17:22, 15 Jun 2005 (PDT)

It is, but it still retains some Approval advantages such as FBC compliance. Besides, acceptance-psychology counts as a point, except in this pure-merit poll. [MO]

Yes, I noticed that; it was the only way I could understand your giving approval weighted pairwise a higher score than cardinal weighted pairwise, when AWP is just a limited version of CWP. I don't know if salability is part of your issue with CR, but remember that this poll deals with functional merit rather than salability. Actually, I think that it is more intuitive and easy for most voters to rate candidates on a 0-100 scale than to rank them with an approval cutoff. I suspect that many voters wouldn't understand the approval cutoff, and hence wouldn't use it. Just my opinion. James Green-Armytage 15:20, 16 Jun 2005 (PDT)
I am not considering salability, unless in that you include considerations of whether voters will be able to sensibly use the method. I think approval cutoffs and 0-100 rating are both likely to disappoint in this respect. I consider 0-100 even worse unless it's shown that the optimal strategy is not necessarily approval strategy. Kevin Venzke 15:49, 16 Jun 2005 (PDT)
Can you demonstrate that approval strategy is always the optimal strategy in cardinal pairwise? Because approval strategy is nearly always optimal in ordinary range voting, many people assume that it is also optimal in cardinal pairwise. But I haven't seen a demonstration of that yet. To make the demonstration even harder, assume that the maximizing in scale provision is used... James Green-Armytage 18:25, 16 Jun 2005 (PDT)
No, I can't. Kevin Venzke 23:59, 16 Jun 2005 (PDT)

Condorcet//approval w Cutoff[edit source]

I have joined here in search of a Condorcet method having certain features. I have looked over a few confusing pages like approval-sorted margins, and I have not yet found what I am looking for. Do any of the methods in this poll do the following: Rank all (but one) candidates (no truncation, no ties) and include an approval cut-off among them; the winner is the Schwartz set member approved by the most voters. Jrfisher 12:39, 17 Aug 2005 (PDT)

I will add Schwartz//Approval with explicit cutoff for you. Kevin Venzke 11:43, 18 Aug 2005 (PDT)