Talk:Baldwin's method

From Electowiki
Jump to navigation Jump to search

Cardinal variants[edit source]

Dr. Edmonds, given your interest in the cardinal variant of this system, have you seen IRNR? That might be similar to what you're looking to cover. BetterVotingAdvocacy (talk) 19:08, 24 May 2020 (UTC)

BetterVotingAdvocacy I am aware of Instant Runoff Normalized Ratings. IRNR[1] is very similar to Distributed Voting. The difference being that DV initially gathers cumulative votes but IRNR gathers score votes and converts them to cumulative. I would think that putting the mathematical burden on the voter is not a good idea so I favour IRNR. Anyway, the normalization used by both these systems is different from the rescaling used in Baldwin's method. I have had some discussions with User:Aldo_Tragni recently comparing the two. They both attempt to solve the issue with score voting that the utility scale depends on the candidate list. We have discussed this issue before here. The IRNR/DV normalization does something which attempts to get PR results if you stop the procedure with several candidates remaining. I am pretty sure it does fail since it does not have surplus handling. Instead Baldwin's method maximizes the difference between the most and least endorsed candidate and scales those in between. This gives every voter the same effective power at each round and would not punish a voter for scoring an unviable favourite high and scoring the compromise low. The normalization in IRNR still punishes this. Anyway, I thought that having such a page for what I would consider the best of this type of method would be useful for comparisons. I still prefer Score to Baldwins method as I think that strategy can compensate for the issue of scales more than it can for nonmonotonicity. Although, Approval has neither issue so maybe it is better than both. --Dr. Edmonds (talk) 20:49, 24 May 2020 (UTC)