3 ms·
I think instant runoff voting is optimal, but it's too confusing for an idiot in a hurry. I'd be content with simple approval voting. Every candidate has a chec
by _archon_ 10y ago
I think instant runoff voting is optimal, but it's too confusing for an idiot in a hurry. I'd be content with simple approval voting. Every candidate has a checkbox next to their name on the ballot. If you approve of the candidate, check the box. Most checks wins. Ballot will have a lot of names, but that's the price of democracy.
- dragonwriter 10y ago> I think instant runoff voting is optimal It's very nearly the worst actually proposed single-member ranked ballots system; I can't imagine any way in which it is optimal. It's better than FPTP, but that's a pretty low bar.
- ScottBurson 10y agoI too think AV (Approval Voting) is the best. It's bog simple to explain [ETA: and to implement using existing voting equipment], and completely immune from both strategic candidate selection and strategic voting. Its only downside, IMO, is reduced detectability of vote count tampering: there's no simple relationship between the number of voters and the number of votes, as there is with plurality voting or even IRV. So if you're going to use AV, you have to have an auditable paper trail. That's a very good idea anyway, of course, and lots of places are moving in that direction, so I don't think this is really a barrier. (For very small numbers of voters, such as for elections within an academic society, Rating Voting is even better, but its advantages fade as the number of voters increases. AV is just the limiting case of RV with only two values.)
- vinay427 10y ago> there's no simple relationship between the number of voters and the number of votes, as there is with plurality voting or even IRV This problem still exists in FPTP where some voters may opt not to vote for any candidate for a position.