[EM] Yes/?/No

[Juho Laatu]

One problem with options (2) and (3) is that they make buying and selling votes possible (by the proxies). How about using some IRV like heuristic approach in option (3)? The "?" ratings could be automatically raised to the "Yes" category or lowered to "No" depending on who the forerunner/forerunners are at each stage of the counting process. I don't have any detailed proposal on how that could best be done, but maybe some sensible methods could be constructed.

Maybe also a one shot approach, where all "?" ratings would be automatically changed to "Yes" or "No" at one go, depending on the voter given "Yes" and "No" ratings, would be possible. (i.e. not having multiple serial IRV style rounds)

I'm always a bit afraid of negative votes because of the Dark Horse and related problems. It is quite common that elections have two or three well known strong candidates, and as a result of that competitive setting, those candidates may also get lots of negative votes, although there may be also many other candidates that are clearly worse than they are. (maybe in up to 2/3 of the ballots when there are 3 strong candidates)

In elections where voters are expected not to rank or rate most of the candidates, because of the hight number of candidates, rating/ranking only some of the best candidates is a good practical approach. Ranking some of them might not be much more difficult than approving some of them (and maybe ranking some of them as "No" or "?"). One could also intentionally limit the number of candidates that one can rank, or the number of ranks that one can use. In the latter case the ballot could be quite similar to the Yes/?/No ballot. The ratings/ranks could be e.g. 1/2/3 and default last position if not marked. Multiple candidates could get the same rating/ranking. My point here is to compare the amount of information, difficulty of voting, impact of negative votes, and fairness of results in elections with long list of candidates and only few rated/ranked candidates.

BR, Juho