Michael Ossipoff was asked the following question:

Has anyone looked at the effect of using two or more alternating voting methods one after the other? Most government election systems are basically not just one vote but repeated voting on a regular schedule. Why not establish that you have a system where you switch off election systems like switching off between Approval voting one year and IRV in the next. You could take it further than just taking turns with two election methods, why not three, four, etc.? Perhaps using multiple election methods that are individually well-designed in multiple elections you would dilute some of the problems associated with individual EM systems. Thoughts?

Response by Michael Ossipoff:

Certainly it would be good to try out various methods, and alternating them in successive elections would be one way. My concern would be, however, that, under Green scenario conditions, if we did an Approval election year, the GPUS-preferrers might hope that the socialists will approve the Green, and so the GPUS-preferers might refuse to approve the socialists, hoping to get an easy win by taking advantage of someone else’s co-operativeness. The socialists, knowing that that might happen, might respond by publicly declaring that they won’t approve GPUS. Or maybe they’ll use the anti-chicken-dilemma strategy that I’ve describe–whereby the probabilistically partially support GPUS. But it would be easy to err regarding how much partial cooperation is best, and it could very well happen that the Republocrats would win do that chicken-dilemma meltdown.

That’s why I prefer methods that have no chicken-dilemma, and pass the Mutual Majority Criterion. Benam and Woodall are better still, because they additionally pass the Condorcet Criterion.

But suppose that the election is by rankings, counted by IRV, or by Benham or Woodall–Why not addiionally publish informational counts by other methods? I like the ideal majoritarian methods, MAM in particular. When there’s no chicken dilemma, they’re better than Benham, Woodall, or IRV. The trouble is that there could be a chicken dilemma. But there’s no reason to not do a purely informational rankings-count by MAM.  …or any other method that someone advocates.

IRV is the natural for the Green scenario, because it’s the only non-Plurality voting system mentioned in a political party platform. Five U.S. parties offer it. Four of them are progressive parties. IRV will be implemented if one of those parties wins. Benham is an easy and minimal change from IRV, to gain Condorcet Criterion compliance. It will be obvious why it would be better. There’s an excellent chance that Benham will quickly replace IRV. Not that IRV would _need_ replacing. Though Benham and Woodall are better than IRV, IRV is nevertheless fully good enough.  And it’s the voting system offered by five party platforms.

So, what I’m saying is that the question of which rank-count to offer isn’t in doubt in those parties. They know which one they want to offer. Sure, the EM people will quibble forever, but the parties will institute IRV if they get elected.

I now recommend MAM, which I consider the best ideal majoritarian method, for polling, because evidently chicken dilemma hasn’t been happening in polls. My most recent poll was officially a MAM poll. If there were doubt about which method to use in public elections, of course informational trial ballotings, counted by various rank-counts, might bring out a chicken-dilemma defection that succeeds in Beatpath, but fails in IRV, Benham, and Woodall.

Because successful chicken dilemma defection seems to not be happening in polls, the polls won’t be a useful practical comparison of {MAM, Beatpath} vs {IRV, Benham, Woodall} for official public political elections. Comparison of results shows MAM and Beatpath usually making output rankings that make more sense.  That’s because, under ideal majoritarian conditions, which seem to obtain in polls, the ideal majoritarian methods do the best. At such time as I do another household vote, I’ll probably use MAM, because I don’t expect chicken-dilemma defection in a family vote.

Back to the matter of alternating voting systems, MAM or Beatpath would look pretty good anytime that chicken dilemma defection doesn’t happen and succeed. So, in those alternations, MAM or Beatpath might look great. But who knows when the chicken dilemma could happen.