Do we wish to satisfy various simple, narrow requirements (criteria), or do we wish to attain substantial, consequential goals (objectives)?
I have long been advocating (what I see as) the simplest form of score/range voting, 1 to 10, plus explicit abstention, because I saw it as a way to overcome ‘party lock-in’ (usually termed ‘two-party domination’). (And the electorate would still need to use a ‘hedge strategy’ for that to work.) Actually, the problem it poses would be just as bad if there were a dozen different parties, if they were all fronted by one common movement. And the problem is truly far worse than one of simple domination – the electorate has been totally locked into a two party regime all along. Disruption of party lock-in appears to be much more of a hard-to-attain substantial objective, than simple, narrow criterion. So, here we are dealing with a substantial objective, rather than some simple picked-and-chosen criterion.
Many, many people dwell at great length upon things like the Condorcet criterion. Really, what does meeting this very attractive, albeit narrow criterion actually accomplish in the real world? To satisfy this would do nothing to advance the disruption of party lock-in, which is a very substantial objective. Yet people fixate upon it relentlessly.
Then there’s the ever-fashionable ‘majority favorite criterion’. This beauty insists that if a candidate wins over 50% of the ‘first place votes’ then he or she ought to be declared the winner. For example, in a (currently ubiquitous) ‘choose-one’ method election, if a candidate wins over 50% of the votes, they win, and that surely sounds reasonable. Now, in a ‘RCV’/‘IRV’ method election, the weakest first-rank-place candidates are ‘eliminated’ on individual ballots, and the second-rank-place candidates on that ballot are ‘transferred upward’ to the first rank-place, and this ballot editing is repeated until some candidate achieves an artificial ‘majority plus one’ number of first-rank-place ‘votes’. Since this scheme eventually provides somebody with over 50% of the (edited) first-rank-place ‘votes’, the system is claimed to meet the ‘majority favorite criterion’.
But with score voting, in an election with 10 voters, some candidate might get 8 voters to grant 10 votes, for a total of 80 votes, while some other candidate might get 10 voters to grant them 9 votes, for a total of 90 votes, and the latter candidate would win (in the simplest case). But notice the former candidate (who lost) had won 8 of the ‘first-score-place votes’ while the latter (who won) had won (in the simplest case) only 2 of the ‘first-score-place votes’. This must be a dreadfully wrong violation of the (‘crucially important’) ‘majority favorite criterion’. Never mind that it has nothing whatever to do with any substantial, consequential objectives. This should illustrate the problem with the criterial fixation mentality.
Let us suppose that RCV actually meets the ‘majority favorite criterion’ often via weak candidate eliminations and ballot editing. Then, there is surely no reason to suppose it cannot also meet a ‘supermajority favorite criterion’. All the vote tabulators need to do is go beyond the point where somebody has more than a majority, and simply continue to eliminate weak candidates, and edit the appropriate ballots, until the winning candidate obtains 2/3rds of the (artificial) votes. Surely winning by 2/3rds is far better than having a paltry majority plus one. And while we are at it, why stop there? Why not go all the way to glory and just continue the eliminating and editing until the winner achieves a unanimous victory. Yes, ranked choice voting actually furnishes a unanimous victory in every single election.