In a time where people are worried about computer fraud in elections, would SPAV be best for hand-counting PR methods? You just sort ballots out into piles of how much fractional weight they have.
Hard yes. Not just easier, it’s the easiest one (if you exclude the party based methods, delegated methods, and non-deterministic methods).
Is it feasible to hand-count any of the PR methods that you didn’t exclude? (STV should work) I’m guessing something like RRV or SSS wouldn’t work at a statewide level, so does that rule them out? Is it at least possible to hand-verify the computer result with those methods?
I’d guess Sequential Monroe or any allocated method really should work decently with something like the lexicographic orderings from Jameson, or maybe even fractional surplus handling if rounded.
Also, is SPAV constitutional? Currently the two “equal vote” passing voting paradigms are either you get one vote per open seat (majoritarian) or one vote total to be spent (proportional); SPAV is sort of in between (one vote total that is spent in fractions and can basically never be exhausted completely; utilitarian-proportional)
RRV would be a lot easier then STV, so if it’s possible to hand count STV (it is. Australia has been using it before computers were a thing) then it’s DEFINITELY possible to hand count RRV without having to use computers.
In fact, even some optimal methods like harmonic voting and it’s approval ballot case regular proportional approval voting should also be possible without having to use computers provided that in districts with too many candidates running you use 2 rounds and only use such optimal methods in the 2nd round among the candidates that were qualified in the first round (or even conduct the 1st non-optimal round on the same ballots by only allowing the top n candidates who would win under the sequential version of the optimal voting method to be considered when calculating the winners under the optimal method). Though it would be a lot easier to get a sequential method enacted without having to use computers then it would be to jump through all the hoops required to get one of those optimal methods enacted without computers (though I think it’s still do-able and the possibility of enacting an optimal method (without computers) shouldn’t be prematurely ruled out).
Though with methods like harmonic voting, even if you did have to use computers, you can still validate the results by hand by seeing if the quality value of the single combination of candidates that actually won matched the quality value reported for that combination, and dooing that is really easy under harmonic voting and PAV. You could even randomly pick some other candidate combinations to be checked for good measure.
I went into more detail about this topic here.