On 6/24/23 22:01, Kevin Venzke wrote:
Hello, I saw that it would sometimes be convenient to have all
methods in one calculator, and this also lets me add some new methods
without including a lot of detail or finding a common theme for
them.
So I implemented fpA-max(fpC), MaxMin(PS), TACC, Benham, and
generalizations of Eivind Stensholt methods, as well as some other
ideas I have or haven't posted about before. I mark some criteria
too, such as when Smith candidates are covered. Hopefully of some
interest.
It's at: votingmethods.net/calc
Thanks.
We should be the ones saying thanks :-) So thank you.
But here's an idea about generalizing BPW. In my notation, it would be
fpB, because in an ABCA cycle, the candidate A beats is B, and we want
to reward the candidate whose candidate he beats has the highest first
preference score.
So perhaps max or sum fpB will work as a generalization? I don't know
what metrics you're using to evaluate whether a generalization is good,
so you'd have to check. Alternatively use some weak covering ideas to
eliminate all but three candidates, with A being one, one of the
candidates A beats being B, and C being a third (someone beating A?
Someone beaten by B?), then sum or max of fpB for every possible choice
of B (with eliminations before counting the first preferences).
-km
On 6/24/23 22:01, Kevin Venzke wrote:
> Hello, I saw that it would sometimes be convenient to have all
> methods in one calculator, and this also lets me add some new methods
> without including a lot of detail or finding a common theme for
> them.
>
> So I implemented fpA-max(fpC), MaxMin(PS), TACC, Benham, and
> generalizations of Eivind Stensholt methods, as well as some other
> ideas I have or haven't posted about before. I mark some criteria
> too, such as when Smith candidates are covered. Hopefully of some
> interest.
>
> It's at: votingmethods.net/calc
>
> Thanks.
We should be the ones saying thanks :-) So thank you.
But here's an idea about generalizing BPW. In my notation, it would be
fpB, because in an ABCA cycle, the candidate A beats is B, and we want
to reward the candidate whose candidate he beats has the highest first
preference score.
So perhaps max or sum fpB will work as a generalization? I don't know
what metrics you're using to evaluate whether a generalization is good,
so you'd have to check. Alternatively use some weak covering ideas to
eliminate all but three candidates, with A being one, one of the
candidates A beats being B, and C being a third (someone beating A?
Someone beaten by B?), then sum or max of fpB for every possible choice
of B (with eliminations before counting the first preferences).
-km
