[EM] Feature extraction and criteria for multiwinner elections

[Kristofer Munsterhjelm]

Paul Kislanko wrote:
> See below for the context.
> 
> I thought aloud a few years ago that one way to select a "winning ranked
> ballot" would be to find the ballot that is "most like" the consensus
> ballot. Order the candidates by their Bucklin rank and ballots by the
> Kendall tau "distance" that each ballot is from the Bucklin rank. The ballot
> with the least tau distance from the medians is then used to decide the
> election.
> 
> (Based upon a science fiction short-story from the 1950s, but reasonable
> from a mathematical perspective.)

That sounds a bit like Kemeny. How does it relate to multiwinner 
methods? (Also, what story?)