[EM] The critical importance of Precinct Summability.

[Kristofer Munsterhjelm]

On 2024-08-13 17:23, Filip Ejlak wrote:
> Aside from the general usefulness/importance of summability, in this 
> specific case the Venezuelan system's summability does not make a 
> difference in exposing the election as stolen. The (real) winner got, 
> like, 70% of the votes? In such a situation, you don't need the method 
> to be summable, it being majoritarian is enough. Similarly, if there's a 
> non-summable Condorcet method but election results have no Condorcet 
> cycle, we can still verify the winner with summable information. Of 
> course cycles do happen sometimes, so the question remains if this would 
> be transparent enough or not.
> 
> Perhaps a better way to do presidential elections - I don't know, giving 
> it up for debate - would be to do an actual, manual 2nd voting round 
> with two or three candidates from the Smith set (only if there's a top 
> cycle, of course). One-round summable methods' vulnerability is quite a 
> disadvantage, but two summable rounds might do the trick in this respect.

That could well be a good idea, at least for the "noise" sort of cycle 
that has been observed so far. The 2021 Minneapolis Ward 2 election 
seems to have been mostly up to random noise: 
https://electowiki.org/wiki/2021_Minneapolis_Ward_2_city_council_election

and focusing on the Smith set for a runoff would have made clear if 
there were any real differences in voter opinion between the three.

I do know that some jurisdictions (particularly American) really don't 
like runoffs, so that may count against it. And if the political system 
evolves (due to supporting a wider range of opinions), then it's 
possible that cycles could start happening more often.

-km