Python governance vote (December 2018): Results

[tim.one (Tim Peters)]

Some notes on the tech aspects of the vote: most remarkable is how unremarkable it was Not only was there a flat-out Condorcet (“beats all”) winner, but if we throw that winner out, there’s also a flat-out Condorcet winner among the 7 remaining - and so on, all the way down to “further discussion”. For that reason, all of the Condorcet methods supported by CIVS compute the same total ordering. No ties, and no preference cycles, anywhere. The raw Condorcet criterion on its own was enough to resolve the full ordering. We were lucky in that way.

About a third of the ballots exploited the possibility to express ties. I was surprised that wasn’t larger, but it’s possible some didn’t realize they could; e.g.,

@jackjansen

How can some rows in the ballots listing have multiple entries with the same value? Or was it possible to vote for alternatives being equal, and did I overlook that?

Yup! In fact, if ;you didn’t change anything in the initial ballot presented to you, and clicked “Submit”, you would have “voted” that you had no preferences at all - the same as if you hadn’t voted.

Here’s a breakdown of the number of ballots expressing a given number of distinct ranks:

#ranks	#ballots
2	1
4	4
5	6
6	5
7	5
8	41

Only one ballot gave only two ranks, effectively pretending this was an Approval election:

1	8	1	1	8	8	8	8

Moral of the story: if we do this again, use score voting instead. It’s much easier to understand, is immune to preference cycles, and would almost certainly have yielded the same total ordering .