On Mon, Mar 18, 2024 at 11:10 AM Dima Pasechnik <dimp...@gmail.com> wrote:
> It's very important to note that with multiwinner approval voting, merely > counting the votes per candidate and picking the top ones can lead to > rather unfair results > (unlike in the single winner case). > > For instance, if we elect k=3 candidates out of 6, say, $a,b,c,d,e,f$, and > out of N=19 people, 10 vote for $a,b,c$, and 9 - for $d,e,f$, then, with > approval voting, $a,b,c$ get elected (as $a,b,c$, get 10 votes each, more > than $d,e,f$), and almost half the voters, 9 out of 10, get no > representation of their views. > This is obviously bad - in such a case a fair outcome would be something > like $a,b,d$. Here "fair" has to be quantified, of course. > I've posted some details (and pointed at some solutions) on this here: > https://github.com/sagemath/sage/pull/37501#issuecomment-2004121053 > > It would be interesting to get the anonymised returned ballots and see if > we did well on this occasion. > As well, adjustments ought to be made along the lines outlined above. > The committee has agreed to release anonymized voting records, since the benefit gained for picking a better system for future votes outweighs the privacy risk (we've anonymized both voters and candidates). As I explained here <https://github.com/sagemath/sage/pull/37501#issuecomment-2004649597>, if we had stuck with 5 people on the committee then different system would have had different results, but after extending to 6 based on the approval-voting tie there was broad agreement: Approval Voting, Proportional Approval Voting, Phragmén's sequential rule, Maximin Support and Equal Shares all selected the current committee; only Minimax approval voting had a different result. The anonymized votes are here <https://docs.google.com/spreadsheets/d/12kbJepIKLpLKwLDum70FsSkqTto2St5gbspf3DyrPIc/edit?usp=sharing>, in case anyone wants to do further analysis. David -- You received this message because you are subscribed to the Google Groups "sage-devel" group. To unsubscribe from this group and stop receiving emails from it, send an email to sage-devel+unsubscr...@googlegroups.com. To view this discussion on the web visit https://groups.google.com/d/msgid/sage-devel/CAChs6_km71CEnaVUQX4T5mkGz6zMEZKLjfuXW3fsd1w50wNFSw%40mail.gmail.com.
