On Sat, Nov 16, 2002 at 05:54:59AM +0100, Matthias Urlichs wrote: > > > Another alternative might have been to have the default option win if > > > it's _ever_ a member of the Scwartz set, rather than if it's a member > > > of the Schwartz set after the sequential dropping phases are complete. > That rule might be equivalent to the "drop all options defeated by the > default option" rule, or it might not be. I think it is, based on gut > feeling, but I don't know. It's not. Consider the case where there are three options: A, B, D D being the default option. Suppose people vote: 40 A B D 30 B D A 20 D A B then we end up with: 60:30 A > B 70:20 B > D 50:40 D > A thus eliminating A, and having B win, although if you went through the SSD process you'd eliminate D>A, and A would win. If the 30 people who'd voted "BDA" weren't sincere in that choice, and really meant "BAD", then that's allowing strategic voting to ensure their candidate wins. I'm not really sure that's such a bad thing, since it requires a majority of voters to engage in the tactic of ranking the default option above some other option, and even if you declare the above vote to have no result and go back to further discussion it doesn't stop the same 50 people ensuring A continues to not win (which is desirable -- if a majority of developers don't want to do something, they should be able to make sure that we don't do it). If people start taking this plan too seriously, we just end up with votes like: 40 A D B 30 B D A 20 D A B with 60:30 A > B 50:40 D > A 60:30 D > B and all our votes end up without results. > I'd rather run the algorithm with the full set of votes first, and _then_, > if the default option wins, have a separate rule on what to do next. Nope: see the first vote listed above. You _don't_ want to declare a result when the majority of developers would prefer further discussion or "none of the above". I suppose you could hack things so that "the default option defeats X" is never considered the weakest defeat, but I'm not sure that would do you much good. Cheers, aj -- Anthony Towns <aj@humbug.org.au> <http://azure.humbug.org.au/~aj/> I don't speak for anyone save myself. GPG signed mail preferred. ``If you don't do it now, you'll be one year older when you do.''
Attachment:
pgpcbPofCn29T.pgp
Description: PGP signature