Path: utzoo!utgpu!jarvis.csri.toronto.edu!mailrus!cs.utexas.edu!swrinde!zaphod.mps.ohio-state.edu!wuarchive!kuhub.cc.ukans.edu!sloane From: sloane@kuhub.cc.ukans.edu Newsgroups: news.groups Subject: Re: CfD: Interest Groups Surveys Message-ID: <22212.25d6c149@kuhub.cc.ukans.edu> Date: 12 Feb 90 19:59:37 GMT References: <22196.25d2eed0@kuhub.cc.ukans.edu> Organization: University of Kansas Academic Computing Services Lines: 50 In article , peter@ficc.uu.net (Peter da Silva) writes: > In article <22196.25d2eed0@kuhub.cc.ukans.edu> sloane@kuhub.cc.ukans.edu writes: >> 1. sci.astrology >> 2. rec.astrology >> 3. talk.astrology > > OK, I'll put this in terms of the modifications I just suggested. > >> Now suppose that voter 1 doesn't care whether an astrology group is >> created or not, but wants to vote against the name sci.astrology. What >> would this vote look like? > > Subject: 32 Doesn't this vote mean that I am voting FOR 2 and 3? This voter doesn't want to vote FOR anything, just AGAINST sci.*. >> Voter 4 really likes the idea of an astrology group, but hates the >> sci.astrology name enough that he/she would prefer to not have a group >> rather than name it sci.any. > > Subject: YES 32 > > (yes, this isn't as strong a vote as you might like, but given > the 2/3 rule it will act as a vote against sci) Again, there doesn't seem to be any way to vote against a name without voting for something else. I think preference voting (if you prefer that name) has a lot more going for it. Most of my votes in recent times have been against a name, not for anything The problem I have with STV votes (actually non-binding referendums, NOT surveys) is that they are designed to provide two different types of information about newsgroup creation. First, they need to establish that enough interest exists to warrant a newsgroup, and second they need to decide what the name should be. The current guidelines work well for the first purpose, but don't address the second purpose at all. STV, as least as I have seen it proposed so far, can answer the second question, but fails to answer the first question, since there are only votes FOR something, not AGAINST anything. In the examples above, in order to express an opinion AGAINST a sci.* name, a voter must vote FOR something else, thus biasing the "interest" survey. In the examples above, voter 1 would be counted as being interested in rec.* and talk.*, which is not the case. I think Preference Voting (PV) solves this problem, admittedly at the expense of finding the most prefered name. Rather PV will tend to find the least hated name, which may actually be a good thing. Perhaps the least hated name is the best compromise we can come up with. -- USmail: Bob Sloane, University of Kansas Computer Center, Lawrence, KS, 66045 E-mail: sloane@kuhub.cc.ukans.edu, sloane@ukanvax.bitnet, AT&T: (913)864-0444