sommar@enea.se (Erland Sommarskog) (10/18/89)
(I tried to post another article with similar content, but
I don't think it made it. Apologies if I am redundant.)
Several ideas have been presented here to improve the voting
scheme to handle name conflicts. My idea is similar to some
of the presented, and is a simplified form of the single
transferable vote system. Also, it gives room for a name
czar, or rather a name supervisor.
First the name supervisor(s). (I think there should be more
than one, but not more than 5-6 people.) When a group is
proposed with what they think is an inappropriate name,
they have the right to dictate that what they think is a better
name should be considered. This is implemented so that the
call for votes must cover both names and give them equal
attention. (Assuming that Greg Woods is one of the supervisors,
they have control of news.announce.newgroups.)
When you vote for such a call, you have ten possibilities:
(A is original proposal, B is supervisor's idea.)
YES to both NO to both
YES to both, preferring A YES to both, preferring B
YES to A, NO to B YES to B, NO to A
YES to A YES to B
NO to A NO to A
In the latter four cases you don't have any opinion on creating
a group with the name you don't vote on.
Tallying the straight votes are trivial. The preference votes
(YES to both, but I prefer X) is counted to X in the first
step, where we compare the two names with each other. Winner is
the name with biggest difference between YES and NO votes in
absolute numbers(*). After this the loser's preference vote
are added to the winner. If the winner now passes the 100
more-YES-than-NO limit the group is created with that name.
The advantage with this system over STV and the one proposed by
Bob Sloane is that it is much simpler.
It can be used to resolve sci/rec.acquaria issue, but it cannot
be used for resolving the discussion we had about comp.object.
I have retained the preferrence mechanism from STV, but to be
honest, I don't think it is that terribly important. It would be
nice, but...
I think the point with the name supervisor is important. Without
that we can sit here say TALK.whatever. The group champion will
still call for votes on sci.whatever, and too many would vote in
favour, since they care more about the group than the name. With
this model, they would at least get an alternative.
Why not let the name supervisors be name czars and dictate the
name from the start? Because that would cause to a deluge of flames,
and even more important: name czars are human too, and can make
erroneous assumptions.
(*) One could think that percentage would be better that absolute
difference, when choosing the winner. This is OK, as longs as the
creation criterion is also percentage as Chuq wants it to be. But
if we have different criterions, we run the risk of getting anomalies.
Example:
A: 400 YES 250 NO
B: 50 YES 1 NO
B wins but does not pass the creation criterion.
--
Erland Sommarskog - ENEA Data, Stockholm - sommar@enea.se
"My baby's a 26. On a scale from one to ten, my baby's a 26." - Chicsloane@kuhub.cc.ukans.edu (10/18/89)
In article <372@enea.se>, sommar@enea.se (Erland Sommarskog) writes: > First the name supervisor(s). (I think there should be more > than one, but not more than 5-6 people.) When a group is > proposed with what they think is an inappropriate name, > they have the right to dictate that what they think is a better > name should be considered. This is implemented so that the > call for votes must cover both names and give them equal > attention. (Assuming that Greg Woods is one of the supervisors, > they have control of news.announce.newgroups.) Unfortunatly, I doubt that this scheme would work for the present aquaria conflict. I seem to be seeing 3 major camps, the sci.* people, the rec.aquaria people, and the rec.pets.* group. Recently, there has also been some interest shown in misc.*. Frequently, just two choices aren't enough. > The advantage with this system over STV and the one proposed by > Bob Sloane is that it is much simpler. Hmmm. "Simpler" is tricky thing to pin down. How is "Yes to Both" simpler than "Yes to any name?" Your scheme is "simpler" in that it restrict the choices to only two names. What else does it "simplify?" > It can be used to resolve sci/rec.acquaria issue, but it cannot > be used for resolving the discussion we had about comp.object. My scheme would have worked for either. > I think the point with the name supervisor is important. Without > that we can sit here say TALK.whatever. The group champion will > still call for votes on sci.whatever, and too many would vote in > favour, since they care more about the group than the name. With > this model, they would at least get an alternative. This is the problem I was trying to avoid. In my scheme, the vote taker would call for votes on a particular topic, as stated in the charter, not for a particular name. The name for the group would be decided by the vote. -- 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 "The scientific theory I like best is that the rings of Saturn are composed entirely of lost airline luggage." -- Mark Russell
sommar@enea.se (Erland Sommarskog) (10/22/89)
Bob Sloane (sloane@kuhub.cc.ukans.edu) replies to an article of mine: >In article <372@enea.se>, sommar@enea.se (Erland Sommarskog) writes: >> First the name supervisor(s). (I think there should be more >> than one, but not more than 5-6 people.) When a group is >> proposed with what they think is an inappropriate name, >> they have the right to dictate that what they think is a better >> name should be considered. This is implemented so that the >> call for votes must cover both names and give them equal >> attention. (Assuming that Greg Woods is one of the supervisors, >> they have control of news.announce.newgroups.) > >Unfortunatly, I doubt that this scheme would work for the present aquaria >conflict. I seem to be seeing 3 major camps, the sci.* people, the >rec.aquaria people, and the rec.pets.* group. Recently, there has also been >some interest shown in misc.*. Frequently, just two choices aren't enough. I haven't been following the acquaria conflict. (God, it's real fun to press "k" and see 40 articles just go away!) One could of course extend the scheme so that the vote taker may include how many names he like. This would lead us to the proposal Alien Wells had. But the idea my my scheme was not to settle all name disputes in the votes, only those when the group champion goes for a completely inappropriate name as the acquaria case. It is part of the scheme that the name supervisors has to come up with a name that is likely from to be accepted from the newsreaders. They may think that rec.pets.fish is a suitable name, but no one into acquaria think so, and sci.acquira wins, which is unnecessary when rec.acquaria would have had no problems. (Here's another motive for name supervisors rather than name czars. A name czar can't be expected to grasp all areas, and may miss subtleties that are clear for those interested in the subject.) >> The advantage with this system over STV and the one proposed by >> Bob Sloane is that it is much simpler. > >Hmmm. "Simpler" is tricky thing to pin down. How is "Yes to Both" > simpler than to any name?" Your scheme is "simpler" in that it > restrict the choices to two names. What else does it "simplify?" To be honest, I didn't understand your idea. I only glanced it, but if it can be understood at first glance, then it is too complex. Greg Woods has a point when he talks about verification- >> I think the point with the name supervisor is important. Without >> that we can sit here say TALK.whatever. The group champion will >> still call for votes on sci.whatever, and too many would vote in >> favour, since they care more about the group than the name. With >> this model, they would at least get an alternative. > >This is the problem I was trying to avoid. In my scheme, the vote taker would >call for votes on a particular topic, as stated in the charter, not for a >particular name. The name for the group would be decided by the vote. There will always be some "leading" proposal and not all group champions will includes names they don't like. If you send out a vote on sci.acquaria and has to put in rec.acquaria yourself there won't be much difference from today. People will like the subject and vote YES and sci.acquaria will pass despite some spurious votes for rec.acquaria. (Since many of those opposing sci, won't vote for rec.*, since they vote because of their interest in the name space, not in acquarias.) Your idea is more aimed at cases like comp.object and rec.radio.shortwave where the vote taker has an honest desire to get the most appropriate name. Although resolving those problems in a vote would be nice, there is the verification problem. Chuq said earlier that these problems were best solved with pre-vote polls. I'm not really satisfied, but there is a point. You can't win them all... -- Erland Sommarskog - ENEA Data, Stockholm - sommar@enea.se "My baby's a 26. On a scale from one to ten, my baby's a 26." - Chic