brnstnd@stealth.acf.nyu.edu (Dan Bernstein) (11/13/89)
In article <1989Nov11.034157.12415@alembic.acs.com> csu@alembic.acs.com (Dave Mack) writes: > So your idea of voter happiness is: "If I can't have a group > with the name *I* want, I'd rather not have any group at all. So there!" No. Dave, for each proposed name for the fish newsgroup, can't you quantify how much you'd like to have a newsgroup with that name? Can't you say how well you would accept sci.aquaria, or rec.aquaria, or comp.society.fish? My main criterion for acceptance is how well I think people will be able to find the group under that name. For example, I'd vote NO to comp.society.fish on this basis. Other people have different reasons; but whatever the reasons are, each voter can look at each name and decide how happy he'd be. Now do you understand what I mean by voter happiness? There is one problem: the scale between NO and YES is too sparse. There's not enough room for Dave to say ``I really want rec.aquaria above all else, but I'd prefer sci.aquaria and comp.society.fish to no group.'' That's why I've been advocating a scale between -10 and 10, with NO the same as -10 and YES the same as 10. Dave can then precisely express his vote: rec.aquaria YES (same as 10) sci.aquaria 7 comp.society.fish 3 In contrast, I would put comp.society.fish at -10, since I really do consider comp.society.fish worse than no group at all; and I'd put sci.aquaria around 5. Writing, tallying, and reporting these votes is just as easy as with straight NO/blank/YES. > I would rather see pure MAUVE implemented than > keep the current system. So why don't we just go ahead and use MAUVE without your ``separate name from creation'' stuff? You'll see that it works, and these discussions will cease. > But I think it would better still to take it > one step further and clearly separate the issues of creation and naming. The theory behind approval voting breaks down when you have separate votes on dependent issues. But even in practice, it's a bad idea. If every group name would antagonize more voters than it pleases, how can you possibly say that the group's a good idea? If the group would have failed under each name separately, how can it be created? > Whatever the outcome of any vote, some voters will not be > happy about it. I would like to think this method minimizes the > number of voters made unhappy. Too bad that it doesn't. If instead of the real names we had a vote choosing between comp.protocols.tcp-ip.aquaria and comp.society.fishies---both of which would probably make many voters very unhappy---you would have one of them created, just because people agree ``yeah, we need a fish group''? Let the group fail and wait for people to find the right name. ---Dan rave rave rave