[EM] Median and MCA/Generalised Bucklin (GB)
Kevin Venzke
stepjak at yahoo.fr
Thu Aug 7 20:29:02 PDT 2003
Gervase,
--- Gervase Lam <gervase at group.force9.co.uk> a écrit :
[...]
> From the polls, you know that one candidate is predicted to get a median
> rating of 7. There are some candidates that you think are better than
> him. However, your sincere utilities for the candidate is lower than 7.
> Therefore, you need to put the candidates who you think are better above 7
> on your ballot paper.
In my opinion, it would be impossible to get such a specific prediction.
There's not enough reason for voters to give a "sincere" rating.
> If this were a simple "Approval vote", then you would have no choice but
> to put the candidates in the top most slot. However, on the Median
> Ratings ballot paper, you have the choice of using slots 8, 9 or 10.
Yes, but as voters, do we really know how to put those slots to good use?
> Anyway, regardless of the above, isn't the better strategy to use in this
> 11 level Median Ratings vote the Approval Strategy? That is, use only
> slots 10 and 0.
In the zero-info case, when you don't have any predictions, you should vote
like that. Otherwise it might be possible to use a few of the top slots.
For instance, put your favorite at 10, and second-favorite at 9 (not 8, not 7:
there's no reason to be sincere). That means you won't cause a median 10 tie between
them. BUT, maybe someone worse has a median rating of 10, so you should've rated
second-favorite at 10. ...See? I think it becomes a huge guessing game.
> Or, with the prediction that the candidate is going to get a median rating
> of 7, may be you should put that candidate into slot 0 plus all of the
> ones who are worse, and then somehow spread out the remaining candidates
> between slots 1 and 10 inclusive?
You mean the winner is supposed to win with median rating of 7? What you should
do depends upon the prediction for the other front-runner(s), I think. But in
any case I don't think you should spread out the candidates. You should decide
whom to support, and make sure your vote counts. Forget about sincere ratings.
> I am open to comments/criticisms here...
>
> Thanks,
> Gervase.
I think three ranks is ideal, because each rank is unique, and has to be used in
a specific way. With even a fourth rank, it's not clear when you should make a
distinction in the top two ranks.
Kevin Venzke
stepjak at yahoo.fr
___________________________________________________________
Do You Yahoo!? -- Une adresse @yahoo.fr gratuite et en français !
Yahoo! Mail : http://fr.mail.yahoo.com
More information about the Election-Methods
mailing list