[EM] Correlated Instant Borda Runoff, without Borda

Simmons, Forest simmonfo at up.edu
Wed Dec 28 14:35:18 PST 2005


Paul K. wrote ...
 

We were, in fact, talking about a science fiction short story written in the
1950s. The very idea of figuring out how to correlate such data has been a
thorny problem for AI researchers going even farther back.

But when I talked about the positives, part of it was serious. Exactly how
different is "the candidate might guess how the voters feel about all the
issues and fill out his ballot that way" than "the candidate might guess how
the voters feel about all the issues and tell them on TV that he feels the
same way"?

In the (fantasy) the process is double-blind - the candidates don't know who
the voters are and the voters don't have to put up with candidates.


Forest replies:
 
Good points.  Here's another variation on this idea:  Have the method do the nomination(s).
 
The voter whose questionaire "best correlates" with those of the other voters is nominated.  If that voter accepts the nomination and the other voters ratify it with an approval vote, then the election is done.  Otherwise, go to the second best questionaire, etc.
 
This could get rid of the plague of professional politicians.
 
The only problem I can see is that some mafioso claims to be the guy that filled out the winning questionaire, no matter who really did.
 
This would work better in a small group setting where everybody knows everybody else.
 
A variation for a large electorate would be this:  The hundred voters having the best correlation with the rest of the voters
meet together and nominate someone (possibly one of themselves) as candidate.   Since they have such high correlation among themselves it shouldn't be hard for them to come to consensus on this nomination.   This candidate is put up for approval by the whole electorate.
 
Forest
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/ms-tnef
Size: 5116 bytes
Desc: not available
URL: <http://lists.electorama.com/pipermail/election-methods-electorama.com/attachments/20051228/a24aa5df/attachment-0003.bin>


More information about the Election-Methods mailing list