[EM] Steph: Your rating method
MIKE OSSIPOFF
nkklrp at hotmail.com
Tue Oct 18 21:11:27 PDT 2005
Steph--
You wrote:
Some mail ago (adressed to Mike if I remember well)
was referring to range voting and saying that adding
the values or averaging was the only proper way of
using this information.
I reply:
Just summing each candidate's ratings is the obvious and meaningful way to
count them.
You could say that the winner is the candidate with the highest mean rating,
but then you've got to clarify that if you don't rate someone you give them
a default rating of zero. Just summing the ratings is much simpler to
define.
You continued:
Median_Ratings seems a far better way to handle that information.
Putting an artificially high or low value on a candidate would
almost never affect a candidate score, and if it does it would be
by almost nothing.
I reply:
What criteria would that method meet? What would those median ratings mean?
When you propose a method, it's for you to tell why it's better. What
criteria it meets. You point out that your method doesn't reward extreme
ratings. But you haven't told what criteria it meets or what the median
ratings mean, why they should be important.
Choosing the candidate with the greatest rating sum maximizes social utility
if people rate sincerely.
The resulting method meets FBC and WDSC.
That method is called Range Voting (RV). We used to call it Cardinal Ratings
(CR).
You continue:
Thus unsincere strategies appear useless.
I reply:
What? Are you saying that Gibbard & Satterthwaite were wrong? They showed
that every nonprobabilistsic method can sometimes give incentive for
insincere strategy.
For instance, with your method, do you claim that no one ever has incentive
to vote someone over their favorite? That isn't just a strategy incentive.
It's the biggest and worst strategy incentive.
Mike Ossipoff
_________________________________________________________________
On the road to retirement? Check out MSN Life Events for advice on how to
get there! http://lifeevents.msn.com/category.aspx?cid=Retirement
More information about the Election-Methods
mailing list