[EM] Fluffy the Dog and group strategy

Richard Moore rmoore4 at home.com
Fri Sep 14 22:37:45 PDT 2001

Craig Layton wrote:
> The important
> consideration is that almost all of the voters are highly dissatisfied with
> the outcome, therefore the result isn't a good interpretation or
> representation of the voters' preferences.

In this race, and in many real-world races, there will be 
bad outcomes regardless of the method because either there 
are no good candidates or the electorate is simply too 
divided to choose a compromise. By "good candidates", I mean 
those that high percentages of the voters find acceptable.

> Because I don't
> believe that ratings can be expressed non-relatively, I'm not so
> enthusiastic about mixed methods that seem to regard 50% + 1 of approval
> votes as some magical passport to a political mandate.

I agree that ratings are not absolute. I also agree that 
there is something a little arbitrary about "50% + 1" 
approval. It might be nice if we could find leaders who get 
70% approval or better. The election method "X" and the 
acceptability criterion "Y" in the meta-method I wrote of 
are variables -- they should be chosen depending on the 
application. Some people may like Y to be "50% + 1". At 
least this level of approval suggests that most people 
prefer this winner to the expectation from a random dice 
throw. That isn't saying a whole lot; it's just a safeguard
that might keep the likes of Fluffy from running the 
government. Adjust the level of the safeguard, like a 
thermostat, until you are comfortable but still have a good 
chance of having a winner meet the criterion.

Unfortunately, adding some criterion Y to public elections 
will at times incur the costs and time delays of holding 
(an) additional election(s). This idea would be resisted by 
the public, at least here in the U.S. where it is unfamiliar 
to most and runs up against constitutional requirements 
(Amendment XX) of having terms of office begin and end on 
specified dates.


More information about the Election-Methods mailing list