[EM] utility - some agreement at last...

Warren Smith wds at math.temple.edu
Thu Sep 1 14:08:40 PDT 2005

>robla: The problem with placing paramount importance on "utility" in voting
methods is not that it doesn't exist, it's that there's no systematic,
fair way of measuring utility.

--WDS:  EXACTLY!!!!  GOOD!!!
However, Heitzig has repeatedly and clearly stated that it "does not exist."
I have repeatedly stated that it does exist, it is just hard to measure
and hard to get people to tell it to you honestly.

Great.  Now in fact, I completely agree utility is hard to measure.
Good.  Excellent.  I am glad to see Robla & I are in the same camp.
However, despite that hardness, it is nevertheless possible to use the utility
concept to make important strides.   Those strides are only possible once we
have admitted that utility exists.

What strides?  Well for one thing, we can try to formulate/characterize
optimal voting strategies, whereas without utility, we could not hope to do so.
For another, we can - even without measuring any human's utility for any event -
still compare different voting systems via "Bayesian regret measurements".  It is quite
a wonderful thing, but you can measure Bayesian regets of different voting systems
even without ever determining what any human might mean by saying Bush=64 versus Bush=65.

Now.  Are we to close our eyes to these wonderful advances  in voting theory
because "utility does not exist"?   Are we to adopt voting strategies that demonstrably
can cause wholsesale genocide and 100 year-long wars, purely because we refuse
to admit that utility "exists"?  Not very wise.  I recommend adopting the course
of admitting that utility exists, and then investigating what we can deduce from that

And I go even further: not only does utility exist, but it often
can be approximately measured, and on those occasions
should be measured by any responsible person trying to make any important


More information about the Election-Methods mailing list