[EM] Why Utility is more important than "transvestite Inversion Property" - reply to Venzke, Gilmour

[Warren Smith]

Reply to Venzke & Gilmour about Social Utility

Gilmour & Venzke have again expressed the opinion that Social Utility is merely
yet another voting system criterion, on a par with
Monotonicity, Favorite Betrayal, Condorcet, etc, and therefore my preference for
it is unwarranted, mysterious, biased, and/or somehow cantankerous of me.

They are wrong.

Social Utility is different than all those other property-criteria.
That is because it incorporates them all inside itself.

For example, every time a failure of monotonicity leads to a decrement in Social
Utility, the latter notices it.  And the latter notices it exactly as often as
it happens (i.e. weighting it with the right probabilistic weight) and
notices it exactly as much as its deleterious effect (i.e. weighting
it with the right seriousness-weight).  And a monotonicity failure might
actually in some situation lead to an INCREASE in social utility.  SU notices
that too, again with exactly the right weightings.

And SU does the same for every other property-failure criterion of that ilk,
not just "monotonicity."  Automatically.  With the right probabilistic
weight and seriousness weights.  Including for an infinite number of
important  properties like "Frogodocity" and "Perflumiousness"
that nobody has veer even invented or named yet and never will.  Every time.  Easily.

OK? Social Utility therfore trumps all the other property-failure notions
and is a superior measure of voting system quality.  By far.
Massively.  There is just no comparison.

You cannot argue in the reverse direction. For example, you cannot argue Monotonicity
somehow understands all instances of social utility decrease, weighting
them all appropriately.  This argument is asymmetric.   It only works in
my direction.  Not in the other direction.

It says social utility (Bayesian Regret, is my name for it, but essentially the same thing)
is the right way to compare voting systems and trumps the others.

Now.  You can ignore me.  You can say that you all want to consider your
random weird little proliferating notions of this property or that property,
forever arguing without resolution about how common and how serious they are.
Or: you can actually come to a conclusion.

Social Utility / Bayesian Regret allows coming to a conclusion.

I am not opposed to investigating properties to gain more understanding and thus
adding to the number of footnotes.  But I am opposed to deluding oneself into
dismissing the ultimate goal - increasing social utility - as yet another mere footnote.
That is a horrible mistake.

wds