[EM] How to fix the flawed "Nash equilibrium" concept for voting-theory purposes

[Thomas von der Elbe]

Hello,

Warren Smith wrote:
> For example, consider a 2-way election Gandhi vs Hitler in which 
> everybody votes
> for the (unanimously agreed to be) worst choice: Hitler.
>
> Well, that is a "Nash equilibrium" because no single voter can change
> the election result!
>
> Indeed, essentially every possible vote pattern in every possible
> large election, is a Nash equilibrium.    

If the election-method is proxy-voting, then a Nash equilibrium seems to 
not exist. Because you can then vote for a voter of Hitler. This in 
itself is already a payoff (because Hitler gets less direct votes). But 
even further it's possible to vote for a voter of Hitler who says 
something like: I delegate all my votes to Hitler until I have 50% of 
all votes, then I delegate them to Gandhi.

What do you think?

Thomas