[EM] Condorcet Criterion for plurality.

[Martin Harper]

LAYTON Craig wrote:

> Okay.  Maybe I'm understanding it differently to Markus.  I haven't read the
> literature he refers to, but I would think it was fairly straightforward.  A
> voter has a set of preferences (A>B>C) either sincere or insincere, it
> doesn't matter as long as those are the preferences a voter intends to vote
> (not strategic).  You posit that the preferences exist,

Ok - I don't think that is accurate.

I have a sincere set of preferences. I do not have a set of preferences I intend
to vote. How I intend to vote will not be known until I know what the voting
system is - how I may mark the ballot paper - and so forth. It will also change
depending on my knowledge of the rest of the electorate.

I understand that all these factors in deciding how to actually vote are
strategic influences - but I don't think that changes the fact that until I
*know* how much information will be available - until I know that I'm in 0-info
plurality, for instance - I don't have a how to vote card.

> even when they
> cannot be fully expressed (say, in Approval or Plurality).  The preferences
> of plurality voters would show that there can be a Condorcet winner who may
> not get elected under plurality.

Part of my distaste is that I'm not sure that this hack is even necessary -
plurality is probably flawed enough that you can miss the sincere CC even if all
the electorate have perfect information on the sincere preferences of all the
rest of the electorate. I'll be back after some scribbling... :)