[EM] Relevance of Consistency

[Blake Cretney]

Well, my previous long post on a weakened consistency criterion kind of
blew up in my face.  So, here's a list of assertions that didn't survive
the light of day.  This won't make much sense if you haven't already
read that post, on the other hand I have trouble recommending you do so
at this point.

First, I was wrong about claiming that only Ranked Pairs fitted my
rule.  Markus's beatpath winner method does as well.  This means that my
rule doesn't declare a single winner in every case (in violation of my
1-Ranking property).  On the other hand, these cases are rather obscure.

More seriously, remember how I mentioned that a proof that that rule
would pass my consistency property was yet to come?  Well, I've now
found a contradictory example, so that could be a long wait.

Also, not so much an error as a misleading oversight, I neglected to
point out that because my consistency property is just a weakening of
the usual consistency criterion its easy to show that the normal
consistency passing methods are consistent by this standard too.

Anyway, I'm still going to hunt for a better group-decision rule.

---
Blake Cretney


----
For more information about this list (subscribe, unsubscribe, FAQ, etc), 
please see http://www.eskimo.com/~robla/em