There seems to be some disagreement among the various polls. Nothing
unusual there. But let me mention a few criteria that are important
to consider. For one thing, are the polls supposed to measure and rank
a team by its whole season, or by how strong the team is RIGHT NOW?
To me, the ranking at the moment is most important to "know" and
therefore a system of weighting recent games over less recent games
should be a component of a good poll IMO. For example, would you, as
a theoretical opponent, want to play Minnesota now or back in Dec.? How\
about Maine--play them now or in Dec.? I think, obviously, that
Minnesota should not be overly penalized for their early season pratfalls.
   Another component of a good poll is to adjust very lopsided scores
to "merely" reflect a good beating. I don't believe that 11-2 is sig-
nificantly worse than 4 or 5 to 1 in the whupping dept.
   Yet another component of a good poll, for me, would be a disregard for
empty net goals. Many games end up 4-2 or even 5-2 which were REALLY
3-2 games.
    In these matters, I am referring to COMPUTER polls (which sometimes
have such quantified parameters) but the same ideas should prevail in
opinion polls too. In my opinion.