Patrick's idea of changing the bracketing so that 1E plays 2E and 1W
plays 2W if necessary to maintain the pairings occurred to me last
night, and makes a certain amount of sense given that the 1E-2W, 1W-2E
bracketing is something of a historical accident.  However, the
tournament bracket in the appendix of the Championships Manual clearly
spells out the existing bracketing, so I don't see how they can change
it.  (And in general, I think the cdommittee changing the rules in
mid-season, even if we don't like the old rules , is a Bad Thing.)

On Thu, 14 Mar 2002, Patrick Abegg wrote:

> Two Cornell wins produces one of those "Pairwise Paradoxes" where they lose
> to BU in the direct comparison, but beat some other teams that BU will not
> (unless they win out as well). This could lead to some discussion on the
> committee as to exactly how to apply the comparison rules. My guess is that
> they use the direct comparison and give the bye to BU.

But the thing is that Cornell and BU are not the only teams in the
running for that bye.  You could just as easily say they should look
at the direct comparison between St. Cloud and BU and give it to SCSU,
or between Cornell and SCSU, and give it to Cornell.

The comparisons among not-obviously-irrelevant teams in the scenario
where all the conference tourney games are won by the higher seeds are
as follows:

 3 Minnesota (W)    5 .6226    __SCBUMSMiMe
 4 Cornell (E)      4 .5893  Mn  SC____MiMe
 5 St Cloud (W)     3 .6059  ____  BUMSMi__
 6 Boston Univ (H)  3 .6058  __Cr__  MS__Me
 7 Mich State (C)   3 .5930  __Cr____  MiMe
 8 Michigan (C)     2 .5844  ______BU__  Me
 9 Maine (H)        1 .5807  ____SC______

Two of these teams have to get byes.  It's pretty clearly not going to
be Michigan or Maine, and once you throw out these, the number of
comparisons won would be 3-2-2-2-1.  Tossing out Michigan State (the
1) gives a choice of two out of Minnesota (beats SCSU and BU), Cornell
(beats Minnesota and BU), SCSU (beats BU) and BU (beats Cornell).  So
looking at the comparisons among the four teams in the running for the
last two byes, Minnsota and Cornell win two out of three and the
othewrs win one out of three, so by this argument, Cornell and
Minnesota would get the byes.  That's not to say this is what they
*would* do in this case, but it's the algorithm used by my script to
make the suggested bye choices, and it's similar to the resolution of
an analogous tangle for the last two at-large bids in 1999.
Unfortunately, past experience with NCAA indicates that there's little
point in trying to get an answer about how to resolve this situation
unless it actually occurs, and even then we might not get an
explanation of what was done.
                                          John Whelan, Cornell '91
                                                 [log in to unmask]
                                     http://www.amurgsval.org/joe/

Enjoy the latest Hockey Geek tools at slack.net/hockey