A few introductory observations, followed by some substantial
information, followed by analysis and opinion.
 
        1) The NCAA tournament seedings illustrate why Charlie and I
have degrees in computer science and physics rather than psychology.
        2) In planetary science, there was a statistical observation
called Bode's law, which identified a pattern in the spacing of the
innermost half dozen or so planets plus the asteroid belt, but which
appears to have been just a coincidence.  It's important to keep that
in mind when trying to draw definitive conclusions from small numbers
of data points.
        3) We really are making progress in our knowledge of the
seeding process.  There were absolutely no doubts about which twelve
teams made the tournament, who got byes, and how the teams were to be
seeded withing each region.
 
        Now the meat: I asked someone in the NCAA for an explanation
of the seeding process, and was told the following:
 
"  the committee just liked the arrangement that
" played out this year with a strict "top four stay in the region, bottom
" two move" interpretation.  They talked about an Ohio State-Wisconsin
" switch because they were sensitive to the just-played Michigan
" State/Ohio State game, but decided against it.  They were not very
" concerned with a possible Boston U./New Hampshire meeting because those
" teams have not met since December and went 1-1-1 vs. each other when
" they last played (kind of a deciding game sort of thing).
 
" The committee also did not want to move New Hampshire because of its
" drawing potential in Albany and the fact that it was rated higher than
" both Yale and Princeton.  The Ohio State/Wisconsin question was pretty
" much it and they decided to keep things as they were after following
" orders to: avoid first-round conference matchups, move two teams from
" each region (preferably the bottom two) to provide an East-West flavor.
 
[I was going to paraphrase that, but I figured a direct quote would
minimize the confusion.]
 
        Okay, now for the analysis.  We have known for a while that
the committee's decision on which teams land in which regionals was
based on three factors:
        -- Pairwise comparisons: all else being equal, the bottom two
teams get swapped
        -- Avoiding conference matchups (although if this could be
done by rearranging teams within each regional, that was preferrable
to swapping teams between regionals)
        -- Maximizing attendance at each regional
 
        I thought we were developing a sense of under which
circumstances each of the priorities were more or less important
(avoid conference matchups if possible; do it by shuffling seeds
within regions if you can, but use placement of teams in one regional
or the other if neccessary, unless you feel attendance considerations
should take precedence).  But clearly that's not the case, and there
may be nothing to do under the current system but be prepared for
unexpected results in regionals placement.
 
        From the experience of 1996 (see my other post), we knew
attendance could override comparisons and matchups (sending Lowell
West in place of Cornell) and matchups could override comparisons
(sending LSSU East in place of Minnesota).  I was assuming that, in
cases where a conference matchup could only be avoided by sending
different teams to different regions, that's what the committee would
do, unless they played the attendance card.  But I think we've learned
here that unless all the priorities are in agreement, we can't say
what the committee would do in even a relatively simple situation.
 
Now a bit of backtalk:
 
Adam Wodon says
 
>Well. .... Congrats to Keith Instone, who was right on the money ... You
>see everyone --- so "Cut and Dried" that the selections and seedings
>were known as soon the games ended.  Does this make believers out of the
>few remaining holdouts?
 
        The selections were known, as were the byes and the seedings
within each region.  But as which teams were placed in each region,
that was *not* known.  Keith assumed correctly that the committee
would ignore conference matchups and go by the numbers.  Charlie and
I, based past years, and the claim that the NCAA doesn't want
intraconference games in the regionals, assumed incorrectly that they
would make two simple changes to avoid those matchups.
 
        There are lots of other statements from the last 36 hours I
have responses to, but in the interest of time (mine), let me just
make a few more statements of my own.
 
        As Adam pointed out, This is not precisely the same situation
as with Minnesota last year, since that involved shifting teams within
a region.  However, between the example of Lake State in 1996 and the
fact that they thought about shipping Ohio State East this year (even
if they didn't do it), I'd say that it is *not* true that they will
never use placement of teams in regionals to avoid conference
matchups.  I believe the situation is this:
 
        -- If they can avoid a conference matchup by shuffling seeds,
they'll do it; with only two priorities (pairwise comparisons and
avoiding intraconference games), it's easier to manage.  This also
means that if given the choice of shuffling seeds or shifting teams
between regions to avoid a matchup, they'll leave the regions intact.
 
        -- They may still shift teams between regions to avoid
conference matchups, but there's an additional issue to consider in
these cases, namely attendance.
 
        Now it does seem a little bizarre to me that faithfulness to
the comparisons is more of a consideration when determining who gets
to stay in their own region than in seeding the regionals, and I see
where J. Michael Neal's beef is coming from.  Of course, my personal
opinion is that considering the potential draw is unfair, while paying
strict attention to conference considerations is just fine.  Call me
an idealist.
 
        I am disappointed that the selection procedure turns out to be
less of a deterministic algorithm than I'd thought.  And it does seem
that the issue of HE losing a representative on the committee is a
potentially significant one, with decisions on who plays where being
left largely up to the committee's discretion.  But at least the
determination of who's in the tournament is nearly unambiguous.  (The
"nearly" is due to a technical issue that will arise only rarely.)
 
                                         John Whelan, Cornell '91
                                               <[log in to unmask]>
                      <http://www.cc.utah.edu/~jtw16960/joe.html>
 
        Learn about the NCAA selection process on the web at
       http://www.slack.net/~whelan/cgi-bin/tbrw.cgi?pairwise
 
HOCKEY-L is for discussion of college ice hockey;  send information to
[log in to unmask], The College Hockey Information List.