At 9:03 PM -0600 3/9/98, Adam Wodon wrote:
>It had been assumed that the committee went strictly by PWR  (PWR by the
>way, is a monicker given to this system by Keith Instone, Tim Brule and
>others -- I don't go back that far -- It didn't actually have a name).
>
>Since the committee never actually named the system --- other than to
>say "we go by the 'criteria'" -- they were just as confused, in reverse,
>when this whole PWR thing kept cropping up.
 
 
Tim gets credit for the "PWR" name. I get credit for the concept. I call it
"Criteria Comparison Record" since when I do it, I present it as a won-loss
record. (With a different set of criteria definitions, ie no RPI
tiebreaker, you can imagine ties, so I decided to use a won/loss/tie
representation rather than just wins that PWR goes by.)
 
Here is some of the history to explain the confusion:
 
1. The software I wrote to help the NCAA determine selections and seedings
has "CCR" (aka PWR) built into it, to be used if they want to. I was never
sure how much they used it and how much they relied on other things. The
seeding process was especially vague. But my software did lots of different
things (like let the NCAA set any RPI weightings they wanted), so it was up
to them to decide how to use the system. I learned of the selections just
like everyone else.
 
2. Reverse-engineering over a few years indicated that CCR/PWR was REAL
GOOD at predicting the results. That is probably when I presented the idea
to HOCKEY-L, but I forget. I believe I presented it as "this method gets
the same results but only those in the room know for sure". Someone would
have to check the archives.
 
3. Adam's Joe March interview was a lot of news to me, too. I only deal
with Phil Buttafuoco @ NCAA and the interview helped me understand how Phil
was using my software and what he was telling the committee members. I
believe Phil was using the concept of CCR/PWR for the most part but not all
of the way until the end. Joe's confusion about the terms and his discovery
that both "sides" are on the same page lead me to that conclusion. Also,
the NCAA did not use my software last year, so computing CCR/PWR may not
have been automated for them. Being hard to compute by hand, they have
taken a shortcut. This could explain variations in last years method, but I
have no idea if that is the case.
 
Use of CCR/PWR has never been "legislated" like the exact crieria
comparison rules, so the committee is free to use it as it sees fit, and
when it does not see fit.
 
 
 
There are many "holes" that have been tightened up over the years, like
defining "teams under consideration" by a .500+ record (for many years no
one really knew who was under consideration).
 
There are still a few more holes, places where the process is vague and
each year you have to ask the committee what they are doing this time. For
many of the finer points, like what happens when two teams have the same
CCR/PWR, we have to rely on the committee to tell us what they are doing in
those cases. My experience has been that these sort of problems are not
solved until they crop up and make a difference. Basically, the committee
solves those problems when they arise ("make it up as they go along" you
could say).
 
Again, they are a few holes left, but with the peppering of quality
questions that members of this list have hit them with, it looks like the
committee has been forced to think about many of the holes ahead of time.
 
Hopefully the committee will continue to fill the holes and someday
FORMALLY publish the entire selection and seeding process. Hockey is still
light years ahead of the other sports, so I do not expect it tomorrow, but
eventually.
 
 
Keith
PS They are back to using my software this year, so let's hope for a job
well done.
 
HOCKEY-L is for discussion of college ice hockey;  send information to
[log in to unmask], The College Hockey Information List.