Jason Moy posts the following explanation according to the USCHO staff:
 
> Here is the best guess as to what happened.
>
> The Committee added the 10 picks that were considered no brainers.
 
[snip]
 
> Using the RPI they then determined the so called bubble teams
 
> Northern Michigan
> Quinnipiac
> Notre Dame
> RPI
> Princeton
> Ohio State
 
This would explain why Niagara was not included, but I'm still
doubtful that straight RPI is still used in the process.  We tried
asking this question last year and the best answer we could get out of
the NCAA was something to the effect of "the bubble is chosen looking
at the RPI and comparisons" which doesn't really nail it down too
much.
 
> They immediately through out Quinnipiac. And ranked the teams based on
> criteria ( a mini PWR)
>
> Northern Michigan   3
> Notre Dame          2
> Rensselaer          2
> Ohio State          2
> Princeton           1
>
> Northern then earned a birth and Princeton was thrown out. The remaining
> three teams form a loop
>
> OSU beats Notre Dame who beats Rensselaer who beats OSU
>
> They then looked at the individual comparisons.
 
[which showed that ratings percentage index was used as a tiebreaker
in both of Rensselaer's comparisons]
 
> So the comparison actually went the following way, without any ties being
> broken:
>
> OSU            1.5
> RPI              1.0
> Notre Dame  0.5
 
I don't think they would have done this, given what happened in 1997
when Vermont, Cornell and UNH were all being compared to decide who
went out West.  Quoting from
<http://www.uscollegehockey.com/tournament/032097.html>:
 
     Cornell edged Vermont in the criteria, while Vermont defeated New
     Hampshire, and UNH defeated Cornell.  So with the teams even in
     that sense, the RPI was used to break the tie, and Cornell was
     the odd team out.
 
I went back and looked at the details of those comparisons, and
UNH-Cornell was the only one for which RPI had to be used as a
tie-breaker.  If they had applied the logic you attribute to this
year's selections in that situation, the comparison would have been
 
Cornell 1.5
UVM     1.0
UNH     0.5
 
and UNH would have been sent West instead of Cornell.
 
I think it's more likely that given the need to choose two teams out
of the five-team bubble you listed, they dropped Princeton (who are
farther below the cutoff than NMU are above it) and re-evaluated,
which put NMU and OSU in with two comparisons each.
 
However, we won't know for sure until someone at the NCAA tells us
what happened.  That's why I'm waiting to post a "here's what
happened" of my own like
http://www.slack.net/~whelan/cgi-bin/tbrw.cgi?pairwise.1998
 
                                          John Whelan, Cornell '91
                                                  [log in to unmask]
                                     http://www.amurgsval.org/joe/
 
Play along at home at http://www.slack.net/~whelan/cgi-bin/tbrw.cgi?brack
 
HOCKEY-L is for discussion of college ice hockey;  send information to
[log in to unmask], The College Hockey Information List.