[Date Prev][Date Next][Thread Prev][Thread Next]
[Date Index] [Thread Index] [Author Index]
Return to main CEDA-L Archive Page

Re: MPJ


While Steve Hunt's philosophical questions deserve careful discussion, I
would like to answer some of the technical questions.

1) How may A's would be required to successfully match the tournament?

Because of the power of the backtracking algorithms, I predict that with
30-35% of the pool as A's, approx. 90-95% of the matches would be mutual
A's with less than 1% of the matches being C's.

I am amazed at the ability of backtracking to maximize preferences.  I
gravely predicted before this nats that some judges would not hear rounds
(since two critics were struck by over 160 teams and 21 were struck by
over 1/2 of the teams).  However, every 8-round judge but one judged at
least 7 rounds - the other judged 6 AND every partial judge met their full
obligation.  One critic who could only hear 18 teams in the tournament
when strikes and regional constraints were combined, judged all rounds for
which they were obligated.

2) Would guest judges and potentially non-preferred critics be assigned
rounds?

The assignment of "less-preferred" critics to rounds is almost entirely a
function of how tight the pool is.  As an illustration, at SIU we had
approx 6 slots of excess judging in the pool which meant that
hard-to-place critics could be passed over in favor or more highly
preferred critics.  As a result, EVERY prelim round had a mutual A BUT
some critics judged fewer than their full complement of rounds.  At Towson
State, Ken operated with a "fixed" pool - I pre-assigned partial
commitment judges to the rounds they would be available for and used only
ONE more judge per round than was required.  As a result, several rounds
in novice had mutual B's but everyone judged their commitment (actually I
reran the tournament later with a new feature that allows bactracking
across divisions and eliminated almost ALL novice B's.  The non-usage of
less-preferred critics is NOT an inherent feature of mutual preference but
rather the number of additional critics in the pool.  CEDA could manage
this to insure that all judges met their commitment.

3) Could the tabroom handle the huge volume of data entry that is
required?

This year, we entered strikes until nearly 4:00 AM. (200 teams x ~30
strikes = 6000 data points with doublechecks on all).  Despite the huge
data entry task, we again did not have any team successfully demonstrate
that they were assigned a judge that they had struck.  If mutual pref were
used, we would have approx 30000 data points, a daunting task.  BUT, we
have made the commitment to never let data entry serve as a limit on
tournament operation. I am in the process of designing a SCANTRON option
which would permit automated entry and validation of pref sheets (it
actually takes quite a bit of time to verify that sheets are filled out
correctly).  Our strategy would be to read pref sheets into the computer
and then print out the computer file.  The team that submitted the sheet
would have the responsibil;ity for verifying accuracy of the computer file
by comparing it to their original.  Such a system would actually reduce
tabroom data entry and reduce the possibility of error.

Even if such a procedure proved impossible, we would use muliple data
entry teams to accomplish the task - this is not our first choice however.

In any case, MPJ must be adopted on merits or rejected for their lack.
The technical consisderations will be accounted for regardless of how we
manage the tournament.

Sincerely,

GARY
References:
Archive created by Jonathan Stanton (jonathan@cs.jhu.edu)
Return to main CEDA-L Archive Page