RESULTS: MITBAT III (3-3-07) @ MIT (Cambridge, MA)

[Concluding Unscientific Post-It Note]

Your team is invited to participate in the Massachusetts Institute of Technology Beaver Academic Tournament III (MITBAT III) that will take place on Saturday, March 3, 2007. Registration will be from 8:00 am to 8:45 am with rounds starting at 9:15 am. Running smoothly, we estimate the tournament finals to end between 6:00 and 6:30 pm. Preliminary rounds will end mid-afternoon.

Following the success of the last two years, the tournament will again feature Real Time Swiss Pairing. That is, the rounds will be paired based on the performance of your team thus far through the tournament -- including the round you just played -- without any delays. We have tried and proved we do in fact run at the pace of the slowest round. Once the final game of the round finishes, the pairings can be made immediately! I don’t think you will have the opportunity to experience this elsewhere, so come and see it in action!

Teams will be accepted on a first-come, first-serve basis, so please register as soon as possible. The current cap of teams is around 40 [though this may increase if we can get more space]. The registration deadline is Friday, February 17, 2006. If you register by the early registration deadline, January 27, 2006, you will receive a $10 discount from your school’s total fee. For our convenience and your benefit, we request that you register as soon as you can.

I want to expand this tournament to make it as big as possible, and the only way I can do that is which your help. Please consider coming if you can! And especially for teams who have longer trips to make, I would like to keep a list of seriously interested teams so that every one can see every one else who is thinking about coming and I’m hoping that will help convince everyone that it is worthwhile to come! Our website will feature up to do lists showing registered and seriously interested teams. That said, please don’t everyone hold off -- then no one will come!

This tournament is open to teams from all geographic areas and levels of experience. The questions are from NAQT, and the top 15% of the field qualifies for NAQT 2006 High School National Championship Tournament http://www.naqt.com/hsnct/index.html. Additionally, the top four teams qualify for the PACE National Scholastics Championship in June 2006.
MITBAT III will be held at the Massachusetts Institute of Technology campus. This is a great opportunity to bring your aspiring students to visit the campuses of MIT, Harvard, and other universities as well as to experience the wonderful city of Boston!

We will be using an untimed NAQT high school format [see http://www.naqt.com for details]. Every round will have twenty tossups with unrelated bonuses. There will be a single elimination playoff tree for teams that break. Depending on the number of teams registered, there may possibly be a consolation playoff as well. All teams will be guaranteed to play at least 5 games, but very likely will get 6 or 7, and at least 16 teams will advance to the first stage of playoffs.

In order to register, please email or mail a copy of the registration form at the end of this letter. The registration fee is $80 for the first team from a school and $55 for each additional team, with a $10 discount from the school’s total fee for registering by the early registration date. There is no limit on the number of teams that maybe registered from any one school. There is a $10 discount from a school's total fee for each working buzzer system brought. The questions are available for purchase: $10 for teams participating in the tournament and $15 for non-participating teams. Registration fees may be mailed in advance or paid at the tournament. Checks should be made out to "MIT." Note: the registration fee for new teams that have never competed in any quiz bowl tournament before is $50 for the first team and $45 for subsequent teams from the same
school.

My e-mail address, [email protected], <mailto:[email protected]> can be used for any questions that you may have regarding the tournament or visit http://web.mit.edu/collegebowl/www/mitbat/. I will be happy to help anyone with information about travel, living arrangements, or anything else as best I can. Please keep in mind Boston has an excellent and safe subway and commuter rail system that may help you minimize travel and accommodation cost. Shortly before the tournament, we will contact registered teams to confirm participation. We look forward to a large and
competitive tournament and hope to see you on March 3, 2007! Please contact us with any questions.

Sincerely,

Mark Seifter

Tournament Director, MITBAT III

EDIT: We will be using IS-64.

[Concluding Unscientific Post-It Note]

Field as of 2/4/07:

Registered

* Aiken High (1)
* Manlius Pebble Hill School (1)
* E.O. Smith (3)
* Georgetown Day School (1)
* Norwich Free Academy (2)
* Stuyvesant (2)
* Smithtown (1)
* Dolgeville Central School (1)

Total teams registered: 12

Not registered, but have been in contact

* New Milford High School (2)
* Concord-Carlisle (2)
* Coxsackie-Athens (3)

This list is also on our website, http://web.mit.edu/collegebowl/www/mitbat/

[Aerion]

The final standings for yesterday's MITBAT III are:

1. Aiken High School
2. E. O. Smith A
3. Manlius Pebble Hill
4. E. O. Smith B

Top Scorers:

1. Amy Varallo, Aiken, 73.33 PPG
2. Maxwell "The Hammer" Ziskin, Williston Northampton, 56.00 PPG
3. Gavin Byrnes, Manlius Pebble, 55.83 PPG

Congratulations! Complete results and stats are on our website, http://web.mit.edu/collegebowl/www/mitbat/.

We are planning to hold MITBAT IV on Saturday, March 1, 2008.

[fender_outta_hock]

As a member of E. O. Smith A I would like to thank MIT for a job well done in running this tournament. It was a great deal of fun to play in, and very well moderated. A very good tournament.

[Zip Zap Rap Pants]

Haha that's pretty funny two tournaments on the same day ended up with 23 teams and tried to get all teams to play extra games (play in games for MIT consolation games for us). Did this tournament actually run on the debugged Swiss system, or was it done with SQBS, or did they just go with the old incorrect system?


Celebrating Jamestown's 400th anniversary with my 400th post.

[Aerion]

Was the old system incorrect? (It stopped working sometimes, but I thought it generated valid pairings.) I don't have any idea what's up with the old pairer, since Lyric wouldn't give us the code on account of the fact that it was a software engineering war crime. This year, we used a new pairer that I wrote. It seemed to work great. I'll probably be putting source code somewhere, sometime, if anyone is interested.

SQBS doesn't do a very good job of avoiding repeats, and I don't think it avoids pairing teams from the same school against each other. I actually just tried to test it a little, but it crashed on a stack overflow of some kind. (Probably wine's fault, and not SQBS's.)

Our pairer is willing to float teams as far up or down as is needed to ensure that there are no repeats and no games between teams from the same school. It worked like a charm yesterday - the only delay during the initial rounds was due to a rules protest.

[mhanna]

Thanks MIT for a job well done. My group really enjoyed the first-class moderators and the interaction with the teams.

Mac Hanna
Aiken

[rleavitt]

A great experience-- thanks for organizing! The moderators were friendly and swiss pairing gave us some even competition. Thanks for posting the stats so quickly!

[Zip Zap Rap Pants]

Was the old system incorrect? (It stopped working sometimes, but I thought it generated valid pairings.) I don't have any idea what's up with the old pairer, since Lyric wouldn't give us the code on account of the fact that it was a software engineering war crime. This year, we used a new pairer that I wrote. It seemed to work great. I'll probably be putting source code somewhere, sometime, if anyone is interested.

SQBS doesn't do a very good job of avoiding repeats, and I don't think it avoids pairing teams from the same school against each other. I actually just tried to test it a little, but it crashed on a stack overflow of some kind. (Probably wine's fault, and not SQBS's.)

Our pairer is willing to float teams as far up or down as is needed to ensure that there are no repeats and no games between teams from the same school. It worked like a charm yesterday - the only delay during the initial rounds was due to a rules protest.

Well I know one problem was the old system paired good teams against each other too early, for instance I think RM had to play TJ in round 2 at the GSAC that used it.

[Aerion]

Well I know one problem was the old system paired good teams against each other too early, for instance I think RM had to play TJ in round 2 at the GSAC that used it.

It's tough to avoid that issue if you don't rank the teams beforehand. I had no clue about the relative strengths of our teams, other than that Aiken was flying in from SC, so they were probably pretty good. (That turned out to be correct.) As a result, I assigned pairing numbers at random and paired the first round accordingly (with a couple exceptions). Normally, if your teams come in ranked, as is done in chess-style Swiss sys, pairings within the same group are based on the pre-tournament ranking. Instead, I opted to rank teams within each score group based on their point margin so far in the tournament. This tended to keep the high-scoring teams away from each other for the first few rounds.

I don't think it's too big of a problem if good teams face each other early, although it does sort of kill the climax at the end. Since the top teams still get high seeds in the single-elim portion, it doesn't seem like a real issue.

Also, our final seeding tiebreak was to guess what animal I was thinking of. The tiebreak didn't come into play, but for the curious, the answer was "emperor penguin." (Not a dragon. Sorry.)

--Jason ("the guy with the dragon on his shirt")

[Zip Zap Rap Pants]

Well I know one problem was the old system paired good teams against each other too early, for instance I think RM had to play TJ in round 2 at the GSAC that used it.

It's tough to avoid that issue if you don't rank the teams beforehand. I had no clue about the relative strengths of our teams, other than that Aiken was flying in from SC, so they were probably pretty good. (That turned out to be correct.) As a result, I assigned pairing numbers at random and paired the first round accordingly (with a couple exceptions). Normally, if your teams come in ranked, as is done in chess-style Swiss sys, pairings within the same group are based on the pre-tournament ranking. Instead, I opted to rank teams within each score group based on their point margin so far in the tournament. This tended to keep the high-scoring teams away from each other for the first few rounds.

I don't think it's too big of a problem if good teams face each other early, although it does sort of kill the climax at the end. Since the top teams still get high seeds in the single-elim portion, it doesn't seem like a real issue.

Also, our final seeding tiebreak was to guess what animal I was thinking of. The tiebreak didn't come into play, but for the curious, the answer was "emperor penguin." (Not a dragon. Sorry.)

--Jason ("the guy with the dragon on his shirt")

It actually has nothing to do with pre-ranking. Early on, let's say you have 8 2-0 teams by round three. At that point I think you're supposed to have the top undefeated team in ppg play the lowest 2-0 in ppg, like a seed basically, or something like that. Evan Silberman can explain it better than I can.

[Aerion]

It actually has nothing to do with pre-ranking. Early on, let's say you have 8 2-0 teams by round three. At that point I think you're supposed to have the top undefeated team in ppg play the lowest 2-0 in ppg, like a seed basically, or something like that. Evan Silberman can explain it better than I can.

No, at that point, you typically have the top 2-0 team play the 5th 2-0 team, 2 v. 6, 3 v. 7, and 4 v. 8. If one of those pairings is illegal for some reason, you attempt to exchange the lowest-ranked team with the lowest-ranked team possible in order to achieve a legal pairing, etc.

The pre-ranking serves to define what we mean by "top" team. In most chess tournaments, the players keep a consistent ranking throughout the tournament (their "pairing number"), and are always ranked in the same way regardless of their performance during the tournament. Of course, in chess, there are no PPG or other stats to speak of, just wins, losses and draws.

In any case, if you correctly rank the teams beforehand and pair by that rank, you are likely avoid having the top 2 teams play each other until there are only 2 or 3 undefeated teams left. If you rank on the fly with PPG and pair according to that stat, you are likely to get a similar result, although with a little more variance (if somebody is having a particularly good or bad round). The problem comes if you pair teams within a score group at random, or according to a poorly-chosen pre-assigned rank, which I suspect is what Lyric was doing (but again, without the source, I can't be sure).

[dschafer]

The program used at the GSAC, as I recall, paired 1 vs. 2, 3 vs. 4, etc. within each group.

The JIAT ranks the teams mostly by random for the first round, then re-ranks within each group by PPG. After that re-ranking, 1 plays n/2, 2 plays n/2 + 1, etc, with rearrangements made if necessary to avoid repeats and/or teams from the same school playing each other.

[Zip Zap Rap Pants]

The program used at the GSAC, as I recall, paired 1 vs. 2, 3 vs. 4, etc. within each group.

The JIAT ranks the teams mostly by random for the first round, then re-ranks within each group by PPG. After that re-ranking, 1 plays n/2, 2 plays n/2 + 1, etc, with rearrangements made if necessary to avoid repeats and/or teams from the same school playing each other.

Yeah that's what I was talking about. At GSAC in round 2 the straight up top two 1-0 teams (in ppg) played each other in round 2, which was whack (and it was the main problem with the old system, which I believe was the same one used in the first two MITBATs).

[Aerion]

Yeah that's what I was talking about. At GSAC in round 2 the straight up top two 1-0 teams (in ppg) played each other in round 2, which was whack (and it was the main problem with the old system, which I believe was the same one used in the first two MITBATs).

Ewww. Now I'm really glad I rewrote the pairer this year.

ETA: Checking the stats from MITBAT I reveals that yes, the top two scorers from Round 1 were in fact paired against each other, as were #3 and #4. Ick. Anyway, the current MITBAT system follows the FIDE Swiss pairing rules pretty closely, except for the ranking of teams within score groups. We also, obviously, ignore color choice.