First Gothello Machine World Championship Tournament: Report

The tournament actually began around 7:00 PM, due to setup time. Attending were students Andrew Southworth and Paul Mirati (authors of Luser) and Keith Miller (whose entry did not participate). Hosting were Bart Massey (myself) and Jamey Sharp, Intel Linux Laboratory Administrator. Jamey's assistance both with preparing and running the tournament and preparing this writeup was invaluable, and is gratefully acknowledged.

The Contestants

Listed are all entrants in the order in which they were seeded by the tournament director based on pre-tournament program evaluations.

SeedAuthorsProgram
1southworth-miratiLuser
2burkeThunderhammer
3longunshallow
4supancichgothello
5parkerplaygoth
6zuttermeistergthDz
7neiroukhGigathello
8tanakajunya
9wang-liwang-li-Grossthello
10linChen_Llo
11tangtang-Grossthello
12huanghuang-Grossthello

The Tournament

The tournament consisted of 6 rounds, in an order and format chosen by the tournament director. Each program had an un-loaded Linux box to itself every round, so there was no opportunity for resource contention.

Round 1: Competitors vs. Grossthello

The first two rounds were played against Grossthello at depth 3, with a time limit of 2 minutes per side. All of the Grossthello instances ran on the same machine. Programs which won both games in the first round were given a bye in the second round. Programs which failed to win against Grossthello in any of the first four games were eliminated from the tournament. After both rounds the programs were ranked in the order of winning record, with ties broken by tournament seeding: this is the order shown in the table.

Round 1Round 2
AuthorsProgrambwbw
burkeThunderhammerww--
longunshallowww--
parkerplaygothww--
tanakajunyawlww
huanghuang-grossthellowlww
zuttermeistergthDzl(t)wtw
supancichgothellolwl(i)l
wang-liwang-li-grossthellowlwl
neiroakhgigathellowlwl
southworth-miratiLuserllwl
linchen_Llollll
tangtang-grossthellollll

The game annotated (t) was lost on time. The game annotated (i) was lost on an illegal move in the won position.

Round 3: Power Protect

The remaining 10 programs not eliminated in the first round were paired in power-protect fashion: the highest-ranked program against the lowest, and so forth. A round was played with a time limit of 4 minutes per side, and programs losing both games were eliminated. In the tables, a heading of b indicates that the first-listed program played black, w that the first-listed program played white.

Authorsbw
burkeww
supancichll
Authorsbw
longww
southworth-miratill
Authorsbw
parkerlt
wang-liwt
Authorsbw
tanakall
neiroukhww
Authorsbw
huanglw
zuttermeisterwl

Round 4: Power Protect

Of the 7 programs remaining after round 3, Thunderhammer was given a bye. The remaining six teams were ordered roughly power-protect (but without duplicating previous matchups), and a round was played with a time limit of 4 minutes per side. Programs losing both games were eliminated.

Authorsbw
longww
parkerll
Authorsbw
neiroukhww
zuttermeisterl(t)l
Authorsbw
huangww
wang-lill

The gthDz loss on time this round was in a position where the program could have tied the game by filling an "extra" eye after an opponent pass. This program had a lot of trouble with opponent passes.

The win by the "protected" side in every case here suggests that the ranking converged correctly.

Round 6: Semifinal

The semifinal was played with a time limit of 4 minutes per side. Programs losing twice were eliminated.

Authorsbw
burkeww
huangll
Authorsbw
longll
neiroukhww

A minor upset here was Gigathello over Unshallow, a match that really showed the effectiveness of Gigathello.

Round 6: Final

The final was played at 8 minutes per side.

Authorsbw
burkeww
neiroukhll

More Information

Game records are available upon request. They are difficult to read due to a formatting botch by the tournament director.


Last Modified: 2001/12/14
Bart Massey, <bart@cs.pdx.edu>