CSc 180 -- Spring 2013
Intelligent Systems


2013 Tournament Results - "Borgo Pass"



The 11th Annual AI Strategy Game COntest took place April 9-11 2013.
The atmosphere was very lively, and we had many spectators in the form
of students, faculty, alumni, and previous champions and contestants.
The game this year was "Borgo Pass", which is a sort of reduced chess
with a "vampire" theme.  An interesting feature is that each player's
"vampire" (analogous to the "king" in chess) can capture its own pieces -
and the game is structured so that is a relatively common occurrance.

I would like to thank DeWayne Alexander for helping to run the event,
Francis Tanada for helping with the scorekeeping, and Chris Liberatore
and Mike Vollmer for helping to proctor some of the programs.  They all
helped make the event run very smoothly, and we even finished on time!

As always, each match consisted of two games, so programs could lose
individual games along the way and still win by amassing more total
victories.  There were many VERY strong entrants, as evidenced by the
meager 5th place finish by the instructor's program "Orlock".  As the
event drew to a close, it became a two-horse-race between Matt Rojas'
"Ramathorne 3" and Nick Pearson's "Honest Abe".  Both programs used
bitboards for highly efficient move generation.  "Ramathorne 3" took
the extra step of trying to predict the opponents' moves to get a jump
on the tree search.  But in the end, it was a surprise bug in "Honest Abe"
that made the difference - it didn't rear it's head until all the way
into the finals, when "Honest Abe" played an illegal move, costing it
a tie for 1st place.

Tied for 3rd and 4th were Kyle Matz' "Adamas" and Manuel Sabin's
"Van Helsing", both of whom were 2-0 against "Orlock", but 0-2 against
the top two programs.  After "Orlock", in 6th and 7th were Tate
Chamberlain's "NOS-4-A2" (which benefitted from automatically evolving
its evaluation function) and Donald Todd's "Wesley Snipes", which
featured a sophisticated evaluation function.

Many of the programs improved substantially between day 1 and day 2.
Even the winner "Ramathorne 3" had lost an early round to the 10th place
finishing "Talzin" (Kenneth Barnett), but came back with numerous
improvements.  Programs "Larper" (Ewert) and "Luna" (Tovashova) were
unable to perform at all on the first day, but notched some strong wins
the second day.  Most notably, Justin Robbins' "Tiddlywinks" had numerous
bugs on day 1, but won 8 straight games on day 2, finishing well into the
top half and securing the coveted "Grand Hamster" award for the most
improved program of the tournament.

"Ramathorne 3" was written in C++, and utilized minimax with alpha-beta
pruning, iterative deepening, bitboards, history tables, and move prediction
to search at least 7 plies at each move.  Given that "Borgo Pass" has a
branching factor approaching that of chess - around 30 - "Ramathorne 3" is
thus able to analyze on the order of two billion moves within the 5-second
time limit - a remarkable achievement for a 3-week undergraduate assignment.

Complete results are shown below, including results from prelims,
eighths, quarters, semis, finals, and all consolation rounds
(finals results shown are combined with already-played semis):           Total
                                                                     -----------
 1. Ramathorne3  (Rojas)...... f:7-1 s:3-1 q:4-0 e:3-1 p:4-0              18-2
 2. Honest Abe   (Pearson).... f:6-2 s:3-1 q:4-0 e:4-0 p:2-0              16-2
 3. Adamas       (Matz)....... f:3-5 s:5-1 q:3-1 e:4-0 p:4-0              18-6
    Van Helsing  (Sabin)......(f:3-5)s:0-4 q:3-1 e:4-0 p:4-0              13-5
 5. Orlock       (Gordon)..... f:1-7 s:4-2                                 5-7
 6. NOS-4-A2     (Chamberlain)...... s:2-2 q:2-2 e:3-1 p:4-0              12-6
 7. WesleySnipes (Todd)............. s:0-6 q:2-2 e:2-0 p:4-0               8-8
 8. Ashen        (Yu)..................... q:0-4 e:1-3 p:4-0 c:3-1         8-8
 9. LennieSmall  (Kramer)................. q:0-4 e:3-1 p:0-2 c:2-2         5-9
10. Talzin       (Barnett)................ q:0-4 e:3-1 p:2-0 c:1-3         6-8
11. Tiddlywinks  (Robbins)...................... e:1-3 p:1-3 c:4-0 c:4-0  10-6 [GH]
12. Dejarik      (Devereaux).................... e:0-4 p:4-0 c:1-3         5-7
    PierreUnwise (Swartz)....................... e:0-4 p:2-2 c:1-3 c:2-0   5-9
14. ParaCUDA     (Cuffe)........................ e:2-2 p:2-2 c:3-1         7-5
15. TrnslMadness (Alsaadi)...................... e:1-1 p:2-2 c:3-1 c:1-1   7-5
    VampireChess (Linnyk)....................... e:--- p:0-4 c:2-0 c:3-3   5-7
    EricNorthman (McParland).................... e:1-3 p:0-2 c:2-2 c:0-2   3-9
18. Blade        (Creswell)..................... e:0-4 p:1-3 c:3-1 c:1-1   5-9
    StakeDinner  (Brown)........................ e:2-2 p:2-2 c:1-3 c:0-2   5-9
21. LaptopCrashd (Dong)......................... e:1-3 p:0-4 c:0-2 c:2-0   3-9
    Shiggy       (Hansen)....................... e:--- p:0-4 c:2-2         2-6
    Blast Zone   (Gatt)......................... e:--- p:1-3 c:0-2         1-5
24. Luna         (Tovashova).................... e:--- p:--- c:2-2 c:0-2   2-4
25. Glitter      (Kipers)....................... e:--- p:1-3 c:1-3         2-6
    Aesop        (Anderson)..................... e:--- p:0-4 c:1-3         1-7
27. Penumbra     (Meyers)....................... e:--- p:1-3 c:0-2         1-5
28. Eradicator   (Lopez)........................ e:--- p:0-4               0-4

As always, the top programs played very strongly throughout, certainly
stronger than their human handlers.  It is doubtful that any of us authors
would place in the top 10 against such a powerful field, were we to try
and match wits with our programs.  Although none of the authors incoporated
a GUI into their programs, Chamberlain's "NOS-4-A2" featured clever use of
color in it's display to make the game easier to follow.

There were a wide variety of opinions regarding which heuristics work the
best for this game.  "Ramathorne 3" and several others used what appeared
to be an effective strategy of sacrificing pieces to lure the opponent's
"vampire" out into the open, and then attacking it from a distance.  As a
result, most of the games were wild, tactical affairs where it was
difficult to tell which side was ahead.  It also is unclear whether it is
advantageous to move first or second, although it appeared that both
Rojas and Pearson's programs won more quickly when moving first.

"Borgo Pass" was a great game and it is still far from being solved.

    

2012 results