1st Derek Oldbury Memorial Man v Machine

The 1st Derek Oldbury Memorial 'Man v Machine' Open
"Annotated games from the tournament"

     This was a 7 round freestyle quick-play tournament held at Bristol on September 10th 1994 organized by Richard Pask. 21 players took part including four computer programs, Checkers (Gil Dodgen), Colossus (Martin Bryant), Hustler (Derek Oldbury) and Sage (Adrian Millett).
     The results were as follows, 1st Colossus 13/14 pts, 2nd Checkers 11/14, 3rd Sage 10/14, 4th Tom Landry 9/14. The computers dominated the event as was to be expected as both Pat McCarthy and Bill Edwards were unable to attend and Richard Pask was operating Derek's Hustler program.
     The two pre-tournament favorites were therefore Checkers and Colossus which were drawn together in the very first round! Checkers played a published loss on move 10 but Colossus later missed the win, allowing Checkers to scrape a draw by Petterson's drawbridge.
     Colossus then went on to win every game (including a real demolition job on Sage in the 5th round), but Checkers allowed two more draws (one against 15 year old Danny Martin). Sage started well with 4 straight wins but then after losing to Colossus in round 5 also went down to Tom Landry in round 6.
     Thus Colossus ran out the winner by two clear points.
This is even more impressive considering that Colossus was running on a mere 33MHz 486DX, whereas Checkers was on a 90MHz Pentium(!!) and Sage was on a 66MHz 486DX2.
     The tournament was sponsored by the BDF and the EDA and also contributions from the computer programmers. It is hoped that the tournament will become a regular fixture.

Round 1
Black:Colossus White:Checkers

11-15 23-19 8-11 22-17 9-13(a) 17-14 10x17 19x10!?(b) 7x14 25-22 4-8 27-23 11-16 29-25 2-7 22-18 6-10 18x9 5x14 24-20?(c) 16-19 23x16 12x19 26-22 17x26 30x16 14-18 31-26 13-17 21x14 10x17 16-11 8x15 25-21 18-23 26x10 7x14 20-16 17-22 28-24 22-26 24-19 14-18?!(d) 19-15 26-30?(e) 32-28?(f) 30-26?(g) 15-10 18-23 16-11 26-22 11-7 23-27 7-2 27-32 2-6?(h) 32-27 6-2 27-23?(i) 28-24 23-27 24-20 27-23 2-7 22-18 7-11 23-27 11-7 27-31 7-2 31-26 2-7 26-22 7-2 18-23 2-7 23-27 7-2 27-24 2-7 22-18 Draw

(a) Colossus loves the tactical complexity of the Laird and Lady (Mixed) opening, always ready for a fight.
(b) And Checkers does not disappoint by choosing the Black Doctor! Derek would have approved!!!
(c) All sound to here but now Checkers falters. This given to lose in BDP & 25-22, OMTD 23-19.
(d) 26-31 19-15 1-6 wins outright avoiding the trickier bridge ending.
(e) The remaining play, by two of the strongest computer programs in the world, shows the tremendous difficulties they still have with these 4x4 endings. I checked the play with Chinook's 8-piece databases and it states that this allows a draw whereas 26-31or 18-23 still win.
(f) But Checkers gives the win back! 15-10 draws.
(g) And Colossus says "I don't want it!". 1-6* wins.
(h) Checkers once again falters. 28-24* is forced to draw.
(i) "no... really... you keep the draw...". 22-18 wins.

Round 5
Black:Sage White:Colossus

9-14 22-17 11-15 23-19 7-11(a) 25-22 5-9(b) 17-13 11-16 26-23 16-20 30-26 2-7?!(c) 21-17 14x21 29-25 21x30 19-16 12x19 23x16 30x23 27x2 20x27 32x23 8-12?(d) 23-19 10-15 19x10 6x15 13x6 12x19 22-17 1x10 2-7 10-14 17x10 19-24 28x19 15x24(e) 7-11 24-28 10-6 28-32 6-1 3-8 11-16 8-12 16-11 White Win

(a) I was disappointed to see Sage choose the Whilter as it always seems to me to be too drawish.
(b) pp gives 11-16 first but this move order seems perfectly sound.
(c) But this is very weak! When Colossus started pitching the pieces after this move I initially wondered what the hell was going on. It can be very unnerving operating your own program!
(d) And this loses. BDP correctly gives 3-7 as Black's only sound draw.
(e) Now into Colossus' new 6-piece endgame databases. Richard Pask was watching as this endgame evolved and started discussing with another player if it might be a win. I told him Colossus now knew it was so. He was visibly impressed.

     Many people on both sides of the Atlantic have asked me when I will be releasing the latest version of Colossus to the general public as it is now clearly the strongest PC program available. Well the new release will be called Colossus '94 and will hopefully be available sometime in October. What most people don't know is that I have also been working on the Chinook program with Dr. Schaeffer in preparation for its aborted match with Tinsley in Boston. It was this 'distraction' from my own program which caused a delay in its being launched earlier.
     However the delay will be well worthwhile for everyone. Colossus' book database (and now Chinook's too!) contains nearly 84,000 moves! This entire database has been 'verified' by Chinook. Where errors in pp were found they have been corrected and also all of Chinook's
own cooks have been added to the book!! In return for the book database, Chinook donated its 6-piece endgame databases to Colossus, some 2.5 billion positions online!  Needless to say the new Colossus is formidable! Watch this space for announcement on pricing/system-requirements etc in the near future.


2nd Derek Oldbury Memorial 'Man v Machine'

Derek E. Oldbury - 7/9/1994