Author: Kurt Utzinger
Date: 06:46:24 03/17/02
I hope my English is good enough to follow my thoughts:
Tournament T2002 by chessfriend Rolf Bühler, Zurich
Comments by Kurt Utzinger, Wetzikon / 17th March 2002
***************100 rounds / 6 engines*************
3 x Athlon 1.3/256
Engine-Matches [normally=1 round per day]
Fritz6-GUI
64 MB hashtables
ponder=off
4 men TBS
Each program with its own opening book
TheKing312X with Nz7book.ctg
Book Options: Use Book, Tournament book, Minimum games=2, Optimize
40/120' + 20/60' + 30' for rounds 1-10 [normal book options]
40/80' + 20/40' + 30' for rounds 11-25
60/120' + 30/60' + 30' for rounds 26-100
Cross table 1
<pre>
Program Score % Av.Op. Elo + -
Draws
1 TheKing312X : 58.5/100 58.5 2390 2450 62 51
39.0 %
2 Fritz7002 : 56.5/100 56.5 2392 2438 64 47
43.0 %
3 Shredder6 : 48.0/100 48.0 2402 2388 47 69
40.0 %
4 Junior7 : 48.0/100 48.0 2402 2388 49 69
36.0 %
5 ChessTiger14 : 47.5/100 47.5 2403 2385 46 68
41.0 %
6 Hiarcs732 : 41.5/100 41.5 2410 2350 55 62
33.0 %</pre>
Individual statistic of the single programs:
<pre>
(1) TheKing 3.12 X : 100 (+ 39,= 39,- 22), 58.5 %
Fritz 7002 : 20 (+ 6,= 9,- 5), 52.5 %
Shredder 6 : 20 (+ 9,= 7,- 4), 62.5 %
Junior 7 : 20 (+ 10,= 8,- 2), 70.0 %
Chess Tiger 14.0 : 20 (+ 7,= 7,- 6), 52.5 %
Hiarcs 7.32 : 20 (+ 7,= 8,- 5), 55.0 %
(2) Fritz 7002 : 100 (+ 35,= 43,- 22), 56.5 %
Shredder 6 : 20 (+ 5,= 12,- 3), 55.0 %
TheKing 3.12 X : 20 (+ 5,= 9,- 6), 47.5 %
Junior 7 : 20 (+ 6,= 8,- 6), 50.0 %
Chess Tiger 14.0 : 20 (+ 9,= 5,- 6), 57.5 %
Hiarcs 7.32 : 20 (+ 10,= 9,- 1), 72.5 %
(3) Shredder 6 : 100 (+ 28,= 40,- 32), 48.0 %
Fritz 7002 : 20 (+ 3,= 12,- 5), 45.0 %
TheKing 3.12 X : 20 (+ 4,= 7,- 9), 37.5 %
Junior 7 : 20 (+ 7,= 5,- 8), 47.5 %
Chess Tiger 14.0 : 20 (+ 5,= 11,- 4), 52.5 %
Hiarcs 7.32 : 20 (+ 9,= 5,- 6), 57.5 %
(4) Junior 7 : 100 (+ 30,= 36,- 34), 48.0 %
Fritz 7002 : 20 (+ 6,= 8,- 6), 50.0 %
Shredder 6 : 20 (+ 8,= 5,- 7), 52.5 %
TheKing 3.12 X : 20 (+ 2,= 8,- 10), 30.0 %
Chess Tiger 14.0 : 20 (+ 6,= 11,- 3), 57.5 %
Hiarcs 7.32 : 20 (+ 8,= 4,- 8), 50.0 %
(5) Chess Tiger 14.0 : 100 (+ 27,= 41,- 32), 47.5 %
Fritz 7002 : 20 (+ 6,= 5,- 9), 42.5 %
Shredder 6 : 20 (+ 4,= 11,- 5), 47.5 %
TheKing 3.12 X : 20 (+ 6,= 7,- 7), 47.5 %
Junior 7 : 20 (+ 3,= 11,- 6), 42.5 %
Hiarcs 7.32 : 20 (+ 8,= 7,- 5), 57.5 %
(6) Hiarcs 7.32 : 100 (+ 25,= 33,- 42), 41.5 %
Fritz 7002 : 20 (+ 1,= 9,- 10), 27.5 %
Shredder 6 : 20 (+ 6,= 5,- 9), 42.5 %
TheKing 3.12 X : 20 (+ 5,= 8,- 7), 45.0 %
Junior 7 : 20 (+ 8,= 4,- 8), 50.0 %
Chess Tiger 14.0 : 20 (+ 5,= 7,- 8), 42.5 %</pre>
--------------------------------------------------------------------------------------------
**********COMMENTS**********
1. Introduction
When Rolf Bühler and myself decided to organize a great tournament of the best
available chess programs running in engine matches under the Fritz-GUI and to
make this tournament public, I did not know how much work this would involve for
me. For quite some time we argued about the modus and the participants and then
decided to play with a small number of programs over 50-100 rounds. And I think
we did the right thing because such a tournament says in our opinion much more
than an event with 20-30 programs to play 2 or 3 rounds. At first we planned to
play at time control 120'/40 but soon realized [after 10 rounds] that it would
not be possible to play one round each at a normal working day. And so we
changed the time control to 40/80'+20/40'+ 30' for rounds 11-25 and then to
60/120'+30/60'+ 30' for rounds 26-100. Thus, the programs had about 2 minutes
per move at their disposal which are good conditions considering the hardware,
three Ath 1.2/256.
--------------------------------------------------------------------------------------------
2. TheKing312X
That we choose TheKing312X to participate in our tournament was for two reasons.
First of all we thought it would be interesting to see how strong the program of
Johan de Koning would be with a good opening book [Nz7book.ctg] in comparison
with the other top engines. And furthermore Rolf Bühler and myself have tested
numerous CM8-settings in more than 4500 games at 40'/40 over the last 18 months.
And so we started with TheKing312X giving this engine a CM8-setting that has
proved to be very well. And you may be surprised, this setting is not mine, but
the original CM_8777-setting of Stefan Kleinert, Germany. There are different
versions of this setting on the market and to avoid any confusion, we give below
the exact details of the setting used in our T2002 tournament.
2.1 Settings für TheKing312X unter CM8-GUI [cm_8777.cmp]
<pre>A/D (Attack/Defender) [ -2] M/P (Material/Position) [ 10]
S/P (Strength of Play) [ 100] C/C (Control of Center) [ 105]
Rand (Randomness) [ 0] MOB (Mobility) [ 115]
MSD (Max Search Depth) [ 99] K/S (King Safety) [ 150]
SS (Selective Search) [ 12] P/P (Passed Pawns) [ 110]
C/D (Contempt for Draw) [ 0] P/W (Pawn Weakness) [ 125]</pre>
<pre>OWN OPPONENT
Queen [11.9] Queen [11.9]
Rook [ 6.5] Rook [ 6.5]
Bishop [ 4.1] Bishop [ 4.1]
Knight [ 3.9] Knight [ 3.9]
Pawn [ 1.1] Pawn [ 1.1]
Transposition Table [64 MB] Ponder Off</pre>
2.2 Settings für TheKing312X unter Fritz-GUI [InBetween.ini]
[Client2Server]
; TheKing 3.12 X or CM 8777 (Kleinert Stefan)
xboard := xboard\npost\nnew\ncm_parm tts=67108864 ttu=8\ncm_parm opk=599585
opp=103 opn=121 opb=127 opr=121 opq=123 myp=103 myn=121 myb=127 myr=121 myq=123
cc=112 mob=123 ks=160 pp=117 pw=133 cfd=0 sop=100 avd=-2 rnd=0 sel=12 md=99\n
[Server2Client]
; Set transalation of commands from the server here
; format word := translation
2.3 Special remarks regarding TheKing312X settings
It must be emphasized that in all our tests and the 4500 games we have played at
40'/40, it was the original TheKing312 engine that has been used that means
TheKing 3.12
123 KB (125'952 Byte)
18th October 2000
and neither TheKing 3.12c [124 KB=126'976 Byte, 12th February 2001] nor TheKing
3.12d [124 KB=127'488 Byte, 5th July 2001]. We do not know whether this may be
of any significance regarding playing strength as the three versions should more
or less be identical. As far as we know, the orginal engine [18th October 2000]
does clear the hashtables after each move whereas the versions 3.12c and 3.12d
keep the hashtables in memory. We have not tested this because in view of the
good results with the original engine we had no reason to change things. But it
is important to know if people all over the world should obtain differing
results in their tests. And I believe to have read sometime that Johan de Koning
considers the clearing of the hashtables after each move to be a good [best]
method.
And furthermore I should like to mention, and this is also important, that we
are always using
WinBoard adapter
23th November 2000
22 KB [22'528 Byte]
for engine matches under the Fritz-GUI.
--------------------------------------------------------------------------------------------
3. Statistics
Cross table 2
<pre>
Program Score % Av.Op. Elo + -
Draws
1 TheKing312X : 58.5/100 58.5 2390 2450 62 51
39.0 %
2 Fritz7002 : 56.5/100 56.5 2392 2438 64 47
43.0 %
3 Shredder6 : 48.0/100 48.0 2402 2388 47 69
40.0 %
4 Junior7 : 48.0/100 48.0 2402 2388 49 69
36.0 %
5 ChessTiger14 : 47.5/100 47.5 2403 2385 46 68
41.0 %
6 Hiarcs732 : 41.5/100 41.5 2410 2350 55 62
33.0 %</pre>
Generated with Frank Schubert's Elostat v1.1 [Start ELO = 2400]
As we can see, the winner TheKing312x is only two points ahead of Fritz7002
which does not mean anything over 100 rounds. In the middlefield we have with a
margin of 8.5 and 9.0 points three almost equal programs [Junior7, Shredder6
and ChessTiger14] whereas Hiarcs732 takes last place with 6.0 points less than
the middlefield.
Despite the above facts and the 300 games played I am unable to declare which
program is the strongest one. If considering the margin of error, it becomes
clear that statistically spoken any of the programs in the T2002 tournament
could have won the title. There were simply not played enough games.
To illustrate what I mean, please compare the results of tournament leader
TheKing312X with the last placed Hiarcs732. In the worst case TheKing312X could
have obtained only 2399 ELO [2450 minus 51] but Hiarcs732 could at best have
reached 2412 ELO [2350 plus 62] which is still better than the worst case
scenario of TheKing312X. And eventually, do not forget that TheKing312X won its
match against Hiarcs732 by the close result of 11.0-9.0. And in view of this,
nobody can say for sure which one of these two is the better program. According
to the statistic-program 'Who is better' of Steve Maughan [programmer of UCI
engine Monarch] one would have to play 30 games between TheKing312x and
Hiarcs732 with the result of 21.0-9.0 in favour of TheKing to be able to say
with a certainity of 97 percent that TheKing312X is the better program. Another
calculation of Christophe Théron [programmer of ChessTiger] says: if you want 90
% reliability in your conclusions and have played 30 games, then you must assume
a +/- 12 % error margin in the winning percentage of the winner which translates
to a +/- 63 elo margin of error.
Conclusion:
Quite a huge number of games are needed to say something concrete about playing
strength between two programs. In any case, a result of 8.0-2.0 does not mean
anything. In this respect I would like to recall my experiment of a match over
100 rounds at time control 40'/40:
Gandalf 4.32g vs Program X
Games 1-10
3.0-7.0 [win program X]
Total 3.0-7.0 for program X
Games 11-20
6.5-3.5 [win Gandalf]
Total 9.5-10.5 for program X
Games 21-30
5.0-5.0 [draw]
Total 14.5-15.5 for program X
Games 31-40
3.5-6.5 [win program X]
Total 18.0-22.0 for program X
Games 41-50
4.5-5.5 [win program X]
Total 22.5-27.5 for program X
Games 51-60
3.0-7.0 [win program X
Total 25.5-34.5 for program X
Games 61-70
5.0-5.0 [draw]
Total 30.5-39.5 for program X
Games 71-80
8.0-2.0 [win Gandalf]
Total 38.5-41.5 for program X
Games 81-90
7.0-3.0 [win Gandalf]
Total 45.5-44.5 for Gandalf
Games 91-100
5.5-4.5 [win Gandalf]
Final match result 51.0-49.0 for Gandalf
Enough statistics now. But the intention of this all seemed very important to me
in order to demonstrate that one needs a lot of games and a lot of time to say
for sure with a given reliability that program A is better than program B. And
nowadays things have become more complicated due to the influence of opening
books, learning functions, tablesbases, hashtables and so on.
--------------------------------------------------------------------------------------------
4. T2002-Tournament or 10 World Chess Computer Championships
The World Chess Computer Championships seldom last more than 10 rounds.
Therefore the following interesting experiment: Why not divide the 100 rounds
T2002 tournament into 10 World Chess Computer Championships? And as you can see
from the cross tables below, each program with the exception of Hiarcs732 would
at least once have obtained the title. In view of these facts, I do no longer
give any computer championship the importance that people normally do. Each
tournament is just an unique event for me and once the winner is program X the
other time program Y. Despite these arguments, such events are of course very
important for the further development of computer chess.
<pre>
1. WM [world champion=Junior7]
Runden 1-10 T2002 Ath 1.3/64
Zurich, 2002.01.01 - 2002.01.05
Score Ju Ch Th Fr Hi Sh
------------------------------------------------
1: Junior7 7.0 / 10 XX =1 0= 11 == 11
2: ChessTiger14 6.0 / 10 =0 XX 11 10 1= ==
3: TheKing312X 5.5 / 10 1= 00 XX 1= 10 =1
4: Fritz7002 5.0 / 10 00 01 0= XX 11 1=
5: Hiarcs732 3.5 / 10 == 0= 01 00 XX 10
6: Shredder6 3.0 / 10 00 == =0 0= 01 XX
------------------------------------------------
30 games: +15 =10 -5</pre>
<pre>
2. WM [world champion=ChessTiger14]
Runden 11-20 T2002 Ath 1.3/64
Zurich, 2002.01.06 - 2002.01.13
Score Ch Fr Th Hi Sh Ju
------------------------------------------------
1: ChessTiger14 7.0 / 10 XX 1= 1= 1= == =1
2: Fritz7002 5.5 / 10 0= XX 01 =1 == 1=
3: TheKing312X 5.0 / 10 0= 10 XX =0 =1 =1
4: Hiarcs732 5.0 / 10 0= =0 =1 XX =0 11
5: Shredder6 4.0 / 10 == == =0 =1 XX 00
6: Junior7 3.5 / 10 =0 0= =0 00 11 XX
------------------------------------------------
30 games: +7 =14 -9</pre>
<pre>
3. WM [world champion=TheKing312X / Fritz7002]
Runden 21-30 T2002 Ath 1.3/64
Zurich, 2002.01.13 - 2002.01.21
Score Th Fr Ju Hi Sh Ch
------------------------------------------------
1: TheKing312X 6.5 / 10 XX == 1= =1 1= 01
2: Fritz7002 6.5 / 10 == XX == 11 == 1=
3: Junior7 6.0 / 10 0= == XX 10 11 1=
4: Hiarcs732 4.5 / 10 =0 00 01 XX =1 1=
5: Shredder6 3.5 / 10 0= == 00 =0 XX =1
6: ChessTiger14 3.0 / 10 10 0= 0= 0= =0 XX
------------------------------------------------
30 games: +3 =14 -13</pre>
<pre>
4. WM [world champion=Fritz7002]
Runden 31-40 T2002 Ath 1.3/64
Zurich, 2002.01.22 - 2002.01.29
Score Fr Th Sh Hi Ch Ju
------------------------------------------------
1: Fritz7002 7.0 / 10 XX 0= 1= 1= 11 1=
2: TheKing312X 6.0 / 10 1= XX 01 1= =1 =0
3: Shredder6 6.0 / 10 0= 10 XX 10 11 1=
4: Hiarcs732 4.5 / 10 0= 0= 01 XX 0= 11
5: ChessTiger14 3.5 / 10 00 =0 00 1= XX 1=
6: Junior7 3.0 / 10 0= =1 0= 00 0= XX
------------------------------------------------
30 games: +11 =10 -9</pre>
<pre>5. WM [world champion=TheKing312X]
Runden 41-50 T2002 Ath 1.3/64
Zurich, 2002.01.30 - 2002.02.05
Score Th Ju Ch Sh Hi Fr
------------------------------------------------
1: TheKing312X 6.5 / 10 XX 1= =1 == == =1
2: Junior7 5.5 / 10 0= XX == 01 =1 1=
3: ChessTiger14 5.5 / 10 =0 == XX =1 01 1=
4: Shredder6 5.0 / 10 == 10 =0 XX 01 =1
5: Hiarcs732 5.0 / 10 == =0 10 10 XX 1=
6: Fritz7002 2.5 / 10 =0 0= 0= =0 0= XX
------------------------------------------------
30 games: +9 =15 -6</pre>
<pre>
6. WM [world champion=Junior7]
Runden 51-60 T2002 Ath 1.3/64
Zurich, 2002.02.06 - 2002.02.13
Score Ju Fr Th Sh Ch Hi
------------------------------------------------
1: Junior7 6.5 / 10 XX 1= 1= 01 =0 11
2: Fritz7002 6.0 / 10 0= XX 11 =0 1= =1
3: TheKing312X 5.5 / 10 0= 00 XX =1 11 1=
4: Shredder6 4.5 / 10 10 =1 =0 XX 10 =0
5: ChessTiger14 4.0 / 10 =1 0= 00 01 XX 10
6: Hiarcs732 3.5 / 10 00 =0 0= =1 01 XX
------------------------------------------------
30 games: +11 =9 -10</pre>
<pre>
7. WM [world champion=TheKing312X]
Runden 61-70 T2002 Ath 1.3/64
Zurich, 2002.02.13 - 2002.02.17
Score Th Fr Hi Ju Sh Ch
------------------------------------------------
1: TheKing312X 8.0 / 10 XX == 1= 11 11 =1
2: Fritz7002 7.0 / 10 == XX == =1 =1 11
3: Hiarcs732 5.0 / 10 0= == XX 10 10 =1
4: Junior7 4.0 / 10 00 =0 01 XX == =1
5: Shredder6 4.0 / 10 00 =0 01 == XX =1
6: ChessTiger14 2.0 / 10 =0 00 =0 =0 =0 XX
------------------------------------------------
30 games: +11 =13 -6</pre>
<pre>
8. WM [world champion=TheKing312X]
Runden 71-80 T2002 Ath 1.3/64
Zurich, 2002.02.18 - 2002.02.25
Score Th Fr Sh Hi Ch Ju
------------------------------------------------
1: TheKing312X 6.5 / 10 XX =1 01 10 01 11
2: Fritz7002 6.0 / 10 =0 XX =1 =1 1= 01
3: Shredder6 5.5 / 10 10 =0 XX 11 =0 =1
4: Hiarcs732 4.5 / 10 01 =0 00 XX 1= 1=
5: ChessTiger14 4.0 / 10 10 0= =1 0= XX 0=
6: Junior7 3.5 / 10 00 10 =0 0= 1= XX
------------------------------------------------
30 games: +12 =9 -9</pre>
<pre>
9. WM [world champion=Fritz7002]
Runden 81-90 T2002 Ath 1.3/64
Zurich, 2002.02.26 - 2002.03.05
Score Fr Ch Sh Hi Th Ju
------------------------------------------------
1: Fritz7002 7.0 / 10 XX 01 01 1= 11 1=
2: ChessTiger14 5.5 / 10 10 XX =1 =1 == =0
3: Shredder6 5.0 / 10 10 =0 XX == =0 11
4: Hiarcs732 4.5 / 10 0= =0 == XX =1 01
5: TheKing312X 4.5 / 10 00 == =1 =0 XX =1
6: Junior7 3.5 / 10 0= =1 00 10 =0 XX
------------------------------------------------
30 games: +9 =12 -9</pre>
<pre>
10. WM [world champion=Shredder6]
Runden 91-100 T2002 Ath 1.3/64
Zurich, 2002.03.06 - 2002.03.13
Score Sh Ch Ju Th Fr Hi
------------------------------------------------
1: Shredder6 7.5 / 10 XX == 1= 11 == 11
2: ChessTiger14 7.0 / 10 == XX =0 =1 11 11
3: Junior7 5.5 / 10 0= =1 XX =0 01 11
4: TheKing312X 4.5 / 10 00 =0 =1 XX =1 01
5: Fritz7002 4.0 / 10 == 00 10 =0 XX =1
6: Hiarcs732 1.5 / 10 00 00 00 10 =0 XX
------------------------------------------------
30 games: +8 =10 -12</pre>
Considering the fact that TheKing312X would have obtained the title four times,
this program is a fair winner of the T2002 tournament.
--------------------------------------------------------------------------------------------
5. Opening report
<pre>
Data file: T2002.pgn (300 Partien)
Report: (300 Partien)
Generated with Scid 3.2, 2002.03.16
5.1. RESULT TRENDS
-----------------
5.1.2 Results lenghts and frequencies
Score Game Lenght Frequency
1-0 =-= 0-1 1-0 =-= 0-1
Report-Games 51.3% 60 78 62 32.0% 38.6% 29.3%
Alle Games 51.3% 60 79 62 32.0% 38.6% 29.3%
5.1.3 Shortest wins (White)
1: 1-0(28) Shredder6 - Hiarcs732, Zurich 2002 [21]
2: 1-0(30) TheKing312X - Fritz7002, Zurich 2002 [102]
3: 1-0(31) Junior7 - Fritz7002, Zurich 2002 [70]
4: 1-0(31) TheKing312X - Hiarcs732, Zurich 2002 [2]
5: 1-0(32) Fritz7002 - Hiarcs732, Zurich 2002 [27]
5.1.4 Shortest wins (Black)
1: 0-1(32) TheKing312X - Shredder6, Zurich 2002 [111]
2: 0-1(34) Shredder6 - Hiarcs732, Zurich 2002 [82]
3: 0-1(36) Hiarcs732 - ChessTiger14, Zurich 2002 [110]
4: 0-1(36) ChessTiger14 - Hiarcs732, Zurich 2002 [2]
5: 0-1(38) Hiarcs732 - Junior7, Zurich 2002 [104]
5.1.5 Move orders reaching the report position
Move ECO Frequency Score AvElo Perf AvYear
1: e4 B00a 124: 41.3% 47.1% 2002
2: d4 A40a 103: 34.3% 51.9% 2002
3: Nf3 A04 36: 12.0% 61.1% 2002
4: c4 A10 29: 9.6% 56.8% 2002
5: b3 A01 5: 1.6% 30.0% 2002
6: g3 A00t 3: 1.0% 66.6% 2002
__________________________________________________________
TOTAL: 300:100.0% 51.3% 20</pre>
<pre>Games : 300 (finished)
White Wins : 96 (32.0 %)
Black Wins : 88 (29.3 %)
Draws : 116 (38.7 %)
Unfinished : 0
White Perf. : 51.3 %
Black Perf. : 48.7 %
ECO A = 50 Games (16.7 %)
ECO B = 74 Games (24.7 %)
ECO C = 50 Games (16.7 %)
ECO D = 70 Games (23.3 %)
ECO E = 56 Games (18.7 %)</pre>
With the exception of Junio7 no program could avoid a short defeat. And
Hiarcs732 had to suffer even five times. In most cases the losses were not the
result of the choosen opening book line but the insufficient feeling regarding
King's safety. And sometimes the program started unsound attacks before having
finished development of pieces.
--------------------------------------------------------------------------------------------
6. General impressions
Despite the statistical equality of the six chess programs I have feelings and
emotions that do not correspond with the pure statistic material. And in such a
long tournament happens things you cannot explain.
After round 10 we set the book options to 'optimize' but nevertheless had
sometimes the feeling that this did not work in the sense of their inventors. On
the other hand we can assure that it seldom happened that a program showed an
evaluation of more than +/- 0.50 and in such cases only one program believed
that it stood so good wheras the opponent did not accept this. During the whole
tournament we permitted to restart games in only two cases where it was obvious
that the opening book was faulty when one could have interrupted the game with
either 1-0 or 0-1. And furthermore there were only two games we had to repeat:
once because it became necessary to kill a double and the other time due to an
exceptional shutdown of the PC.
Another - but not very important - question is: How many games ended with a draw
instead of a win or with a loss instead a draw as a result of a hashtable
collision? We do not know, but in some games we had the feeling that the
unexpected outcome could have been the result of a collision of hashtables. As
an example I give the game Junior7-Hiarcs732 of round 97 where in the following
position
8/R1bk3p/2r3p1/1Q1p1q2/3Pp3/7P/4NP2/4K3 b - - 0 50 Hiarcs732 played h7-h6 and
lost. After the correct Kd6 Junior7 would not have been able to win the game.
6.1 TheKing312X
The Dutch program with the CM_8777-settings plays a powerful and most efficient
attacking chess. Dynamic positions are dealt with at best. The program likes to
sacrifice material, especially pawns, to open lines at the right moment thus
activating its pieces to a maximum. And even if a pawn sacrifice does not lead
to a win, the opponent can only survive with a good defence and many such games
are really dramatic to look at. TheKing312X is playing very strong with the
queen. And the program recognizes repetition of moves and permanent chess
situations better than any other program. The endgame is rather convincing and
despite the fact that the program has no access to the tablebases in the
searching process, it is often TheKing312X to have more knowledge. I can only
say that I like the playing style of TheKing312X very much.
6.2 Fritz7002
Frans Morsch program is probably the most outbalanced chess program of the
world. Fritz7002 plays all phases of a game well, can also play rather
dynamically with a good end game. Fritz7002 is strong at king attacks but the
cases where pawns are sacrificed to get positional advantage are more seldom
than with TheKing312X. We noticed that Fritz7002 has won many equal endgames
although sometimes it went vice versa. We also have the feeling that some more
aggressive algos would not at all be bad for the program. On the other hand, due
to the good results there is of course nothing to blame. We deliberatly
abstained from using Fritz7006 from round 50 on as we did not want to falsify
the outcome of the tournament by different program versions.
6.3 Junior7
It is a must to have the Israelin program. Junior7 plays in a much different way
than his competitors, it plays a most attractive chess ... and the outcome of
each game is unclear, all can happen. Junior7 likes the dynamic way of chess but
contrary to TheKing312X his pawn sacrifices are much more speculative and [in my
opinion] too often goes the program wrong. A classical example could be seen in
the game Junior7-Shredder6, round 93,
r2q1rk1/1b2bpp1/p3pn1p/1p1pN3/2nP3B/P1NBP3/1PQ2PPP/2R1K2R w K - 0 where Junior7
castled kingside thus giving away a pawn for really nothing. A dangerous
opponent is Junior7 in unclear situations where the enemy king does not stand
well and in such cases the program has the behaviour of a snake that is always
ready to bite. The program has also the tendency for attacking moves like h2-h4
or h7-h5 but this also happens when such moves do heavily weaken the own
kingside position. The endgame game [not the technical ones] is probably the
weakest point of Junior7 because the program sometimes loses games that should
under no circumstances go lost: Example position
3r4/p4k1p/Bnp2bp1/4p3/P2nP3/1P6/1K1BN1PP/3R4 b - - 0 27 of the game
Junior7-ChessTiger14, round 19, that ended 0-1. This critic does of course not
bear in mind that Junior7 is on the other hand also able to play good endgames.
6.4 Shredder6
Frankly spoken I must admit that with Shredder6, the program of Stefan
Meyer-Kahlen, I have feelings of hatred and feelings of love at the same time.
Predecessor Shredder5 belonged to one of my favourite programs for analysing
purposes, it played positionally sound and showed a safe chess with a good
endgame. The latest Shredder6 behaves more aggressively and is faster at
tactics. But I am still not sure if this is in favour of the program. Anyway,
there is one thing that attracts attention: in the T2002 tournament we have
often been disappointed by the fast manner of how the program was changing
evaluations. If Shredder6 plays a move with let's say a positive score of 1.50
you cannot be sure that this evaluation will not change to -2.30 within the next
five moves. As a good example we can take the game Shredder6-TheKing312X [round
21], 8/3r2pk/2r2p1p/2p2P2/P5P1/1P1n3P/2QBK3/8 b - - 0 51 where Shredder6 after
51.Bd2 showed +2.30/15 in his favour only to loose the game at move 64. And at
first, we thought that this weakness would be responsible for the fact that the
program took last place after 50 rounds. But the longer the tournament went on,
the better was the play of Shredder6 and the winner of the last ten rounds was
Shredder6. Without thoroughly studying the games of Shredder6 I am unable to say
much more about this program who is able to play very good games but also bad
ones.
6.5 ChessTiger14
I call this program the French Karpov and indeed, ChessTiger14 has a similar
playing style with a doing nothing but do it well strategy and always ready to
bite if the opponent does not much care. The program started excellent with 13
points out of 20 games and took temporarily first place. But then the unusual
happened: 3 points only in the next 10 games and 6.5 points out of 20, that was
not the usual standard of ChessTiger14. I think no other program had less luck
in this tournament and more than once remained ChessTiger14 with unsufficient
material for a win. But the program must not blame for his 'bad' place because
his total score is only half a point behind Junior7 and Shredder6. In any case I
like ChessTiger14 as a very strong program with a reliable assessment of a given
position and as in previous times with the programs of Richard Lang can hardly
believe how is it possible to build a strong positional 'player' that is also so
strong at tactics.
6.6 Hiarcs732
The last place of Hiarcs732 looks quite normal. Marc Uniacke's program is the
oldest one of all participants in this tournament. But to our great surprise,
Hiarcs732 had only to fear Fritz7002. On the other hand one must admit that
chess programming has made further progress since the release of Hiarcs732 in
1999. And in many situations this progress can been seen when the latest
programs are fighting with Hiarcs732 who is tactically and in the endgame
somewhat weaker. But the weakest point of Hiarcs732 nowadays is the missing
feeling as far as King safety is concerned. And it was therefore logical that
Hiarcs732 was the program to be rather often defeated in short games. And after
all we can say; it's high time for Hiarcs8.
--------------------------------------------------------------------------------------------
7. Closing words
We had much fun with this tournament. The games will be put on sombody's
homepage. Jonas Cohonas promised to do that but I forgot his address.
Thanks go to Stefan Kleinert for his usual kind assistance and gratulations to
him for his fine CM_8777-settings. To create them he needed 8 months and they
were published first in March 2001. Also thanks to Alexander Kure [Austria] who
was responsible for the Nz7book.ctg for Nimzo732, a book that even nowadays
belongs to the best ones you can get. And finally many thanks to my chessfriend
Rolf Bühler, Zurich who has played the 300 games on his 3 Athlons 1.3/256.
This page took 0 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.