YABRL: List strongest free engine, but ...

[Robert Allgeuer]

Geschrieben von:/Posted by: Robert Allgeuer at 24 April 2004 00:10:55:

Under my test conditions List 5.12 scored higher than Ruffian 1.0.1 and is now the strongest free engine in YABRL. It also scored almost equal to the leading Ruffian 2.1.0, just some tiny 5 points behind.
However - as posted in this forum a few days ago - List has a bug and does not know how to mate against a lone king. I therefore had to make a decision, and decided to test in such a way that all opponents resign a lost game. If the opponent engine did not support resignation, I adjudicated such a game. This was done in order to avoid that engines that do resign have a disadvantage.
As a bottomline this means that the resulting rating is the one of List under the assumption that all opponents resign, if they are lost (resign threshold 7 pawn units). If one would let play all games to the end, List´s rating would drop significantly.
List won the direct comparison against Ruffian 2.1.0 and 1.0.1; interestingly Anmon 5.30 is the engine that scored the highest against List.
List played in the ChessPartner GUI with a book built from Dann Corbit´s book.pgn file.
For conditions and tools please refer to the link below. Next I will put all these results into perspective by adding a strong commercial engine, Chesstiger 15.
Robert



    Program                     Elo    +   -   Games   Score   Av.Op.  Draws
 01 Ruffian v2.1.0            : 2675   18  27   798    70.7 %   2522   24.9 %
 02 Ruffian v2.0.0            : 2675   17  27   840    71.6 %   2514   25.8 %
 03 List v5.12                : 2670   19  28   696    69.6 %   2526   26.6 %
 04 Ruffian v1.0.1            : 2648   17  23   956    68.8 %   2511   27.2 %
 05 DeepSjeng v1.6ntb         : 2627   20  25   779    65.0 %   2519   24.9 %
 06 Ktulu v4.2                : 2591   20  22   855    60.5 %   2517   24.3 %
 07 SmarThink v0.17a          : 2589   20  21   859    59.7 %   2521   25.5 %
 08 Thinker v4.5b             : 2589   21  20   816    59.6 %   2521   33.2 %
 09 Crafty v17.14DC           : 2588   18  18  1038    60.1 %   2517   33.0 %
 10 Crafty v19.06DCntb        : 2583   20  19   901    58.8 %   2521   30.5 %
 11 Aristarch v4.21           : 2578   18  19  1057    58.6 %   2517   24.7 %
 12 Aristarch v4.37           : 2573   21  18   839    57.6 %   2520   36.6 %
 13 Delfi v4.3                : 2567   20  19   980    57.2 %   2517   25.4 %
 14 Crafty-MPC v18.15DC       : 2561   20  19   964    56.4 %   2517   27.3 %
 15 Delfi v4.2                : 2561   25  25   580    58.1 %   2504   27.2 %
 16 SmarThink v0.16b++        : 2559   21  21   836    58.0 %   2503   24.3 %
 17 Little Goliath 2000 v3.9  : 2557   19  18  1060    55.6 %   2518   26.4 %
 18 Crafty v18.15DC           : 2557   22  22   741    59.0 %   2493   29.0 %
 19 El Chinito v3.25          : 2556   22  20   820    55.5 %   2517   27.6 %
 20 SoS 3                     : 2555   19  18  1059    55.3 %   2518   22.8 %
 21 Yace Paderborn            : 2552   20  18  1040    55.2 %   2516   25.9 %
 22 Pepito v1.59 profile      : 2550   20  18  1060    54.7 %   2518   25.6 %
 23 Aristarch v4.4            : 2547   36  34   319    54.1 %   2518   20.4 %
 24 SoS 4                     : 2542   23  20   839    53.4 %   2519   24.1 %
 25 Green Light Chess v3.00   : 2542   20  17  1060    53.4 %   2518   26.0 %
 26 Yace v0.99.56             : 2541   33  29   380    54.3 %   2511   26.6 %
 27 Little Goliath 2000 v3.5  : 2536   31  25   440    53.6 %   2511   30.9 %
 28 Tao v5.6                  : 2526   25  20   776    50.6 %   2522   24.5 %
 29 Anmon v5.30               : 2525   24  19   820    51.0 %   2518   26.8 %
 30 Amyan v1.59               : 2513   18  23   931    49.8 %   2514   26.0 %
 31 Pharaon v2.62             : 2509   17  21  1057    48.6 %   2518   24.2 %
 32 Crafty v19.01DC           : 2499   24  19   815    50.4 %   2496   25.5 %
 33 LambChop v10.99           : 2495   18  20  1057    46.5 %   2519   22.5 %
 34 Ktulu v3.9                : 2491   19  24   779    48.6 %   2501   26.1 %
 35 Gromit v3.8.2             : 2489   18  20  1037    45.8 %   2518   23.2 %
 36 SlowChess v2.89b          : 2488   19  21   917    46.0 %   2516   23.9 %
 37 KnightDreamer v3.2        : 2486   19  21   960    45.4 %   2518   24.8 %
 38 Anmon v5.22               : 2483   19  22   899    46.4 %   2508   26.7 %
 39 Comet B44-2               : 2480   18  20   980    44.4 %   2519   26.4 %
 40 Amy v0.8.3                : 2476   20  19  1049    43.9 %   2519   19.0 %
 41 SoS v11-99                : 2476   33  34   359    46.0 %   2504   17.3 %
 42 Tao v5.4                  : 2475   19  19  1039    44.1 %   2516   21.0 %
 43 Dragon v4.4.3             : 2470   19  20   925    43.3 %   2517   25.8 %
 44 Comet B62-3               : 2458   20  19   960    41.4 %   2518   25.5 %
 45 Francesca M.0.0.9         : 2442   20  18  1058    38.9 %   2520   25.0 %
 46 PostModernist v1.007      : 2441   21  18   980    38.9 %   2520   25.0 %
 47 Comet B60                 : 2434   22  21   780    41.2 %   2496   25.6 %
 48 Leila v0.53h              : 2426   23  18   978    36.9 %   2520   21.0 %
 49 Tcb v0045                 : 2422   22  18   979    36.3 %   2520   24.3 %
 50 Resp v0.19                : 2404   23  17   960    33.9 %   2519   23.4 %
 51 Nejmet v3.07              : 2389   25  18   876    33.2 %   2510   22.3 %
 52 SlowChess v2.78           : 2378   27  19   790    33.5 %   2497   19.6 %
 53 Exchess v4.03             : 2329   30  15   959    24.9 %   2521   21.4 %
 54 Beowulf v2.2              : 2303   33  14  1020    22.4 %   2519   17.5 %

Games        :  23676 (finished)
White Wins   :   9615 (40.6 %)
Black Wins   :   8100 (34.2 %)
Draws        :   5961 (25.2 %)
Unfinished   :      0
White Perf.  : 53.2 %
Black Perf.  : 46.8 %

(3) List v5.12                : 696 (+392,=185,-119), 69.6 %
Ruffian v2.1.0                :  20 (+ 10,=  5,-  5), 62.5 %
DeepSjeng v1.6ntb             :  20 (+  6,=  8,-  6), 50.0 %
SmarThink v0.17a              :  20 (+ 11,=  5,-  4), 67.5 %
Thinker v4.5b                 :  20 (+ 10,=  7,-  3), 67.5 %
Crafty v17.14DC               :  20 (+  8,= 10,-  2), 65.0 %
Ktulu v4.2                    :  20 (+  5,=  6,-  9), 40.0 %
Crafty v19.06DCntb            :  20 (+  8,=  6,-  6), 55.0 %
Aristarch v4.21               :  20 (+ 11,=  6,-  3), 70.0 %
Aristarch v4.37               :  20 (+  9,=  7,-  4), 62.5 %
Delfi v4.3                    :  20 (+  8,=  6,-  6), 55.0 %
Crafty-MPC v18.15DC           :  20 (+ 13,=  5,-  2), 77.5 %
El Chinito v3.25              :  20 (+ 14,=  6,-  0), 85.0 %
Little Goliath 2000 v3.9      :  20 (+  8,=  8,-  4), 60.0 %
SoS 3                         :  20 (+ 10,=  6,-  4), 65.0 %
Pepito v1.59 profile          :  20 (+ 13,=  7,-  0), 82.5 %
Yace Paderborn                :  20 (+ 12,=  4,-  4), 70.0 %
SoS 4                         :  20 (+ 10,=  3,-  7), 57.5 %
Green Light Chess v3.00       :  20 (+  7,=  8,-  5), 55.0 %
Anmon v5.30                   :  20 (+  5,=  5,- 10), 37.5 %
Pharaon v2.62                 :  18 (+ 11,=  3,-  4), 69.4 %
LambChop v10.99               :  20 (+ 15,=  3,-  2), 82.5 %
Gromit v3.8.2                 :  20 (+ 15,=  5,-  0), 87.5 %
KnightDreamer v3.2            :  20 (+ 14,=  3,-  3), 77.5 %
Ruffian v1.0.1                :  20 (+  9,= 10,-  1), 70.0 %
Comet B44-2                   :  20 (+ 14,=  1,-  5), 72.5 %
Tao v5.6                      :  19 (+ 10,=  6,-  3), 68.4 %
Amy v0.8.3                    :  20 (+ 14,=  5,-  1), 82.5 %
Dragon v4.4.3                 :  20 (+ 13,=  7,-  0), 82.5 %
Comet B62-3                   :  20 (+ 13,=  4,-  3), 75.0 %
Francesca M.0.0.9             :  19 (+ 12,=  4,-  3), 73.7 %
PostModernist v1.007          :  20 (+ 13,=  6,-  1), 80.0 %
Leila v0.53h                  :  20 (+ 17,=  0,-  3), 85.0 %
Tcb v0045                     :  20 (+ 14,=  4,-  2), 80.0 %
Resp v0.19                    :  20 (+ 14,=  3,-  3), 77.5 %
Exchess v4.03                 :  20 (+ 16,=  3,-  1), 87.5 %





YABRL (Yet Another Blitz Rating List)

[Roger Brown]

Geschrieben von:/Posted by: Roger Brown at 24 April 2004 00:52:00:
Als Antwort auf:/In reply to: YABRL: List strongest free engine, but ... geschrieben von:/posted by: Robert Allgeuer at 24 April 2004 00:10:55:

However - as posted in this forum a few days ago - List has a bug and does not know how to mate against a lone king.
As a bottomline this means that the resulting rating is the one of List under the assumption that all opponents resign, if they are lost (resign threshold 7 pawn units).

Hello Robert,
Doubtlessly you will disagree with my remarks. I agree that as TD you can implement whatever rules you deem fit. However I submit that if the aim is to come up with some sort of rating list that an engine which cannot mate a lone king - an endgame routine for a beginner - then that engine cannot be the strongest engine in any universe.
Think about it. I am a patzer. I cannot mate a lone king. Do you really think that the other engine which dances around for 50 moves is doing something wrong in salvaging the half point? In the real world I agree it would be insulting to play on in such a situation - hence my concession that the rules are yours to make - but in computerchess it seems unfair to bias the results this way.
Robert Hyatt (Crafy) specifically is against that sort of arrangement. Complete the mate is his motto. Of course his GM guests are always granted the concession of resignation. You will not get many GM's to play against your program if you do not know when to call it a day....

I therefore had to make a decision, and decided to test in such a way that all opponents resign a lost game. If the opponent engine did not support resignation, I adjudicated such a game. This was done in order to avoid that engines that do resign have a disadvantage.

Hmmm, you seem to giving an adavantage to engines which cannot resign...


Believe me, I have lost a few of those. The win was there but I didn't have someone to save me from my blunders, tactical and otherwise......


If one would let play all games to the end, List´s rating would drop significantly.

Doesn't this statement tell you that the process you are using is not fair to the other engines in the list?

Next I will put all these results into perspective by adding a strong commercial engine, Chesstiger 15.

Any bets on the outcome?

Thanks for the labour of love. You ran many games. For free. It is appreciated.
Later.

[Michael Byrne]

Geschrieben von:/Posted by: Michael Byrne at 24 April 2004 03:46:29:
Als Antwort auf:/In reply to: YABRL: List strongest free engine, but ... geschrieben von:/posted by: Robert Allgeuer at 24 April 2004 00:10:55:

Under my test conditions List 5.12 scored higher than Ruffian 1.0.1 and is now the strongest free engine in YABRL.
However - as posted in this forum a few days ago - List has a bug and does not know how to mate against a lone king.

Robert,
With all due respect to your opinion, I wholeheartedly disagree with your assesment that: " ...List is now the strongest free engine ..."
Especially in light of one of your other statements "...List has a bug and does not know how to mate against a lone king."
Perhaps, you should qualify your statement and say, "when playing under my rules, i.e, that is when game is +7 and I adjudicate the win to the winning side, List is the strongest free engine." But in a real tournament, any of the other operators should not resign against List and act like they are from Missouri and say "show me".
That is not to say that I find your tournament uninteresting.
Best,
Michael

[Michael Scheidl]

Geschrieben von:/Posted by: Michael Scheidl at 24 April 2004 04:37:38:
Als Antwort auf:/In reply to: YABRL: List strongest free engine, but ... geschrieben von:/posted by: Robert Allgeuer at 24 April 2004 00:10:55:

(...)
However - as posted in this forum a few days ago - List has a bug and does not know how to mate against a lone king.

Sounds more like a tbs. access bug to me... normally, an engine which has correct tbs. access has no other choice so to speak, than to play a mating sequence in tbs. postions. I never saw tbs. problems reported about List.
I looked into the other posting, where you write List was running in Chess Partner with alls 5 piece tbs. (in the older posting you wrote, conditions were
Winboard 4.2.3
WB Tourney Manager 0.60 (Jori Ostrovskij)
Uci2wb 2.0 (R. Pfister) for UCI engines.
I wonder if that List 5.12 problem was observed in Arena, too (or in Fritz when the GUIs tbs. access is switched off). Could it be a kind of protocol communication problem between List and CP? Do you notice normal tbs. accesses by List 5.12?
(Currently I cannot test it becaue due to a computer replacement, my installations are still incomplete.)
Regards,
M.Scheidl

[Robert Allgeuer]

Geschrieben von:/Posted by: Robert Allgeuer at 24 April 2004 09:25:11:
Als Antwort auf:/In reply to: Re: YABRL: List strongest free engine, but ... geschrieben von:/posted by: Roger Brown at 24 April 2004 00:52:00:

However - as posted in this forum a few days ago - List has a bug and does not know how to mate against a lone king.
As a bottomline this means that the resulting rating is the one of List under the assumption that all opponents resign, if they are lost (resign threshold 7 pawn units).

Hello Robert,
Doubtlessly you will disagree with my remarks. I agree that as TD you can implement whatever rules you deem fit. However I submit that if the aim is to come up with some sort of rating list that an engine which cannot mate a lone king - an endgame routine for a beginner - then that engine cannot be the strongest engine in any universe.
Think about it. I am a patzer. I cannot mate a lone king. Do you really think that the other engine which dances around for 50 moves is doing something wrong in salvaging the half point? In the real world I agree it would be insulting to play on in such a situation - hence my concession that the rules are yours to make - but in computerchess it seems unfair to bias the results this way.
Robert Hyatt (Crafy) specifically is against that sort of arrangement. Complete the mate is his motto. Of course his GM guests are always granted the concession of resignation. You will not get many GM's to play against your program if you do not know when to call it a day....
I therefore had to make a decision, and decided to test in such a way that all opponents resign a lost game. If the opponent engine did not support resignation, I adjudicated such a game. This was done in order to avoid that engines that do resign have a disadvantage.

Hmmm, you seem to giving an adavantage to engines which cannot resign...
Believe me, I have lost a few of those. The win was there but I didn't have someone to save me from my blunders, tactical and otherwise......

If one would let play all games to the end, List´s rating would drop significantly.
Doesn't this statement tell you that the process you are using is not fair to the other engines in the list?

Next I will put all these results into perspective by adding a strong commercial engine, Chesstiger 15.
Any bets on the outcome?

Thanks for the labour of love. You ran many games. For free. It is appreciated.
Later.

Hi,
I know the dilemma.
List is there and has a bug, unfortunately. But is is a highly interesting engine.
To deal with this problem there are several possibilities:
1) Do not test it
2) Just run it and whatever games it produces are added to the rating list. Disadvantage: this will unfairly influence the whole rating list, because those engines that resign will lose games (because they assume that their opponent can mate), while those that do not resign will get a relatively high number of draws against a strong opponent. Their rating will go up compared to the engines that resign. Also List´s rating will be a strange, in my view rather worthless hybrid rating and will in particular depend on how many of its opponents do resign and how many do not resign.
3) Reconfigure all engines so that they do NOT resign. This is impossible because some engines simply DO resign.
4) Configure all engines to resign (and adjudicating thoses games of engines that CANNOT resign). The disadvantage here is essentially that it covers up the bug. It measures List as if the bug were fixed, which however is not the case.
I think that most testing of List we have seen so far - without that it ever has been discussed - is according to approach 2, close to 4: that is most tourneys are run with resign on where possible, but there will also be some engines in these tourneys that cannot resign.
My main goal is to have an accurate measurement of playing strength (as far as possible), and in particular I do not want to bias ratings of the OTHER engines just because one single engine has a bug, so approach 2 for me is out. 3 is impossible, which leaves 1 and 4. I went for 4 because List is highly interesting; for some other engines I would possibly go for 1.

CT15: I had a few test games (but not against Ruffian yet in order to keep the suspense). I think it is for the number one spot.... But I also consider to run Ruffian 2.0.2 with the Leiden book soon, which may also score higher, because in fact in the meantime it is quite apparent that Ruffian 2.1.0 is not the strongest Ruffian.

A long post, sorry ...
Robert



YABRL (Yet Another Blitz Rating List)

[Robert Allgeuer]

Geschrieben von:/Posted by: Robert Allgeuer at 24 April 2004 09:28:47:
Als Antwort auf:/In reply to: Re: YABRL: List strongest free engine, but ... geschrieben von:/posted by: Michael Byrne at 24 April 2004 03:46:29:

Under my test conditions List 5.12 scored higher than Ruffian 1.0.1 and is now the strongest free engine in YABRL.
However - as posted in this forum a few days ago - List has a bug and does not know how to mate against a lone king.

Robert,
With all due respect to your opinion, I wholeheartedly disagree with your assesment that: " ...List is now the strongest free engine ..."
Especially in light of one of your other statements "...List has a bug and does not know how to mate against a lone king."
Perhaps, you should qualify your statement and say, "when playing under my rules, i.e, that is when game is +7 and I adjudicate the win to the winning side, List is the strongest free engine." But in a real tournament, any of the other operators should not resign against List and act like they are from Missouri and say "show me".
That is not to say that I find your tournament uninteresting.
Best,
Michael

Well, on purpose I wrote "strongest free engine in YABRL", which I thought says exactly this, it is not meant as a general statement.
I am aware of the problem and I knew when posting that this is controversial, I tried to explain my reasoning in my response below to Roger.
Robert



YABRL (Yet Another Blitz Rating List)

[Robert Allgeuer]

Geschrieben von:/Posted by: Robert Allgeuer at 24 April 2004 09:43:28:
Als Antwort auf:/In reply to: Re: YABRL: List strongest free engine, but ... geschrieben von:/posted by: Michael Scheidl at 24 April 2004 04:37:38:

(...)
However - as posted in this forum a few days ago - List has a bug and does not know how to mate against a lone king.

Sounds more like a tbs. access bug to me... normally, an engine which has correct tbs. access has no other choice so to speak, than to play a mating sequence in tbs. postions. I never saw tbs. problems reported about List.
I looked into the other posting, where you write List was running in Chess Partner with alls 5 piece tbs. (in the older posting you wrote, conditions were
Winboard 4.2.3
WB Tourney Manager 0.60 (Jori Ostrovskij)
Uci2wb 2.0 (R. Pfister) for UCI engines.
I wonder if that List 5.12 problem was observed in Arena, too (or in Fritz when the GUIs tbs. access is switched off). Could it be a kind of protocol communication problem between List and CP? Do you notice normal tbs. accesses by List 5.12?
(Currently I cannot test it becaue due to a computer replacement, my installations are still incomplete.)
Regards,
M.Scheidl

List definitely does access tablebases in my configuration, this I have checked. I have not tested a configuration of List without tablebases, and whether the problem appears there, though. One thing I remember: Aristarch had a tablebase bug fixed (either from 4.37 to 4.40 or from 4.40 to 4.41), List uses the same DLL as Aristarch, but there was no bug-fix.
Not overly many results have been posted about List, I also assume that most tourneys run with resign on so that they may not run into the problem too often.

If I want to test Chesstiger and Rebel (which is rather unstable in Winboard and WBTM) I have to use another GUI (in my case ChessPartner) and essentially relax the GUI-rule. I have checked CPU usage (99%) and memory consumption to be identical to running under WB for each single engine.
I plan to update the conditions and attach it to one of my next posts.
Yes, it accesses.
I am not really running Arena here, have just installed it once and not really used. But I could try to try it out ...
Robert



YABRL (Yet Another Blitz Rating List)

[Robert Allgeuer]

Geschrieben von:/Posted by: Robert Allgeuer at 24 April 2004 11:38:36:
Als Antwort auf:/In reply to: Re: YABRL: List strongest free engine, but ... geschrieben von:/posted by: Robert Allgeuer at 24 April 2004 09:43:28:

I wonder if that List 5.12 problem was observed in Arena, too (or in Fritz when the GUIs tbs. access is switched off). Could it be a kind of protocol communication problem between List and CP? Do you notice normal tbs. accesses by List 5.12?

I have reproduced the problem in Arena 1.07. In addition this is also on a different computer, so conditions are independent, yet the bug occurs there as well.
The situation is a follows:
- With tablebases enabled (5 piece EGTBs) List 5.12 cannot mate against a lone king and moves aimlessly. This was observed both in ChessPartner 5.3 and Arena 1.07
- With tablebases disabled List 5.12 can mate, which suggests that the problem is a tablebase access problem.
Further observation:
- List always displays "Mate in 500", never something like "Mate in 4". So all mating moves appear to be of identical score for List, which may explain why it chooses just an arbitrary move (according to the tablebase they are all won anyway) rather than the one move that mates immediately.
I assume the author of List does not read this forum; I also do not know the e-mail address of him, so I cannot submit the problem to him.
Robert



YABRL (Yet Another Blitz Rating List)

[Roger Brown]

Geschrieben von:/Posted by: Roger Brown at 24 April 2004 12:45:57:
Als Antwort auf:/In reply to: Re: YABRL: List strongest free engine, but ... geschrieben von:/posted by: Robert Allgeuer at 24 April 2004 09:25:11:

A long post, sorry ...

You have nothing to be sorry about.

I read every word.

Later.

[Norm Pollock]

Geschrieben von:/Posted by: Norm Pollock at 24 April 2004 15:06:25:
Als Antwort auf:/In reply to: List tablebase access bug geschrieben von:/posted by: Robert Allgeuer at 24 April 2004 11:38:36:

I think List 5.12 has a sharp drop-off in strength without the tbs. The best solution is to adjudicate where necessary. Imho, List (with tbs) is second only to Ruffian as the best free engine.
-Norm

[Michael Scheidl]

Geschrieben von:/Posted by: Michael Scheidl at 24 April 2004 20:18:23:
Als Antwort auf:/In reply to: List tablebase access bug geschrieben von:/posted by: Robert Allgeuer at 24 April 2004 11:38:36:

(...)

I assume the author of List does not read this forum; I also do not know the e-mail address of him, so I cannot submit the problem to him.

Thanks; I have forwarded the info to the CSS-Forum.
(I think the problem did not occur when the Fritz GUI was used, which AFAIK is the case for some ratings lists like CSS and FRC, because as mentioned that GUI handles tbs. positions autonomous by default... but I'm not 100% sure though what happens with an UCI Engine then.)
Regards,
M.Scheidl

[Michael Scheidl]

Geschrieben von:/Posted by: Michael Scheidl at 25 April 2004 17:24:29:
Als Antwort auf:/In reply to: Re: List tablebase access bug geschrieben von:/posted by: Michael Scheidl at 24 April 2004 20:18:23:

(...)

Thanks; I have forwarded the info to the CSS-Forum.
(I think the problem did not occur when the Fritz GUI was used, which AFAIK is the case for some ratings lists like CSS and FRC, because as mentioned that GUI handles tbs. positions autonomous by default...

Meanwhile St.Zipproth (of Aristarch), who also hosts the List Download, wrote that this is a known bug which will be fixed in the next version. (He also acknowledged the remark about using List in Fritz, but said the bug would occur there too, when the GUIs tbs. access would be switched off.)
http://f23.parsimony.net/forum50826/messages/97921.htm
Regards,
M.Scheidl

[Robert Allgeuer]

Geschrieben von:/Posted by: Robert Allgeuer at 25 April 2004 19:17:45:
Als Antwort auf:/In reply to: Re: List tablebase access bug = known bug geschrieben von:/posted by: Michael Scheidl at 25 April 2004 17:24:29:

(...)

Thanks; I have forwarded the info to the CSS-Forum.
(I think the problem did not occur when the Fritz GUI was used, which AFAIK is the case for some ratings lists like CSS and FRC, because as mentioned that GUI handles tbs. positions autonomous by default...

Meanwhile St.Zipproth (of Aristarch), who also hosts the List Download, wrote that this is a known bug which will be fixed in the next version. (He also acknowledged the remark about using List in Fritz, but said the bug would occur there too, when the GUIs tbs. access would be switched off.)
http://f23.parsimony.net/forum50826/messages/97921.htm
Regards,
M.Scheidl