Subject: Re: Rodent games on Pi Tue Jan 26, 2021 8:58 pm
Karpov
CM9000 Karpov vs Rod4 Karpov
A King's Indian: Samisch that went very deep in book...I was thinking, draw. CM was even up a pawn, as this is part of the Samisch. But by move 30, White was on the defensive, cramped, and with a very weak Bishop; Black was in total control and had a Bishop pair staring down the White King. By move 40, Rod4 Karpov had a pawn on a3, doubled Rooks on the "b" and waiting for that final mistake. Although, there was no real blunder, SF eval says 44.Rc1 is Game Over. After 63. .. RxR, I called it.
Rod4 personalities have been handing it to CM9000 personalities, however I noticed that CM moves much, much too fast. At 30 sec/mv...CM had made the first 50 moves in under 10 minutes! I'm going to give CM 60 sec/mv next game (but keep Rod4 @ 30 sec).
Subject: Re: Rodent games on Pi Wed Jan 27, 2021 12:34 am
Fischer
CM9000 Fischer vs Rod4 Fischer
Below a "handicap" match: Rodent4 Fischer (Black) @ 30 sec/mv and CM9000 Fischer @ 60 sec/mv. A Ruy Lopez: Open Berlin (of course 1. Pe4) and Rodent4 Fischer gives up the exchange with 20. .. Bd6, but with decent compensation. VERY evenly calculated; At move 31, it is a flatline R+4P vs B+6P position....I let it go to move 60 and called a R+2P vs B+3P draw.
Now as to timing: Yes, CM was on 60 sec/mv and Rod4 30 sec/mv. But at the 60 move point, I wrote down both times used: CM: 29:31 Rod4: 25:34 So, CM is ultra, ultra conservative....by a factor of 2 (@60:sec/mv selection, CM still met the 30 sec/mv time limit!!!). CM used about 4 seconds more per move than Rod4 in this game........at double the time selection. I will probably do something creepy next personality game, like 56 seconds/mv for CM! So, here's the game:
[pgn] [Event " Handicap R: 30 CM 60s/mv"] [Black "Rodent IV v0.32 Fischer Brain"] [Date "Jan 26, 2021"] [Result "1/2-1/2"] [Round "A draw....timing looks good"] [Site "Rodent= NO book"] [White "CM9000 Fischer"]
Subject: Re: Rodent games on Pi Wed Jan 27, 2021 11:53 pm
Botvinnik
CM9000 Botvinnik vs Rod4 Botvinnik
A Symmetrical English and CM Botvinnik looked good for about 15 moves. But Rod4 Botvinnik started to take over after 16. Nd5? (That looks like a powerful post for the Knight....but SF says "Nope"). SF then says CM 30. Kf1? is bad (that didn't look all that bad, either). After Rooks trade, Rod4 plays 33. .. Ra2, and now it does look real bad for White...the CM Rook and Bishop are totally hemmed in. Rod4 just squeezed slowly, tighter and tighter and by move 50, it was Game Over. A very patient, relentless, Petrosian-like win.
[pgn] [Event " Handica R: 30 CM 56s/mv"] [Black "Rodent IV v0.32 Botvinnik Brain"] [Date "Jan 27, 2021"] [Result "0-1"] [Round "Botvinnik does python (the snake) like squeeze"] [Site "Rodent= NO book"] [White "CM9000 Botvinnik"]
1.c4 c5 2.Nf3 Nf6 3.Nc3 Nc6 4.d4 cxd4 5.Nxd4 e6 6.Nxc6 bxc6 7.e4 e5 8.Be3 Qa5 9.Bd3 Bc5 10.Bxc5 Qxc5 11.O-O O-O 12.Na4 Qd4 13.Qc2 Rb8 14.Rad1 c5 15.Nc3 Qd6 16.Nd5 Bb7 17.Rfe1 Bxd5 18.exd5 Rfe8 19.b3 g6 20.Re3 Qb6 21.Qd2 Kg7 22.Rh3 h5 23.Rg3 h4 24.Rh3 Rh8 25.Re1 Rbe8 26.Bc2 Qb4 27.Qxb4 cxb4 28.a3 a5 29.axb4 axb4 30.Kf1 d6 31.Ra1 Ra8 32.Rxa8 Rxa8 33.Rxh4 Ra2 34.Bd1 Rb2 35.f4 e4 36.Rh3 Nd7 37.Re3 Rd2 38.Be2 Nc5 39.h3 Rb2 40.Bd1 f5 41.g3 Rh2 42.h4 Kf6 43.Be2 Rh1+ 44.Kf2 Rb1 45.g4 fxg4 46.Bxg4 Rb2+ 47.Be2 Nd3+ 48.Kg1 Rb1+ 49.Kg2 Nxf4+ 50.Kf2 Kf5 51.Bf1 Ke5 52.Rg3 Kd4 53.c5 Kxc5 54.Bc4 Kd4 55.Rg4 Rb2+ 56.Kg1 Nh3+ 57.Kf1 Rh2 58.Rg3 e3 59.Rg4+ * [/pgn] Note: I forgot to make Black move 58. .. e3 on the DGT board, So, when I made move 59. Rg4, the Pi took it as a take back. Attempts to correct the board position = the Pi locked up.....my bad🤪🤪🤪....Rod4 was about to win the Rook, anyway.
Sc@lly likes this post
Piman
Posts : 114 Join date : 2020-11-23
Subject: Re: Rodent games on Pi Sat Jan 30, 2021 12:26 am
Anderssen
CM9000 Anderssen vs. Rod4 Anderssen
CM Anderssen selects the enterprising King's Gambit, but promptly loses with 9. Nd5?? This was a swashbuckling opening during Anderssen's time, but today even the book line 8. d3 leaves White down over 2 pawns. The game took 56 moves and Rodent4 was FF to mate. [pgn] [Event " Handica R: 30 CM 50s/mv"] [Black "Rodent IV v0.32 Anderssen Brain"] [Date "Jan 29, 2021"] [Result "0-1"] [Round "CM early mistake ...loses"] [Site "Rodent= NO book"] [White "CM9000 Anderssen"]
Subject: Re: Rodent games on Pi Sat Jan 30, 2021 9:38 pm
Anand
CM9000 Anand vs. Rod4 Anand
A Ruy Lopez Berlin and an absolutely 100% flatline draw....there's only Bishops and some pawns left! Until, CM Anand 32. BxB???? Game Over. SF eval instantly sees that White has lost the opposition. CM should never have traded Bishops....and definitely should have "seen" that.
So far, Rod4 on Pi3B has been crushing the equivalent "GM personalities" on CM9000. I am surprised...I thought CM9000 on PC was stronger. to the very impressive performance of Rodent4.
There's Alekhine left, which will be the next game. I will then try "THE ChessMaster" (CM9000 default) against Rod4 Bosboom and "Default" ....let's see what happens then.
[pgn] [Event " Handica R: 30 CM 60s/mv"] [Black "Rodent IV v0.32 Anand Brain"] [Date "Jan 30, 2021"] [Result "0-1"] [Round "CM blunder loses certain draw"] [Site "Rodent= NO book"] [White "CM9000 Anand"]
Subject: Re: Rodent games on Pi Sun Jan 31, 2021 8:21 pm
Alekhine
CM9000 Alekhine vs. Rod4 Alekhine
Another 100% draw and a late CM misstep loses the game. A Semi-Slav and for 47 moves: Drawn. Then, one "fingerfehler" by CM9000, 48. Re3?? is Game Over. The CM endgame simply can't keep up with Rodent4. Well, done Rod4, pouncing on the miscue.
[pgn] [Event " Handicap R: 30 CM 60s/mv"] [Black "Rodent IV v0.32 Alekhine Brain"] [Date "Jan 31, 2021"] [Result "0-1"] [Round "CM blunder and Rod wins"] [Site "Rodent= NO book"] [White "CM9000 Alekhine"]
Subject: Re: Rodent games on Pi Mon Feb 01, 2021 8:16 pm
CM9000 ChessMaster (Default) vs. Rod4 Bosboom
Now THAT'S the CM I remember. A deep English: Symmetrical and unlike the previous English in which CM personality Botvinnik couldn't get the job done, below.....the default CM9000 personality ChessMaster was absolutely terrific; simply Stockfish-like. SF analysis says, Rod4 Bosboom 21. .. Qb6?? is actually Game Over! ChessMaster never looked back; Bosboom valiantly tried an exchange sac with... 26. .. Nxb4? which made things even worse for Black. CM may have missed a couple of stronger moves here and there (36. Qh5+ and 37. Qh5+), but that's me getting very picky when CM is up a Rook! Mate at move 69 and here's another CM strength feature: Quick moves. CM calculated 69 moves to mate in 19:19 ON 45 SEC/MOVE!! That's 17 seconds a move...significantly faster than Rod4 which adhered much closer to its 30sec/mv setting (also conservative at 20-25 sec/mv on average). A very strong game by CM and it seems to indicate that the "GM personality" versions of CM are markedly weaker than the the default "ChessMaster" program...the default is VERY strong.
Subject: Re: Rodent games on Pi Mon Feb 01, 2021 8:41 pm
Piman wrote:
CM9000 ChessMaster (Default) vs. Rod4 Bosboom
Now THAT'S the CM I remember. A deep English: Symmetrical and unlike the previous English in which CM personality Botvinnik couldn't get the job done, below.....the default CM9000 personality ChessMaster was absolutely terrific; simply Stockfish-like. SF analysis says, Rod4 Bosboom 21. .. Qb6?? is actually Game Over! ChessMaster never looked back; Bosboom valiantly tried an exchange sac with... 26. .. Nxb4? which made things even worse for Black. CM may have missed a couple of stronger moves here and there (36. Qh5+ and 37. Qh5+), but that's me getting very picky when CM is up a Rook! Mate at move 69 and here's another CM strength feature: Quick moves. CM calculated 69 moves to mate in 19:19 ON 45 SEC/MOVE!! That's 17 seconds a move...significantly faster than Rod4 which adhered much closer to its 30sec/mv setting (also conservative at 20-25 sec/mv on average). A very strong game by CM and it seems to indicate that the "GM personality" versions of CM are markedly weaker than the the default "ChessMaster" program...the default is VERY strong.
It was white's passed pawns that but black away there.
Piman likes this post
Piman
Posts : 114 Join date : 2020-11-23
Subject: Re: Rodent games on Pi Mon Feb 01, 2021 10:05 pm
Correct. I had ranted so much diatribe in the commentary already, I left out:
Rod4 Bosboom 18. .. Be6? allows 19. Nxb5 and now the White CONNECTED, passed pawns will "rule the day" (whatever that nonsense, "rule the day", means🤪).
Piman
Posts : 114 Join date : 2020-11-23
Subject: Re: Rodent games on Pi Mon Feb 01, 2021 10:30 pm
Additional comments (since I'm on absolute lockdown in this big East Coast snow storm):
I know that I've posted, ad nauseum, these Rodent4 eval games....but it is a very enjoyable program. In fact, my opinion is .....that between Rodent and Chessmaster, you don't really need many (or any) more computer chess programs. Each program has this huge array of "personalities" that can satisfy virtually any chess player (even my buffoonery). That being said, Rod4 is a bit too strong....I prefer Rodent3, which has a lot of bottom dweller personalities, that even after a couple, you have a decent chance against. Oh...and tomorrow...maybe even tonight: The final eval game:
Default CM9000 vs Rod4 Default.
I think I'll even up the time settings to 1 min/mv (maybe 90 sec/mv for CM).
Sc@lly likes this post
Piman
Posts : 114 Join date : 2020-11-23
Subject: Re: Rodent games on Pi Tue Feb 02, 2021 1:54 am
Rod4 L21 2800 Elo on Pi 1 min/mv vs Default CM9000 on PC 1 1/2 min/mv.
The game between the highest levels of each program on their respective hardware. A Sicilian: Kan and Black has to be extremely accurate; CM9000 12. .. Qa5? is not accurate (SF says d5). By move 18, White has a significant advantage: 2 pawns vs 1 on the Queenside AND the Bishop pair....by move 21, White has a Queenside passed pawn and the Bishop pair. At 27. .. Rc6? CM tries a "Hail Mary: Exchange sac to get rid of the "a" pawn. It doesn't work: Game Over. I played out the complete game and Rodent4 2800 never stumbled and just ground down the CM position with the 2R vs R+N advantage to an 84 move mate.
A very well played finish by Rodent4 capitalizing on the early CM miscue.
Subject: Re: Rodent games on Pi Tue Feb 02, 2021 10:29 pm
Thanks for all these games. @Piman,do you think that Rodent IV weakening code is somewhat lacking?
I played some games against Chessmaster personalities: by myself, using Rodent or my unreleased private engine. It was a good engine for its time, but style differences are a bit too small for my taste. Often it is able to keep up with Rodent in the midgame, but gets into problems in the endgame. The same goes for another great sparring partner - ProDeo. I regard it a minor success - back when Ed organised YATC tournaments, Rodent's endgame was seriously lacking. It has been fixed by search, not by evaluation, so even now Chessmaster is able to create a dangerous outside passer a bit too often to my taste.
Perhaps it is caused by Chessmaster design philosophy. I have read somewhere that The King (Chessmaster) engine is extension-based, and perhaps these extensions were tuned with middlegame play in mind. Late move reduction in comparison is a lazy idea, but it gives depth advantage that is difficult to surmount, and it is a killer in the endgame. Perhaps we all would be better off style-wise, had it not been invented, but for a modern engine it is a must.
Sc@lly and Piman like this post
Piman
Posts : 114 Join date : 2020-11-23
Subject: Re: Rodent games on Pi Wed Feb 03, 2021 12:26 am
Missed one: Dynamic
Rodent IV Dynamic vs. CM Polgar
A Sicilian (B33: Open) and Polgar is programmed with very provocative, 4. .. Qb6. However, after 20. Qf6, Black can no longer castle. Polgar tries to free "herself" with 21. .. g5?? Game Over. It took over 50 moves, but Dynamic took the winning Q vs. 2R position successfully to mate; no mistakes. For "personality" vs. "personality" match ups, Rod4 is too strong...mostly wins, with occasional draws vs. CM9000. For CM9000 to keep up with Rod4, the highest level, ChessMaster, must be used. [pgn] [Event " 30s/mv Cm 45 s/mv"] [Black "CM9000 Polgar"] [Date "Feb 2, 2021"] [Result "1-0"] [Round ""] [Site "Rodent= NO book"] [White "Rodent IV Dynamic Brain"]
Subject: Re: Rodent games on Pi Wed Feb 03, 2021 12:49 am
nescitus wrote:
Thanks for all these games. @Piman,do you think that Rodent IV weakening code is somewhat lacking?
I played some games against Chessmaster personalities: by myself, using Rodent or my unreleased private engine. It was a good engine for its time, but style differences are a bit too small for my taste. Often it is able to keep up with Rodent in the midgame, but gets into problems in the endgame. The same goes for another great sparring partner - ProDeo. I regard it a minor success - back when Ed organised YATC tournaments, Rodent's endgame was seriously lacking. It has been fixed by search, not by evaluation, so even now Chessmaster is able to create a dangerous outside passer a bit too often to my taste.
Perhaps it is caused by Chessmaster design philosophy. I have read somewhere that The King (Chessmaster) engine is extension-based, and perhaps these extensions were tuned with middlegame play in mind. Late move reduction in comparison is a lazy idea, but it gives depth advantage that is difficult to surmount, and it is a killer in the endgame. Perhaps we all would be better off style-wise, had it not been invented, but for a modern engine it is a must.
Hello Nescitus,
1. Your Q: Piman,do you think that Rodent IV weakening code is somewhat lacking?
I can't answer that, because I don't even understand the question But, If you are asking me: is your weakening code actually weakening any of your Rodent4 personalities? A: I don't think so But it doesn't matter: There's Rodent3 with plenty of weak personalities.
I thought your Rod4 philosophy was to make all the GM personalities play as strong as possible on Rodent, with tweaks just to simulate their particular style.
2. WRT CM9000 (The King)....I have found historically that The King is strong middlegame, not so strong endgame: Exactly as you have pointed out.
3. Back to Rod4: as stated on a previous post: just too strong for me....it's GM stuff. I'm a patzer....your dumb Rodent3 personalities are perfect for me.
4. "so even now Chessmaster is able to create a dangerous outside passer a bit too often to my taste." I'm not a programmer.....but, looks like you just have to make some endgame tradeoffs. I haven't done an extensive analysis (in fact, no analysis at all) but, in my opinion, your endgame had no problems against CM9000. And I think your super strong middlegame (at least against CM) set the table for endgame success.
In sports, a team can hide its weaknesses by working on and making it's strengths....overwhelming.
Home 
. However, after 20. Qf6, Black can no longer castle. Polgar tries to free "herself" with 21. .. g5?? Game Over. It took over 50 moves, but Dynamic took the winning Q vs. 2R position successfully to mate; no mistakes.