total chess training has wrong problems

[Uri Blass]

For example I solved the following tactics in CT-art that is part of the program and I was disappointed to find that the program did not accept some moves that I suggested that rybka also suggested

Here is the tactics with my comments(I suggested 3.a3 and 9.e8Q but total chess training rejected it.

Something is also wrong in the rating that this software gives me(not only in tactics but also in other subjects like strategy)

My rating is only slightly above 2000 but in most cases this software gives me rating of more than 2300.


[Event "URS 1990"]
[Site "?"]
[Date "2010.01.12"]
[Round "?"]
[White "Glek"]
[Black "Chernikov"]
[Result "1-0"]
[Annotator "so k"]
[SetUp "1"]
[FEN "r1b2rk1/1p3pbp/p3p1p1/2qPB3/8/1B3Q2/P4PPP/3RR1K1 w - - 0 1"]
[PlyCount "27"]

{256MB, Rybka3.ctg, URI-AMD} 1. Bd6 Qxd6 2. dxe6 Qb4 3. Re4 (3. a3 {
(I suggested but ct-art considered it as a mistake)}) 3... Qc3 4. Qxf7+ Kh8 5.
e7 Rxf7 (5... Bf5 6. exf8=Q+ Rxf8 7. Re8) 6. Rd8+ Bf8 7. Rxf8+ Kg7 8. Rxf7+ Kh6
9. g3 {? this blunder is marked by ct-art as !} (9. e8=Q {
I suggested this move but ct-art considered it as a mistake} Bf5 {
does not work because of} 10. Rh4+ {or other alternatives} Kg5 11. Qe7+) 9...
Qc1+ (9... Bg4 10. Rxg4 Qc1+ 11. Kg2) 10. Kg2 Qc6 {??} (10... Bh3+) 11. e8=Q
Bh3+ 12. Kxh3 Rxe8 13. Rh4+ Kg5 14. f4# 1-0

[tomgdrums]

For example I solved the following tactics in CT-art that is part of the program and I was disappointed to find that the program did not accept some moves that I suggested that rybka also suggested

Here is the tactics with my comments(I suggested 3.a3 and 9.e8Q but total chess training rejected it.

Something is also wrong in the rating that this software gives me(not only in tactics but also in other subjects like strategy)

My rating is only slightly above 2000 but in most cases this software gives me rating of more than 2300.


[Event "URS 1990"]
[Site "?"]
[Date "2010.01.12"]
[Round "?"]
[White "Glek"]
[Black "Chernikov"]
[Result "1-0"]
[Annotator "so k"]
[SetUp "1"]
[FEN "r1b2rk1/1p3pbp/p3p1p1/2qPB3/8/1B3Q2/P4PPP/3RR1K1 w - - 0 1"]
[PlyCount "27"]

{256MB, Rybka3.ctg, URI-AMD} 1. Bd6 Qxd6 2. dxe6 Qb4 3. Re4 (3. a3 {
(I suggested but ct-art considered it as a mistake)}) 3... Qc3 4. Qxf7+ Kh8 5.
e7 Rxf7 (5... Bf5 6. exf8=Q+ Rxf8 7. Re8) 6. Rd8+ Bf8 7. Rxf8+ Kg7 8. Rxf7+ Kh6
9. g3 {? this blunder is marked by ct-art as !} (9. e8=Q {
I suggested this move but ct-art considered it as a mistake} Bf5 {
does not work because of} 10. Rh4+ {or other alternatives} Kg5 11. Qe7+) 9...
Qc1+ (9... Bg4 10. Rxg4 Qc1+ 11. Kg2) 10. Kg2 Qc6 {??} (10... Bh3+) 11. e8=Q
Bh3+ 12. Kxh3 Rxe8 13. Rh4+ Kg5 14. f4# 1-0

I am not surprised as I have had a lot of problems with ChessOk software. I downloaded Peshka and purchased the Chess Endgame Training course from them and it never added any problems from the demo! Aquarium was always messy to me as well.

They have good software ideas but bad execution or so it seems to me.

[Albert Silver]

It isn't quite that simple, though I understand perfectly. My rating, according to it, is close to 2600. I'm afraid I don't really blame Convekta for this as it is an almost impossible task to rate a player properly according to their results in a test suite. Obviously, more tactical players will do far better, but what can you do? Instead of worrying about the absolute rating it attributes, I'd suggest using it as a measuring stick: it gave you 2300 now. See if you don't improve this over time. Your improved "2500" rating may be wrong as an absolute rating, but it will still most likely be correct in demonstrating your improved tactical skills compared to your first "2300" results.

As to the correctness of the results, I have a number of genuine cooks, whether in the main lines, or branches in the solutions, that I have found, and then verified with engines. At the time it came out (1999 I think), I think they weren't checked with engines, and simply reproduced the solutions from the book they came from, a Russian classic by Grischuk's trainer. This phenomenon is common, and spread out in chess literature and other tactical aides.

The real meat of it, is being able to do focused work on pattern recognition, with controlled conditions on difficulty. In other words, do 20 problems rated 1600-2000, and then 20 more, rated 2000-2200. The ratings may not be accurate in an absolute sense, but the relative difficulty of the problems from the first group to the second will be.

As always, follow the Russian proverb, "believe it, but verify it."