Saturday, May 21, 2005

What's a patzer to do?

I'm a little bit frustrated with Tasc Chess Tutor (TCT). It is overall a great program, and I loved it for one-move instruction (e.g., mate in one). However, now that I am moving into 2+ move problems, I am running into a dilemma: TCT frequently selects crappy moves.

I include here one position, but it is indicative of my plight. Check out the position at this link. The problem theme is 'Scare away or capture the defender', with black to move. The piece defended is the rook on D4, which my queen wants. Unfortunately it is defended by white's queen. All of these problems involve you forcing the defender to move. TCT says the right answer is Rxa4, capturing the knight. Fritz and Crafty agree. Howver, TCT also would have white play Qxa4 next move! Fritz and Crafty both say this is foolish. The right move for white (according to Fritz and Crafty) is b2-b3. Of course, black still has control and ultimately will end up with the desired rook, but this variation is a few moves down the line and I frankly was hoping to BUILD UP to such stuff. Since I rarely see such combinations myself, this puts me into the time-consuming cycle of checking things with Fritz when I am skeptical of TCT's analysis. Also, with this theme it is given a priori that you have only two moves in the 'correct' sequence, so if I find a multimove combination then I know it is isn't correct in TCT's evaluation.

It might not be so bad if TCT got the first move right, as it did here (after all, if my opponent made that queen move, I would respond by snagging white's rook). But I have also found positions where even the 'correct' first move is not what Fritz would do.

It is a little frustrating that my training program is not playing the way it should play. Is this like the SAT, where you shouldn't over-analyze for fear of getting your percent correct dropped? If I were a confident, competent chess player, this probably wouldn't bother me, but when I doubt TCT I have to go to Fritz to make sure I am not being stupid (I am being stupid about 90% of the time when I have doubts). It is just too time consuming and breaks up the flow of my drills.

Since TCT is just a precircle in the Divine Tragedy, and since usually my skepticism is flat out wrong, I think I will not stress about it too much. When I hit the actual Circles with Convekta's Chess Tactics for Beginners, then I will explore things a little more before moving on to higher circles. This Precircle is only meant to expose me to the basic chess knowledge that any aspiring C-class player should have. It is certainly succeeding at that!

What would MDLM do?

6 Comments:

Blogger King of the Spill said...

I know what you mean. They seem to do this in steps 1-3, sort of explaining why the first move is successful in overly simplistic terms...kind of like setting you up to play against someone making mistakes, just like in a real game.

I think you would find many <1400 opponents rush into taking (Qxa4), and if they do you are ready for the reply :-).

5/22/2005 03:56:00 AM  
Blogger knightwiz said...

I think you shouldn't care too much about what Fritz says on the position. Computer tends to analyse things far different from us, sometimes they discard the "obvious" move based on complex variations that lead to a +0.25 pawn advantage(i can't judge a side to be fourth a pawn ahead anyway). In the exercise you showed, black would be a knight up anyway.

Just as example, I was playing a game recentely, and had a winning endgame if I capture a pawn, but after the game, analysing with Crafty, it doesn't want me to take the pawn. It saw something like "mate in 14" with K+P+B against K+P, but I don't think taking the pawn as a wrong move, as any human player would go for it. Sometimes it's hard to follow computer advices..

just my 2 cents =)

5/22/2005 03:54:00 PM  
Blogger Blue Devil Knight said...

Very good points. Knightwiz, you are right on about that: in the variations TCT prefers, it is never worse than the 'best' line by more than half a pawn . In the type of games I play, this probably won't make a big difference :)

5/22/2005 04:11:00 PM  
Blogger Temposchlucker said...

TCT does so from an educational point of view. Which is extremely helpful to present you the most clear cut examples. To check this out with Fritz is a total waste of time. If you insist on using Fritz make sure brute force is on. 99.9 % of the first moves is ok in TCT.
You will find errors in every problemsolving CD. Whether this is TCT, CT-art, George Renko or whatever.
If you ever do Polgars book "Middlegames" where you are left alone after your opponent plays back the best way, you will learn to appreciate clear cut examples.

5/22/2005 04:57:00 PM  
Blogger CelticDeath said...

What everyone has said is true. You'll find this even in CT-Art 3.0 and Reinfeld's Winning Chess Sacs and Combos book. You are simply being shown the tactic in detail. With some tactics problems, the solution involves the best defense. With others, you are shown simply the idea behind the tactic as if to say "Now, if White/Black takes...then...."

5/23/2005 10:29:00 AM  
Blogger Unknown said...

Yep. This is the tough part of tactics - the reply isn't always "best play". There are a lot of tactics in CT-ART like this. Some of the moves are just plain dumb or contrived.

Don't quit overanalyzing, though. It's really helps develop your skills.

5/23/2005 12:07:00 PM  

Post a Comment

<< Home