Thursday, August 23, 2007

How to Use Personal Chess Trainer 2007


One thing that separate masters from amateurs is the ability to recognize tactical patterns. A popular tool that's available to help amateurs is Personal Chess Trainer. After using it for some time, I believe that how we use the tool makes a big difference.

When you're faced with a position you either recognize the pattern or you have to calculate it. In the first case, you already knew the pattern. In the second case, make sure you spend some time looking at the solution and the elements that made it work! If you don't store an accurate pattern in your short-term memory, how can you expect to recall it during your next game? Our human tendency to look at the problem more than the solution--that's backwards!

If your time for a unit was slower than 3-6 seconds per position, you obviously didn't recognize all the tactical patterns. So repeat it! What's the point of continuing on unless you have actually learned the patterns? There's no prize for finishing a module quickly. If you ensure the knowledge gets into your short-term memory, a good portion of it should reach your long-term memory, and then PCT's frequent reviews will help you to retain it.

On my first pass through PCT I was in too much of a rush, and as a result didn't learn the patterns very well. If it takes me 35 seconds to spot a tactic when I'm told it's there, my chance of spotting it over the board under time pressure is slim.

This second pass, I'm going slow. Slow and accurate. That's the formula that worked on Chess Tactics Server and I think it works here as well.


Glenn Wilson said...

Nice blog!

First, one does not stop studying tactics because one is studying endgames. That is a comment on your previous post I suppose. Endgames are often highly tactical in their own way.

I'm no expert on circles or PCT. I've been using PCT for about 2 weeks today and I have done 47 units so far (42 in Module 2 and 5 in Module 1). I may be higher rated (USCF 2010) than most people using it. I decided to skip Module 1 as it seemed too easy after poking at it a bit. Maybe that was a mistake, we'll see. I'll probably go back and do all of Module 1 just for fun.

In the second half of Module 2 the exercises tend towards four or five moves and I can frequently make the moves as fast as I am able to physically move. Not the first time seeing it, but in third and later repetitions. But, I feel like I am getting more out of it by going slower. Not to "see" the first move of the solution which I see instantly (usually). But to see and understand the alternatives. What I am doing now, before I make the first move, is visualize the final move. It is slower than just slamming out the first move, then the next, then the next. But I am making sure I understand the final point before I make the first move.

I do like that the positions (starting around unit 20) are hard enough that on first exposure I have to work hard to get it and sometimes I don't.

transformation said...

great to see you back, likeForests!

i have great respect for this tool, PCT, based on what i have heard across many quarters, but am already full with GM games, RHP, 1001 Reinfeld et. al., CTS... plenty.

id love to hear more of what you think, and have great confidence you will eloqently and soundly report this.

after my run, and hot bath, im tired, time for bed ... woke frest on 5.6 hours sleep a long time ago, so this day is over now!

likesforests said...

Glenn, thanks for stopping by.

Yeah, I shoulda been reviewing tactics all along. Endgame tactics are somewhat different than middlegame tactics.

I like your approach of visualizing the full sequence before you move a piece. I actually try to do the same thing. I keep at it until I can do it quickly as well.

The moron defense looks quite clever, especially for someone who knows their endgames. Hmm, will think about it.

"USCF 2010... the exercises tend towards four or five moves and I can frequently make the moves as fast as I am able to physically move." You must be a force to be reckoned with over the board! My best success so far is drawing someone at around your level. ;)

likesforests said...

David, yes, I think you are set! Of all your tools, going through GM games sounds the most fun. I know I enjoy playing through good endings.

Glenn Wilson said...

I just finished unit 48 trying to move quickly. It took me an average of 8 seconds per position. So, my current speed is slower than I thought. It seems it takes me about two seconds per position to get oriented and then 0-1 seconds per move. Time above that (up to an average of about six seconds) is "think" time.

One thing I intend to try after finishing this module is to toggle the "White pieces always on bottom" option. That way, at least in black to move positions, the board will be "upside down" compared to what I have been training with. It will be interesting to see how that affects my time and ability. I expect it to slow me down but I'm not sure by how much.

I'm sure that what I have been doing is not the optimal approach but I am also sure that it is valuable.

Glenn Wilson said...

A couple more observations on moving quickly against PCT.

First, some may not know what the time is a measure of. It is a measure of the time you spend moving the pieces. PCT does piece animation as it moves and you can "pre-move" while it is animating through its move and no time will count against you if you finish your next move before it finishes its move.

On sequences where I have several long moves and its moves are short (and quick) I'll get a couple of seconds counted against me if I move as fast as I am able to. But most often, all of my time is a measure of how long it took me to make the first move as I can usually make the other moves in "zero time" (while it is moving) or close to it.

The board is large on my screen (my resolution is 1280 x 1024) and I see no way to adjust that. That affects my time in that I can not visually take in the whole board in one glance. Also it will occasionally add a second or two making "long moves" because of the time to move the mouse to the piece and then make the move (but as I said above, usually not).

I frequently "know" the move to make before I have seen the whole board. Is it correct to start the sequence of moves before seeing the whole board? Is that good pattern recognition or a bad practice?

That will often mean taking on faith based on pattern recognition of a potion of the board that the bishop you are going to need to deliver mate in four is over in the other corner where you need it to be.

Or, should one validate all of the conditions, the combination as a whole before making the first move?

I will often know, for example, that my moves will be Nc6 (attacking the Queen), Qxh7+, Rh5+, Rh8#. I see all that looking at the top half and mostly top right quarter of the board. In the position I am thinking of, that last move is not mate without a bishop somewhere on the long diagonal supporting it.

How much can one and should one rely on pure pattern recognition vs. calculation (even just light validation) before moving in such trainings? In a real game I am not sac'ing the queen just because I think I recognize the patern!

And something interesting happens when one actually calculates before moving against PCT. One starts to realize that PCT often does not play the best defense -- it often plays the thematic defense that best shows the point of the combination. Crafty is quite helpful in these situations to validate what I am thinking ("its not me, its you").

Ultimately this does not represent any real dilemma. I know the difference between developing pattern recognition using PCT and playing my moves in a real game. By developing the pattern recognition I will see the possibilities so I will know to calculate with those possibilities in mind. Of course, in blitz, I might just throw out the moves based on "intuition" without calculating everything out...

I do not see the value of PCT and the circles, etc. in teaching exact moves in exact middlegame positions because I am unlikely to have the entire exact positions repeated in any of my games. I see the value in learning to spot tactical motifs or "chunks" of positions that I can instantly recognize and know to examine for useful possibilities.

My point, if I have one, is I am conflicted at whether attempting to move as fast as possible (especially the first move) or taking in the whole position and then moving is a better approach? Or somewhere in between.....?

Glenn Wilson said... earlier comments about how time is recorded in PCT are not entirely correct. It is correct insofar as how it records the time usage for that exercise during that exercise. For example I started a new module and had 30 seconds to start and "made my moves" in 4 seconds and it shows 26 seconds remaining. The wall clock time was longer because it was also moving during this time with slow animation (is there a way to speed up its move animation? I don't see one).

But, when I view History after just that one exercise it shows I took 14 seconds for that problem. I did a second exercise and took 3 seconds. View History now shows a total time of 28 seconds and 14 seconds average per exercise.

So, my comments about moving slower than I thought I was (based on Viewing History) may be wrong. And view History does not match the clock time displayed during the exercise but evidently is recording wall clock time per exercise?

likesforests said...

"I am conflicted at whether attempting to move as fast as possible (especially the first move) or taking in the whole position and then moving is a better approach? Or somewhere in between.....?" It's a terrific question. Moving as soon as you recognize a pattern but before you can confirm all its nuances would still seem to practice your pattern recognition, and faster. But verifying everything before you make your move is practicing like a real game and helpful for blitz when you really might have 12s to make a choice. You're not the first to ask. dktransform and I prefer to do it the slow way, but some knights prefer the fast lane. Who knows which is best? :)

girorgio said...