Monday, March 16, 2009

Sacrificing position for opportunity

White's eighth move in the following game fragment is a great example of gaining activity at the expense of pawn structure. The example and commentary are from Chapter 7 of Chess for Zebras. For those interested, at the end of this post I include a bunch of statistics pertaining to the key position (stats about the 23 games in my database that have passed through the key position, as well as Fritz's evaluation of the position).
rowson_initiative.pgn


Pretty cool, huh? I want to use this example to tie together some points that came up during the planning exercise.

Analysis, planning, and time
As I've discussed before there is some consensus that quiet positions don't deserve much detailed analysis. Don't do too much analysis because there are too many reasonable futures for this game so the actual game will quickly deviate from what you imagined previously during your analysis of candidate moves. On the other hand, how can you tell, without doing some analysis, if you don't have a position like the one in this game that both Rowson and Shereshevsky think had a brilliancy on move 8?

Any nontrivial chess position can be examined for hours and you still probably won't feel like you have a complete grasp over all it's complexities. Of course, in a real game you should not waste time. These two facts stand in some tension. Based on the game fragment, Rowson says:
This is another example that highlights the difficulty of chess. Just when you think you have some positional understanding, a famous trainer comes along and tells you that 8 Nf3 is a good move.

Making use of opportunities like 8 Nf3 is hard enough, but even more complex is judging how much time to spend on deciding whether such moves are real opportunities. For instance, sometimes you have a choice between two moves. You know that one is decent and retains your advantage, but you can see that the other might be much stronger and win by force. However, in order to find out whether it wins by force you will have to think for several minutes. Many players just [go ahead with the analysis], only to realize it is not so clear-cut and then play the simpler move but with much less time on the clock.

A key element of practical strength, I believe, is to consider the value of Time at such moments...Once you start thinking about a line it is difficult to stop, so one aspect of being skilled in your use of Time is thinking of the opportunity cost of thinking time: if I lose twenty minutes now, I will certainly miss opportunities later; is it worth the investment?
That's beautifully sums up my problem: I am not good at answering that question. In positions without clear tactical opportunities or hit-me-over-the-head obvious strategic features, I am never sure if I am missing something big (like Nf3), or if I should just hurry up and move and save my time for cases where I know it is crucial for me to find the right variation. I tend to err on the side of taking too long to think, looking for Nf3 type moves when none exist and I should just plonk my Rook on an open file and be done with it.

For those that thought I did too much analysis on the position I posted last week, I would have to respectfully disagree. If you look at my planning session for that post, there is actually very little analysis at all. Indeed, the post ends where the analysis would begin, with a list of candidate moves! In other words, there is essentially zero analysis in my planstorm. Rather, it is all higher-level planning in terms of what pieces should go where, that sort of thing, the exact kind of thing people say to think about when it would be silly to do detailed analysis of game trees. That is, I made plans, which is exactly the sort of thing you are supposed to do in quiet positions, no?

Perhaps the problem (if there was one) was that I focused on long-term versus short-term plans, or that my plans were wrong, or that I discussed too many plans, but it definitely wasn't that I was doing too much analysis!

So how do I improve at planning, and become faster at doing it well? As I said in the comments, hopefully by doing it. "My hope is that I will develop better intuitions by thinking through a variety of "quiet" positions that don't require a lot of tactical analysis." Perhaps as I work through additional quiet positions I will get a better sense of which ones have these opportunities like Nf3, versus those that I should simply move quickly to save thinking time for the sharp positions.

And lest I forget, I should always harken back to the most important chess lesson I have ever learned, Safety First. All this planning and strategy isn't worth squat if I'm dropping pieces.

What follows are statistics and Fritz analysis for white's eighth move.

Some statistics on the position after black's seventh move
In my 3 million game database (average ELO 2170) the position after 7...Nxe5 has occured 23 times, and here are the stats from White's perspective. Nf3 clearly does best, though this hasn't exactly been heavily tested at top-level tournaments.
Move------Frequency -----------%W-L-D---------#W-L-D
1: Nc3---------11------------------ 18-64-18 -----------2-7-2
2: Nf3----------9-------------------67-11-22----------- 7-1-2
3: Be2----------2-------------------50-50-0------------1-1-0
4: f4------------1--------------------0-100-0-----------0-1-0

For what it's worth, Fritz 9 in infinite analysis mode for about five minutes yields the following evaluation of the top five moves for white at move eight:
Move-----Evaluation
1. Nd2------- +0.77
2. Nf3------- +0.66
3. h3 -------- +0.59
4. Be2------- +0.58
5. Bd4 ------ +0.45

13 Comments:

Blogger katar said...

Wow, your blog is suddenly talking about actual CHESS-- a lot of it and at a rather high level. You know... actual positions, plans, moves, and mistakes-- instead of generally applicable improvement methods, thought processes, or other general pedagogical theories. It is not enough to study the ocean, you must submerge yourself in it. I think you will be DANGEROUS if you ever come back to play OTB-- i wouldnt want to play ya.

3/16/2009 01:49:00 AM  
Blogger Blue Devil Knight said...

What's the world coming to, eh?

3/16/2009 02:26:00 AM  
Blogger chesstiger said...

Their are different ways to tackle the problem i guess but one of them is what Silman descrives in his book Reasses your chess with his thinking proces with imbalances.

Also one should find something about thinkingproces of Dan Heisman (Novice Nook) in the Archives of chesscafe.

And maybe the book Excelling at chess calculation of Jacob Aagaard tackles the problem.

So yes, it demands much study and practise but somewhere must be a method that is suitable for you. But i dont know if you even then would come up with moves like Nf3 since that is a move of which the rule of Rowson 'to learn chess you must forget all you ever learned of chess' comes to the horizon.

3/16/2009 05:37:00 AM  
Blogger Blue Devil Knight said...

Chesstiger: have you read this? No offense but you often seem to come to the party a couple of hours late, without a gift. :)

3/16/2009 09:21:00 AM  
Blogger chesstiger said...

No i havent read that one otherwise i would have pointed you to there instead of tiring my fingers with all that typing. :-P

Thanks for pointing me towards there, will read it carefully.

3/16/2009 11:10:00 AM  
Blogger Blue Devil Knight said...

Chesstiger: sorry I came off a bit jerky trying to be funny in my last comment.

3/16/2009 11:53:00 AM  
Blogger Hank said...

This is fun!

I set up Fritz running the Rybka 2.32a engine in Infinite Analysis mode for 10 hours(reaching 21 ply), and came up with the following top 5 candidate moves and evaluations:

Move -------- Evaluation
1. Nd2------- +0.31
2. Be2------- +0.28
3. Nf3------- +0.24
4. Nc3------- +0.16
5. Nh3------- +0.10

Best,
Hank

3/16/2009 03:24:00 PM  
Blogger Blue Devil Knight said...

Hank: that's awesome, thanks!

I saw 'The man who laughs' on Friday night. Holy crap that was an amazing film (with a professional silent film pianist too!).

3/16/2009 04:32:00 PM  
Blogger Hank said...

That film sounded really cool - unfortunately I couldn't make it.

BTW, at 17 ply Rybka had Nf3 as her top choice, but then she changed her mind as she analyzed deeper.

That's the thing about computer evaluation - it tells you which move gives you theoretically the best end position if you play just like a computer for 21 moves, but it doesn't always tell you which move gives you the *easiest* position to play out and possibly win with, as a *human* player! :)

-- Hank

3/16/2009 11:07:00 PM  
Blogger chesstiger said...

"Chesstiger: sorry I came off a bit jerky trying to be funny in my last comment."

No problem, i knew it was said tongue in cheek (sp?).

3/17/2009 06:35:00 AM  
Blogger Polly said...

Actually I found your whole planning exercise very interesting. I had a game this past weekend that was kind of like that. Not the position itself, but the whole idea of being in a middle game with no clear tactics, and trying to figure out what direction I wanted to go. As I was looking at the position I was thinking to myself "Damn this reminds me of BDK's exercise." Those are some of the toughest types of positions to deal with. Often it's easier to make the simple move, when the "brilliant" move is too difficult to evaluate.

3/17/2009 11:17:00 PM  
Blogger Phaedrus said...

BDK,

In a lot of books attention is drawn to the exceptional. 8 Nf3 is a beautiful example. The move is so unusual that I wonder if it is very educational. When we are struggling to get to 2000/2200, it seems to be wisest to focus on the conventional, and try to use this as efficiently as possible.

I would never recommend a student of mine to try to find a move during training that is rejected by a 2500/2600 grandmaster like Rowson on first sight. And I would most certainly not tell him to look for moves like this during a game.

BTW, I love this blog and the fundamental questions that are raised in it. Keep up the good work.

3/18/2009 04:19:00 AM  
Blogger Blue Devil Knight said...

Phaedrus: that is a great point, and one reason after summarizing the next chapter, I am done with Rowson. He speaks to someone who is already hit a plateau with the conventional, and needs to break free. I am nowhere near that stage! I discussed all this a bit here.

I understand his points, but I am not at the point where I should worry too much about them. Rules are good, I should learn to use them before I reject them! :)

3/18/2009 12:33:00 PM  

Post a Comment

<< Home