Calculating like a dog

The following rant is more applicable to over-the-board games rather than correspondence.

The process of calculation in chess has always interested me. I thought it might help me clarify my own understanding if I jot down some notes here. The following is just my current opinion.

Years ago, as a combination as being an engineer and reading Kotov, I believed that the calculation process should be very structured. I even sketched a “flow chart” of the steps I should go through…. “do this… then this… repeat for all moves… then do this… etc.”. Fortunately enough, upon discussing this with a GM, he very quickly dismissed my approach. The main factor being that my method may work well for a computer, but not for the human brain.

I’ve since tried to model the calculation process in other ways – so that I can attempt to understand it better – but ultimately I don’t think it’s feasible. The human process of calculating chess positions is just too complex in reality. However, I’d still like to mention an analogy that I considered to have some value. Although, I also stress that it falls down in many respects too.

I think that a human calculating a chess position is in many way like searching for drugs in a large building, where the “searcher” is a sniffer dog and his handler. It’s important to view the dog and handler as working together as a single unit and not as two separate things. Admittedly, this analogy sounds contrived, or even silly, but it helps me to consider the way I calculate (i.e. search a chess position).

- the handler provides structure to the process. He know’s the layout of the building, and he’s ultimately in command of if/when individual rooms get searched. However, he does not possess any “sense of smell” to follow in terms of where drugs may be located.

- alternatively, the dog’s strong point is it’s “sense of smell”. It’s movements don’t appear to be structured, but instead are more spontanteous, impulsive, etc. In my mind, I imagine the dog darting about in an unpredictable manner, constantly lead by its nose.

But before I go mad here and completely lose the plot, consider how the task would be approached if only one half of the pair attempted to do it alone…

(i) the handler would have to tediously search every drawer, closet, etc. in every room. He’d have nothing to direct his search other than just exhaustively covering the whole building. It could be rather painstaking and slow.

(ii) alternatively, the dog may be fortunate if its sense of smell hits upon the right direction, and leads it quickly to a find. However, it could also pick up a false lead, or no lead to follow at all… then it could go to a “clean” part of the building, and remain there due to a lack of prompting to search elsewhere. It could repeatedly keep searching the same rooms, while some unsearched areas elsewhere still exist. In short, it lacks structure to it’s approach.

Obviously the reality is that the pair work together to help overcome each other’s weaknesses. Ok, back to chess…

I believe that the human mind has similar elements working together. i.e. we have a conscious part that is capable of applying structure to our thought processes. But we also have a subconscious part that is highly automatic, spontaneous, impulsive, intuitive (acting like a “sense of smell&rdquo😉, etc.

Consider what happens if I show you a chess position. Without prompting, you’ll automatically start visualising moves, seeing patterns, assessing, etc. etc. Out of many legal moves, you’ll automatically start picking out main candidates, and maybe even lines will pop into your head. Your intuition will be guiding most of the direction. This “auto pilot” is like the dog. It just appears to hit upon certain leads and impulsively follows them. It may quickly hit upon the right areas; or it may wander aimlessly and miss the point.

Alternatively, at a given moment, you may consciously say to yourself, e.g., “there’s many moves to consider here, let’s list them and work through them one at a time”. This is the handler part, trying to give some control and discipline to the process.

Some notes…

- for many players, myself included, sometimes the dog is guilty of leading the handler too much. i.e. the handler doesn’t know when to force the dog to stop and look elsewhere

- however, the opposite happens too, where the handler doesn’t follow the dog enough and imposes too much structure. Since the dog’s freedom is too limited, it may be following a true lead, but the handler forces it to change direction.

- structure is a necessary part of a good calculation process. However, our brains tend to be naturally lazy… we’d rather just let the dog roam on its own accord. This is not ideal. However, if we force ourselves to be too structured, the task may become too tedious and unnatural. At the risk of stretching this too far, it’s a bit like asking a human to search alone… the reality is that boredoom, etc. would probably affect their interest in the task and hence reduce their effectiveness.

In summary, getting a good balance between our structured and automatic “roaming” thought processes is very difficult. But in order to improve our calculation abilities, I believe we have to consider carefully how they currently work together. And of course, even if we’re aware of an unbalance, training ourselves to do better is another story.

Anyway, given that I ain’t to trying write a book here, I’ll leave it at that. If anyone wants to discuss, I’ll add some more. But more likely, I’m expecting someone to comment that I must have found some drugs recently… ;-)

This is a great post. That's really all I have to say. Rec'd.

I wonder excactly how much calculation is necessary for, say, a master player. I have read where a grandmaster says he sacrificed the exchange, for example, without knowing or calculating to the conclusion of the position. I believe sometimes it is humanly impossible to do it, because the calculations would run into the million and only a computer could crunch the positions. Instead, the master has many "lighthouse" structures in his head which suggest he is on the right course, positions, which in general terms fit the situation he's in. To make it real simple let's say a player sees a combination which gives up a piece or two to get two rooks on the 7th, while his opponent's king is on the 8th. He doesn't see the mate or the countless problems which he can create for his opponent. He simply knows through experience this is a fatal situation. A beginner might think, "If i give up the piece i get two rooks doubled on the 7th, so what? Where do i go from there. He sees a couple of checks, then everything goes beyond his horizon, so he doesn't sac the pieces. The beginner is primarily using calculation, much like a computer, and he isn't so it doesn't work very well. This example is admittedly very elementary. A good player has hundred or thousands of these "lighthouse" scenarios in his head. Apparently, the more he has, the better player he is. I guess this is better known as pattern recognition and the theory of combination or endgame study is based on it.

I agree. In some positions, intuition,etc. will play a significant role since, as you suggest, it may be too complicated to calculate all the subsequent variations.

However, at other times, calculation will be critical. Although pattern recognition is an essential aid, many positions are too specific to rely upon our intuition alone. For example, consider the typical pattern involving a bishop sacrifice on h7 (often followed by a Ng5+ and then Qg4 or Qh5, etc. The "Greek Gift" ). Sometimes it works, sometimes it doesn't. And often calculation is required to try to determine which one it is in a given position.

I think a chess player requires many skills, including knowing when to use calculation or intuition. So, I think for a master player both are very important. He'll use better intuition and calculation in comparison to a novice.

Originally posted by Varenka
I agree. In some positions, intuition,etc. will play a significant role since, as you suggest, it may be too complicated to calculate all the subsequent variations.

However, at other times, calculation will be critical. Although pattern recognition is an essential aid, many positions are too specific to rely upon our intuition alone. For example, consi ...[text shortened]... h are very important. He'll use better intuition and calculation in comparison to a novice.

GMs normally do 60% judgement 40% caculation.

Very interesting post. I think the analogy is quite good. I think you are right when you say it's very hard to have think-in-steps kind of thought-process. It's just unnatural and gets in the way. If there should be one, the thought-process must be natural and pick the best out of the natural strengths of the human brain, not trying to structure them in some un-scientific way.. the brain thinks best for itself =)

I also think that calculation is very important. But if you don't know where to look, naturally a lot calculation will be wasted. I once read about a game where a GM said he only looked two moves forward all the time, and he was still beating a high-rated player. I've also read these articles of GM's speaking as they play, doing many errors in calculational tasks, yet playing very good.

Those who have it all will be the strongest.. but (A) a good understanding of the problems at the board and (B) fantasy/intuition/searching for move ideas and (C) calculating correctly are probably all important. Sometimes more, sometimes less, depending on the situation.