# Pattern Recognition vs. Calculation

When I started playing chess, long ago, it was all about calculation. Nowadays, it's all about pattern recognition.

Have you played a lot of tactics trainer here? If so, have you noticed that only if you've played the tactics exhaustively and they start to repeat that you are calculating more than recognizing patterns?

And in real game play...how many times have you said: "Oh, I've seen that exact same pattern on tactics trainer."

Mark Twain said that history doesn't repeat...it rhymes. Seems to me this is also true of chess.

I might see, for example a smothered mate candidate move somewhat similar to one I practiced on a tactics trainer. That does help. But...not all that much.

I still find myself calculating. Looking at up to three...pushing the envelope to four..."moves ahead" and weighing all the opponent's options.

I think this "pattern recognition" thing is over hyped. What do you think?

I find that the more TT I do...the less calculating I do and on many problems I just go into-"auto pilot" mode for at least the first move...then "feeling" my way through the problem instead of calculating everything from the start.so yeah....I think it's more about pattern recognition than brute calculating.

Pattern recognition doesn't replace calculation. Anyone who has said it does is lying.

Rather, being able to recognize common patterns makes calculating more efficient as you are more likely to recognize the salient features of a position faster if you are intimately familiar with the pattern present.

If you've done enough tactics problems, you'll be aware that a bishop and rook can be used to create a back rank mating opportunity with the bishop cutting off the escape square in front of the king and protecting the rook right next to the king.

If you know that pattern, then a position such as the following:

becaomes trivial to see in a real game. And it's the "real game" aspect that is important because on each move no one is there saying "black to move and win."

Rather, the person playing has to spot the potential in the position, then calculate to see if the idea will really work or not.

Learning patterns isn't about if the exact same pattern shows up in your game or not. Learning patterns is about being able to see similar ideas in your own games so as not to squander opportunity when it arises.

In other words, if you know the pattern, you'll see the tactic. If you don't, very often the tactic will go un-played. Not because you couldn't have calculated out the result had you seen it, but because you aren't likely to have calculated the right line.

Someone who doesn't know the bishop-rook pattern above is very unlikely to calculate out .... Bh3! because they just see that it loses a queen immediately. But someone who knows the pattern will realize that the queen isn't the important feature of the position and will look a bit deeper.

Pattern recognition is not a substitute for calculation but it helps generating candidate moves and directing your calculations.

Tactics trainer is not the best tool to build patterns, because the problems are so different, but it's a test of your current pattern bank and calculation speed.

If your TT training doesn't seem to help, you need to work on acquiring patterns with other tools (books, set of exercises classified by theme, etc.)

No, believe me it helps. That's not what I'm saying.

I think that what Kingpatzer is saying is "right on the money". And, what you've said, also: Pattern recognition is not a substitute for calculation but it helps generating candidate moves and directing your calculations.

You guys are affirming some of my thoughts. I've been working quite diligently to be a pattern recognizer more than a calculator. To a good extent, I have greatly benefited from using tactical trainers, here and elsewhere.

BUT...I still need to calculate. Some of the posts that I've read seem to suggest that calculation is just a minor aspect of chess play.

Now, if this was true then why wouldn't tournament play, for example, be nothing but blitz instead of 40 moves in two hours?

My experience and thoughts are in very much line with Kingpatzer's comment above.  I find that about half the tactics I've done are very much pattern-based, and those are great to recognize during a game. (Although I have yet to sac my queen for a back-rank smothered mate in a real game.) But also, the other half basically get me to think outside the box I would normally think in, and say "what if I sac this piece, just for the heck of it? Hey look, it's mate-in-two!". This develops the habit of always looking for more to calculate in a position, even if there's no obvious pattern.

Max Euwe once said that Tactics is seeing and Strategy is thinking.  Petrosian said that if you don't see a tactical idea in the first minute, you probably won't see it.  Calculation is important when you have the time and mental reserves.  You need to be selective about what you calculate since you have limited time to calculate.  It is a waste of time to calculate random moves if you don't see a pattern or have an idea.  One of the greatest benefits of good pattern recognition, is checking to see if your candidate move is safe.  If I look at a possible candidate move and recognize a tactical possibility that my opponent might use, I am not going to waste precious time calculating if I can find a safer move that accomplishes the same objectives for me.  Besides the time element, endless calculation fatigues your mind and leads to blunders later on.  Calculation is important, but careful selection on what to calculate is necessary.

Metastable:  ...the other half basically get me to think outside the box...

Excellent point. Even though I don't ever seem to encounter the exact same tactic in a real game, I see enough things to correlate my thoughts into a calculation that finds the solution. Sometimes.

In this way, using a tactical trainer is akin to a boxer practicing on a speed bag. Helps with the coordination, reflexes and timing...not exactly what might be actually used in the ring.

I would add that sometimes attempting to replicate a pattern without due calculation can result in disaster. There might be a nuance that needs to be taken into consideration...and calculated.

These are all reasons why it seems to me that these "pattern recognition is everything" postings are egregiously lacking.

For awhile I started to wonder if maybe I'm the oddball. So many players seem to say that you can be a non-thinking idiot and still a very competent chess player just from pattern recognition.

Blake, you also make some excellent points along the same lines.

You know, yesterday I was blundering all over the place. Today, I've gotten my mojo back. I think my board/piece perception was off. The foreseen bishop lurking on the extreme edge of the board, away from the action...it would later sweep down and snarf up one of my pieces. This sort of thing. The kind where you slap yourself on the forehead like Kasparov. You know better than to allow such things to happen. Somebody else must have taken over your body.

That and focusing perhaps too much on the calculations instead of the Zen-like flow from the tactical trainings.

I've gotta add another point that comes to mind...

I started playing chess long before computers. Back then, you'd get a chess book of puzzles and you'd set up your board and play the position out. Maybe you could do five or so in an hour.

Nowadays, with a computerized tactical trainer, it's possible to do 50 puzzles in an hour. This is an order of magnitude increase.

This is truly great. Only thing is...I wish that I could also get an order of magnitude improvement along with all this convenience.

Chess is a demanding goddess who still makes you sweat for a living.

"BUT...I still need to calculate. Some of the posts that I've read seem to suggest that calculation is just a minor aspect of chess play."

I very much doubt that :all strong players are fairly good at calculating, and I think bad calculation ability seriously hampers your progress as a player.

What I've observed is that the stronger the player gets (above 2000 elo), the less often they will try to 'calculate out' the position. That means that in most cases, the strong player will pick a 'natural move' (for him), and only use casual calculations to safety-check things. However, they will revert to strong calculations when the position requires it, but it may be only a couple of times during the game.

Let me give an example to illustrate my point from one of my games :

In the above position, I calculated many variations, trying to find the best way to use my Knights to restrict my opponent's pieces. However, my coach immediately pointed that ...Nxd4 and Nc5 was a simple solution with a technically winning position : no need to calculate when your evaluation is strong enough

Good thoughts, Paul.

Here's another thought on the subject matter...

Repetition can be a good thing in learning a skill. But one must be careful. What I mean, for example, is doing a lot of tactical training. I've been doing that. I can recognize the threats and potentials much more quickly than I could before this training.

I used to prefer a one hour game. Later a thirty minute game. And, nowadays a 10 to 15 minute game. These shorter time spans have come about from doing a lot of tactical problems.

And, has been pointed out above (and elsewhere), if you can't see the threats/attacks fairly quickly...perhaps within a minute (and I'm not talking about the correspondence chess player), then you probably won't.

Yet, a "good thing" can be overdone. Although I will play rapid chess...I shun blitz and bullet. I am not so good that I can play with hardly a moment of thinking/calculation.

I know that really high level players are that good. But, for mere mortals, I have seen too many get caught up in playing knee-jerk moves without even realizing that there is a place for calculations...and that place is not an insignificant one that can be packaged into one-tenth of a second.

Saying that, I won 3-0 against the founder of our local chess club (now defunct)...30 minute games. He then wanted to play blitz. I declined. I do think that he probably would have won. That wouldn't have bothered me. What would have bothered me is to play fast and sloppy just for the sake of doing so and voiding the need for any serious calculation.

Everybody seems to love blitz, though.

Laurent: What I've observed is that the stronger the player gets (above 2000 elo), the less often they will try to 'calculate out' the position. That means that in most cases, the strong player will pick a 'natural move' (for him), and only use casual calculations to safety-check things. However, they will revert to strong calculations when the position requires it, but it may be only a couple of times during the game.

Certainly. I know that the better I get, the quicker I get at perceiving situations and moves. If I can continue to make progress...slow as it is...I know that my time at assessing and then moving will shorten further.

Up to a point, I suspect, then there is "the point of diminishing returns"...one of the restrictions that the chess goddess imposes on mere mortals.

Blitz is fun to me, especially after a tournament, because you see all these ideas and candidate moves as you play, but you carefully verify each of them in a long game.  In blitz you can see something that looks interesting and just go for it without worrying about the "real" consequences, so to speak.

If it takes you 20, 30, or 60 seconds to find a few candidate moves, then I could see how blitz would be super sloppy.

But also in blitz you make good use of schematic thinking as well.  So you plan out a 2-3 move sequence to get to a position you want, and that's one reason some moves are played nearly instantly, and can be a lot of fun.

e4nf3 wrote:

Laurent: What I've observed is that the stronger the player gets (above 2000 elo), the less often they will try to 'calculate out' the position.

I don't know about that...I calculate now probably at least as much as I ever did.  And look at a player like Nunn (lol).

Waffle: If it takes you 20, 30, or 60 seconds to find a few candidate moves, then I could see how blitz would be super sloppy.

Yeah it does...not on all moves, of course. If I did nothing but play chess, maybe I could hone that down.

But...you know...Clint: "A man's got to know his limitations." Concerning speed chess, I do. Many don't. Nor do they care. It's fun. For me, playing fast/sloppy and losing by doing so...well...not what I call fun.

Yeah, I was playing some blitz a few days ago and I'd get a strong position, but only have 30 seconds left, so I'd play really really fast to not lose on time, and now my position is worse, but he loses on time.  Not very satisfying.

So rematch and now I don't want to get so low on time, and I play really fast and sloppy the whole game.  I'm probably a bit worse the whole time but I win on time.  Again not very satisfying.

I am going to post a unusual chess problem, and could you tell me if you used

# Pattern Recognition, Calculation or both.

The answer to come soon!!!
CHCL wrote:

I am going to post a unusual chess problem, and could you tell me if you used

Pattern Recognition, Calculation or both.

Both obviously :p  Unless you're a very strong player / familiar with a similar pattern.

I tried Rxb5 with Qf2 thinking I had a double attack (on b2 behind the rook, and of course the rook itself).  And if rook takes queen I thought I had the obvious mate starting Be5, but RxQ gives additional cover to f6, so I didn't solve it.

So in this case seems I used too much pattern recognition and not enough calculation.  I assumed the deflection was valid without calculating to the end.  I'm pleased that the correct answer is similar to my idea, but of course works to the end as well :)

In a tournament game by the way, I'd be sure to verify it to the end with calculation, before sacing a queen like I'd planned to that is :)

Also I didn't even bother to verify if Raa2 was an answer for black, so I really fell short on the calculation lol.  Ok, so mark me down for way too much pattern thinking on this one.