This is a question that I've often thought about, especially when I hear someone mention, colloquially, "(some) computer engine beat (some) grandmaster" and etc. Many of us are familiar with the achievements of computer chess engines. But do these engines play chess? I tend to think that computers do not; they are merely instruments of abstract symbol manipulation that we interpret as a game of chess.

Any thoughts?

Off-the-shelf chess engines are traditionally fast chess move calculators with analysis capabilities of upto 21 moves deep and access to opening books with millions of games.

I am unsure if that is what it takes for long term strategic planning, but maybe someone else can enlighten us on what it takes for a human to do so (minus the imagination, gut feeling or instinct).

They play chess as much as calculators do math.  There's an input, and an output with programed rules in between.  So I guess it depends on your perspective.

They don't "play" the same way humans do though, they number crunch, looking at literally millions upon millions of irrelevant positions and moves to find some good ones... but they do it so fast... is it stupid or just thorough?  Probably both, which is why they play so well, and why we wonder things like this from time to time.

One interesting point is how much better some people are using these tools than others.  Those 1300 rated amateurs that won that anything goes cc tournament (whenever that was, whatever it was called...) obviously knew how to get the most out of their machines despite their low rating (they beat out teams with titled players).

I would say tha yes, computers play chess.  They look further into the game, see what moves would end with what results under a certain number of possiblities.  In a game where there was virtually no time limit a human could do the same, but the only difference in a computer and a human is the computer does in seconds (or less) what it would take a human several minutes (at best) to do. (I am going under the impressioin of several moves into the future.

No, the way humans and computers play chess is fundamentally very different.  Computers play (like you said) by looking many moves ahead.  Grandmasters, for example, are able to play simuls (against many amateurs) or incredible time odds, say 1 minute vs 60 (again, against amateurs) and win because they need very little (or no) calculation to play very well.

Against each other in a very sharp position, of course it's necessary to calculate a lot, but this would at most be only 100-200 positions after several minutes, nothing like the millions of positions per second done by an engine.

I agree with Orange. Computers never "play" chess. They simply calculate and select the best moves based on its programming.

Nevertheless, the moves of chess engines are pretty accurate and strong nowadays that the fun is taken away from the play.

You chaps do understand that all your personal and financial data  is on computers, right?  I'd be rather careful of disparaging their abilities if I were you.  Nice bank account you have there, be a shame if anything were to happen to it - IYKWIMAITTYD.

Me, I love computers, and the people who run them.  And the government.  And Google.  Come to think of it, I love almost everybody!

I don't think anybody here is disparaging a computer's abilities. They're just saying that a computer solves problems differently than a human.

An example: What is 2 + 2?
I don't need to add anything to reach the answer because I memorized the pattern a long time ago. A computer finds the answer by actually adding the numbers together each and every time it solves the problem. It never memorizes the pattern.

That's the same way a computer plays chess. It never memorizes the patterns. It doesn't have any idea what might be good moves or bad moves until it goes through all its program's gyrations.

I didn't know computers add numbers. I thought it memorized patterns.

They actually add much the way we do except they only have two numbers, 0 and 1. To a computer, adding 2 and 2 works like this:

Two registers (special types of memory locations) inside a processor each contain a binary 2 (0010). When a 1 is added to a 1 the registers carry a 1 to the next bit in the register, much the way we carry a 1 to the next decimal place when we add 5 and 5.

   0010    (=2)
+ 0010    (=2)
------------------
   0100    (=4)  

The difference between us and computers is that we can learn that 2 + 2 = 4 without going through the process of addition. The computer's registers must always perform the addition process because they can't memorize the answer ,or pattern.

But couldn't the computer be programmed to recognize "2+2=4", as a complete sentence? In which case, is the computer adding? When "2+2=..." occurs, doesn't the computer complete the sentence with, "4", without having to go through the process of adding?

Yeap . . .come to think of it, psycology governs this issue about computers.

Computers are such a wonder that we have come to consider them as super humans in the same (dangerously) way in which we are coming to believe that Internet is the "non plus ultra" means of knowledge (we are placing apart the books)

If we could put a chess computer and a human to play at the same velocity of thinking, the computer will (almost) always win. . .Why? Because computers can think of thousands of possible moves by the time you can consider some ten or twelve moves. ¡Amazing. . .eh! Of course, assumig the chess program was written by outstanding chess players. . .

Computer chess programs are writen by humans. GM usually. They instruct computers about relative values of pieces, moves, false moves, etc. Once this is established the computer can start to consider diferent combinations and pick the best one(s) according to preset standards.

The issue is that the computer can consider thounsands of combinations in the same time it takes you to consider ten combinations.

Do you know how the levels of difficulty are set in the chess programs?

Well, the level is defined according to the time allowed the computer "to think". A normal level is about 6 seconds, a very high level could be some 70 seconds…Now, imagine a computer with a 2 minutes thinking time.

And take a look at this.

It is estimated that the possible number of combinations for the first ten moves is about

165.518.829.100.544.000.000.000.000.

So, the computer does not need to know whether you prefer 0-0 or 0-0-0 or if you are one of those that resign on losing the queen. . . It plays what is best at the moment. . . and, of course, having so many moves to choose, obviously, chances are that it picks the best.

PD

--The only great player that accepted computers was Botvinik. I think it was because he was an electrical engineer.

--Some body said tat a rather "fast" computer can evaluate a million moves in a few seconds. It would take it 7 billions years to evaluate all possible combinations. . . much more time than the age of the Universe as we learn today.

¡And we all are amazed if we can solve a chess "mate in two" puzzle in one minute. . .!

Yes, computers play chess....in a different manner than humans. They aren't making pancakes.

Similarly, airplanes fly...in a different manner than birds. People seldom ask. "Do airplanes really fly?"

This difference makes the computer analysis of my games frustrating. It never seems to understand my plan or the position I was trying to reach. It doesn't seem to adequately explain why it's plan was better other than an often, a small material gain.

Conversely, the computer analysis may think a move was fine but another player can instantly recognize I planned to push pawns queen side and so blocking a pawn with a piece may not have been the best idea.

I'm not sure I can adequately answer that question. I'm not an expert programmer and there are lots of tricks programmers use to speed up the way computers do things.

Humans memorize by doing something repetitively until physical changes take place in our brains. Computers can't make physical changes to themselves (yet). Computers could be built so that a part of their hardware will spit out 4 whenever they're told to add 2 and 2. The problem with that is there would have to be a separate piece of hardware for every function and number combination you'd want to perform. That would make the computer very large, very quickly.

Another way to make it seem a computer has memorized something is to store answers in a table and program the computer how to look them up. That's basically how opening databases work. It runs the current board position through a process that comes up with a unique number. It uses that number to look up the correct move(s) that's stored in a table. That would be very inefficient for simple addition but can greatly speed up complex functions. That's why a computer chess program slows way down as soon as it gets out of book. It has to calculate everything to find the best move instead of just looking up the answer in a tablebase.

 As I said, there are probably a number of other ways to speed up computers or make them seem to know things they don't but they can never really memorize a result.

(edited for readability)

Pure question of semantics.

That statement usually ends these types of arguments and philosophical debates. 

This is one of the coolest debates I've seen. The way I look at it, the computer isn't playing chess like we do because it has no emotion. We have our styles that we prefer and influence our decisions. Like how I prefer knights over bishops. A computer is more likely to win but it gets no satisfaction from it, so I wouldn't want to be a computer.

Yes.

so if someone with a photographic memory (sort of thing) just memorized tons of games and positions, would it be considered that they are 'playing' chess or not? what if they did make moves outside of the games the memorized, but it was always based on the outcomes of those memorized games/positions?