Will computers ever solve chess?

btickler, I get exponents just fine.  Do you understand that computing power has been increasing exponentially?  So for the number that represents our current computing potential, the exponent is increasing at a linear rate.  Hence it won't take all that long to reach 10^44, provided it is possible at all.

So the question is whether or not it would ever be possible, given any conceivable technology and any amount of time, to search through 10^44 positions.  You say that is impossible "with today's technology and even tomorrow's imagined technology".  For today's technology, of course; no one is disagreeing with that.  For imagined technology, what makes it impossible?  I can certainly imagine a computer that performs 10^35 operations per second or more.  What we would need would be some hard physical limit that proclaims that it is impossible for a computers to get much faster than they are now.  But as far as I know, there is no such known limit.  Other than that, we just have our intuitions, which really have no bearing on what we can achieve thousands or millions of years in the future, and would probably just lead us astray.

For the second point, you say that cutting 99.9% of positions is surely impossible - actually we can cut much more than that.  Alpha-beta pruning can theoretically cut the search space by a factor of the square root of the search space.  In practice, we can't do quite so well.  For example, in 2007 checkers was solved by searching 10^14 of a possible 10^21 positions, so they were able to cut the exponent to 2/3.  If we can achieve this 2/3 for chess, then instead of 10^44 positions we would need to search 10^29 - 10^30 positions.  Yes, this is infeasible with today's technology.  But if you give me all of eternity?  I have to bet that it will be done eventually.

Not remotely true.  Maybe you should read this 57-page thread.

My "input"...(get it? Input?!...oh nevermind)....would be, why would you want to?

Say if they somehow find the time to map chess from beginning to end, really what be the point playing it?

You can feel that way all you want to, just as long as you realize that all the facts say otherwise .

Chess will be solved. I agree that it isn't possible today. The Utah Data Center has an exabyte and possibly a zettabyte (10^21) data storage capacity. While this may be enough to record every digital byte of information on the planet (phone calls, videos, emails, etc) it is a far cry from 10^46, which is approximately 1/2 the number of atoms in the known universe. Still if we can calculate the number of possible moves in chess, then it should stand to reason that our fastest computer, at 10 quadrillion "floating operation points per second", could "solve many openings from the 6th move. Just as checkers was solved by shortcuts, the same may be done with Chess. In a sense, it would be like reverse engineering Chess. You solve one particular opening strategy at a time, starting well into the lines already established. Granted, it would destroy the game little by little. Players would opt not to play the "solved" lines and choose others, until there are no more to choose from. As was mentioned before, the priority of the supercomputers is not to solve the game of Chess. So it may take until the year 2525.

I agree with you, and I wouldn't bet my life on it either. My only comment about the Shannon number (or the estimate of 10^46.7 positions) is that they include positions from ridiculous games.

If you study games that are "well-played" to look for a possible forced-mate, then the number is much much smaller. The shannon number (10^120) is theoretically interesting, but it includes ridiculous games which makes even the most amateurish games on these forums look like brilliancies.

The answer to the question posed is "no".

"If you study games that are "well-played" to look for a possible forced-mate, then the number is much much smaller."

Once again, with feeling.  Even if you eliminated 99.9% of moves as being "ridiculous", you would only reduce the number to 10^43, which is a thousand times smaller, but still unbelievably hugely wildly larger than any number computers will be able to number crunch.

More exponential non-understanding:

"it is a far cry from 10^46, which is approximately 1/2 the number of atoms in the known universe."

No, no it isn't.  You can't add/subtract exponents like that.  10^46 is not "approx. half" of 10^79...the latter is 1000000000000000000000000000000000 times larger.

But remember, you don't need to examine the entire game tree to find one forced mate. If you find one forced mate, then the rest of the calculation can be ignored.

With this in mind, 10^46 (or any other number on this thread) has little meaning.

It's not unreasonable to say that solving chess (proving it's a win for one side or the other) can happen in our lifetime.

Checkers has been solved. Chess is a big leap, but it could happen within a decade or two.

This only has little meaning if you don't understand basic math.

I already pointed out the odds, even allowing for you to eliminate a whopping 99.9% of the tree:  1 in a billion billion billion billion to find your one forced mate.  If you choose to call this chance "reasonable", then I guess we're at an impasse ...

If you dispute the number, tell everyone a better number, and show your work.

The only thing which stops Chess machines from being infallible is a shortage of silicon and time.

The only restraints are physical, so the resources are limited, even thouh massive. Infinity is only sufficient. We need even more to really get the job done. 

On the other hand, there may be many forced mates.

It's not "1 in a billion billion billion billion"

It could be "billions and billions of forced mates out of billions and billions of chances"

But both sides of this equation are unknowable, so I'm OK to be an impasses too.

Same. Exact. Problem.

Even if there are billion forced mates, then you still have only 1 in a billion billion billion chance of walking into one.

There's just a host of people on this thread who just have no concept of large numbers past a million/billion/trillion...it's enough to make one weep for future of humanity ...

The chance depends on the ratio of forced mates compared to the total number of games.

Sure.  In the entire history of chess on earth, nobody has run into one yet ...I would say the ratio is pretty damn low.  Maybe, though, every single chess position that has not been tried yet is actually a forced mate.  It could happen, right?  There's a non-zero chance of that, ergo, "reasonable".   

And they wonder why people who barely know the rules cheat?

Winnin for most is what they want, especially in a world which calls morality a character defect, or a sign of weakness.

Keep all your centipawns locked up for safekeeping.

My bad. You are correct and I even knew that. I should have said "Has half the decimal points". Your clarification actually helps my argument, but with such high numbers, the solving of Chess by studying every possible move, would take longer than scientists predict the sun will last. We have to cheat to do it, as we did with Checkers. 

Thanks for that...it's a far cry better than "well, I'm going to ignore all the math, reduce the numbers to a vague billions and billions among billions and billions, conveniently dropping 40+ orders of magnitude on one side of the equation, and then just say it's unknowable on both sides of the equation"...

The game then is solved, except that the numbers are only big enough to beat mere humans.

10^79 is a big number, but apparently there are numbers "a thousand" times bigger than that that have been factorised as semi-primes.

The largest prime number is 2^57,885,161-1, but calculating prime numbers and evaluating chess positions are two completely different animals.  Comparing them would be like saying that because you can cook a 1 minute egg, you can compete with an Iron Chef.