Will computers ever solve chess?

Sort:
JubilationTCornpone
Johnkagey wrote:
ptd570 wrote:

Will there ever be a computer strong enough to solve chess to the point where white uses its half tempo advantage to always beat black no matter what moves black plays (in otherwords the same computer can never win with black even after a thousand random games against itself)

 

I beleive one day there will be a computer so strong and so big that it will solve chess completely but perhaps that is 50 or 100 years off, its possible to solve it but we may never see it even in a 100 years

the answer is no.The reason being that there are 20 possible first moves alone meaning the outc ok me can not be certain.

I have certainly been in the "no" camp, and maybe I still am, but I don't think you can say this so clearly.

 

In particular, if you think there are 10^120 possible games, which is more games than atoms in the universe, then that would appear to be a clear "no."

 

But if you realize there actually are only 10^45 positions, which is LESS (not MORE as some have claimed) than the atoms just in the earth, then all of a sudden you don't have a certain "no" anymore.  But you do have something prohibitively hard, and probably beyond practical reach by normal computers, because you simply lack the resources to do it, now and most likely ever.

 

But what Elroch has pointed out to me just today is that I was missing a point about quantum computers, thinking their power scaled as  N^2 when in fact they scale as 2^N (where N is the number of qubits) ... and that really might be a big enough difference to change the answer.

 

Other aspects of the problem might still lead to the answer being "no," but it's much less clear than I thought or than you are making it.

winston_weng
btickler wrote:
vickalan wrote:
btickler wrote:

 

...Go for it.

 

I never said "I think chess will be solved within our lifetimes." Like so much else in your message, you like to invent nonsense, and seem to base your arguments on hyperbole and fantasy.😊

I have to apologize Vickylan...you never did state that explicitly.  You also never said 50 years.  You actually gave a bunch of ridiculous calculations and claimed it could be done in 18 years...check page 61.  You stated on 2 other occasions that chess could be solved in the next couple of decades.  Later you effectively recanted and said 200 years, with ironically even more specious reasoning, on page 109.  Since then you have kept your number to yourself for close to 100 pages.  What that shows anyone paying attention is that (A) you were full of crap to begin with, (B) you later realized you were full of crap but never admitted it, and (C) that now you have given up on even standing behind your retreat to 200 years number.  Because...you know you are wrong.  Just admit it.

I do apologize for giving you credit for 50 years when you believed an even more unreasonable 18-20.

Some of your quotes, in chronological order:

 

"On the other hand, if best play is a draw, chess won't be solved until the entire tree is checked. So we can only expect an answer in our lifetime if one side can force a win. But if perfect play is a draw, we probably won't know it within our lifetime."


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


"As for solving chess, there is some reason to believe it can happen within a decade or two."

 

"So chess might be able to be solved in 18 years."


"These are other advances that will help solve chess, compared to computers in 2007 when checkers was solved. I agree with Campter: Yes - chess may soon be solved."

 

"If I must make a guess, I would say Yes chess will be solved within 200 years or so."

 

The next time you claim "invented nonsense" or hyperbole on whatever subject, hopefully people will remember your blatant falsehood here. 

 

For fun:

You posted your silly tree diagrams on pages 65, 68, 116, 117, 127, and 145.

You posted your triangle diagram on pages 165 and 182.

You posted your Venn diagram on pages 146 and 184.

You posted your moving a sofa animated GIF on page 176...you haven't repeated yourself on that one yet.

Chess.com community guidelines states that you shouldn’t be offensive.

winston_weng
winston_weng wrote:
btickler wrote:
vickalan wrote:
btickler wrote:

 

...Go for it.

 

I never said "I think chess will be solved within our lifetimes." Like so much else in your message, you like to invent nonsense, and seem to base your arguments on hyperbole and fantasy.😊

I have to apologize Vickylan...you never did state that explicitly.  You also never said 50 years.  You actually gave a bunch of ridiculous calculations and claimed it could be done in 18 years...check page 61.  You stated on 2 other occasions that chess could be solved in the next couple of decades.  Later you effectively recanted and said 200 years, with ironically even more specious reasoning, on page 109.  Since then you have kept your number to yourself for close to 100 pages.  What that shows anyone paying attention is that (A) you were full of crap to begin with, (B) you later realized you were full of crap but never admitted it, and (C) that now you have given up on even standing behind your retreat to 200 years number.  Because...you know you are wrong.  Just admit it.

I do apologize for giving you credit for 50 years when you believed an even more unreasonable 18-20.

Some of your quotes, in chronological order:

 

"On the other hand, if best play is a draw, chess won't be solved until the entire tree is checked. So we can only expect an answer in our lifetime if one side can force a win. But if perfect play is a draw, we probably won't know it within our lifetime."


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


"As for solving chess, there is some reason to believe it can happen within a decade or two."

 

"So chess might be able to be solved in 18 years."


"These are other advances that will help solve chess, compared to computers in 2007 when checkers was solved. I agree with Campter: Yes - chess may soon be solved."

 

"If I must make a guess, I would say Yes chess will be solved within 200 years or so."

 

The next time you claim "invented nonsense" or hyperbole on whatever subject, hopefully people will remember your blatant falsehood here. 

 

For fun:

You posted your silly tree diagrams on pages 65, 68, 116, 117, 127, and 145.

You posted your triangle diagram on pages 165 and 182.

You posted your Venn diagram on pages 146 and 184.

You posted your moving a sofa animated GIF on page 176...you haven't repeated yourself on that one yet.

Chess.com community guidelines states that you shouldn’t be offensive.

AND I DON’T WANT A QOUTE FROM BTICKLER. END OF.

Elroch
RCMorea wrote:
Elroch wrote:
RCMorea wrote:
Elroch wrote:
RCMorea wrote:
Flank_Attacks wrote:

.. Obviously, it's speculation ; But since, 'quantum' computers, are still in their development infancy ; Perhaps, there are already, mathematical projections, {or estimates} ; As to how long, it might take, a hypothetical, quantum 'supercomputer' ; Capable, of sorting through, mind-boggling, permutations ; Given that, the 'quantum' approach ; Has supposedly equal, daunting capabilities {!?! }

Quantum computers don't work like that.  They check all permutations at the same time.  A quantum computer with enough power solves chess instantly.  The problem is, "enough power" (meaning enough quantum bits, properly organized and powered) is still on the same order as has already been discussed. 

No, I believe you are wrong here. The computation that can be achieved with a quantum computer is exponential in the number of qubits. So if this resource was used efficiently, a thousand qubit quantum computer might be adequate for chess.

 

I suppose it depends what the exponent is.  If it is 2, then 1,000 only squares to 1,000,000, which is still a small number.

No, you have this upside down. "Exponential in N" means that the correct number is k^N, where N is a 1000 here. You can assume k is 2, but it doesn't matter much: the log to one base is proportional to the log to another. k^1000 is big (I chose 1000 to be generous).

 

Interesting.  I started to write up an example of why you are wrong and proved to myself you are right.  It's 2^N.  In fact, at least at one point, I already knew this but was failing to apply it properly.

So...that would suggest all the positions could be examined within a probably feasible quantum computer.

I'm not sure that changes the answer because I don't actually understand some of the other objections I've come across, especially why quantum computers are said to be bad at problems involving sequence (I could guess that having encoded all these positions they still have to go through them in order, but it would only be a guess).  And would the existence of the machine itself count as a proof, if you couldn't store the output?

For sure, the very existence of quantum mechanics should keep anyone living in this world from being too sure they know the answer to anything.

Solving chess appears highly parallelisable (just very big). For example, start with a quantum computer capable of simultaneously representing all possible legal positions and more (with a quantum computer this may merely mean representing the board in qubits and working with every possible state simulataneoously), Only a couple of hundred bits are needed to represent a board, so a similar number of qubits for a parallel ensemble (plus some extra qubits for whether they are in or out of filters?)

Identify which board positions are decisive in one parallel quantum computation. This provides level zero of a tablebase. [This step already does more computation than the human race has so far with conventional computers].

From this, generate all of the positions which can force a decisive position in one move (by deterministic rules corresponding to the moves of chess backwards). This is basically the same as is done with a conventional tablebase, but (hypothetically) completely in parallel.

Repeat this process a limited number of times (maximum legal length of a game of chess depends on the 50 move rule, and is 8848.5 moves according to one source, which would necessitate 17697 steps, which is not too big.

The table base logically has to include the starting board by that time, so you now know its value, and have solved chess!

Note that there is not at present a quantum computer close to being big enough to represent a chess board as required. I have no idea how practical the computations I describe are, either.

I am not even sure the above is possible with a quantum computer, but it seems plausible.

robertjames_perez

A googol terabyte chess engine can solve all but that is impossible. 

RoyalTriumph

false

vickalan
winston_weng wrote:

...AND I DON’T WANT A QUOTE FROM BTICKLER...

 

LOL, I think I should add that at the end of my posts too!😄

Elroch

Some people don't like to Btickled?

DiogenesDue

I'm sure lots of people would love the ability to speak their minds without response from others.  That is intellectually devoid of value, however.

As for being offensive...like most messageboards/forums, this basically translates to direct bashing ala "u r a moron and a pile of s***".  Pointing out that someone's posts are full of crap doesn't qualify.  You'll notice my account start date...compare that to the endless stream of trolls and sockpuppets that don't last 6 months, then read my posts and their posts, and eventually you will understand the difference.  I don't run around bashing everybody, but I don't suffer fools gladly, either.  I take that back actually...I do suffer fools here constantly.  Everyone on the chess.com does, because its impossible to avoid them with the lax moderation.  What I don't suffer are fools that spout their foolishness everywhere and muddy up everyone else's waters when they can't back it up.

I'm sure Vickylan is not happy that I pointed out his hypocrisy here, but it is what it is.  Don't say it can be done in 18 years with a bunch of faulty logic backing it up and then pretend you never said it, and imply that someone else is lying.  You'll notice he hasn't said a word about it, and is deflecting, as usual.  He knows he's in the wrong here.  What is he going to say?  "I never really felt deep in my heart it would be 18 years/a couple of decades, I just said that off the cuff"?

If there are less positive posts from me than in the past, it's because there are less and less positive things to respond to on the forums here.  Batgirl clearly needs to post more often wink.png.  When a certain online chess site improves their forums to match their online play, this place is toast.  That's the sad truth.

I do have respect for Elroch and RCMorea (and many past posters who got frustrated and moved on from this stream of consciousness blathering long ago), their arguments have some education and rigor behind them.  So I am not implying that everyone in this thread is posting worthless stuff.  I'm just keeping it real and calling out the BS.

fewlio

alpha zero only played 4 hours and beat the best chess engine...it's like a computer writing a program in 4 hours to become best player in the world.  so give it some more time, more than 4 hrs, and it could solve chess.

sloughterchess
fewlio wrote:

alpha zero only played 4 hours and beat the best chess engine...it's like a computer writing a program in 4 hours to become best player in the world.  so give it some more time, more than 4 hrs, and it could solve chess.

 

sloughterchess
sloughterchess wrote:
fewlio wrote:

alpha zero only played 4 hours and beat the best chess engine...it's like a computer writing a program in 4 hours to become best player in the world.  so give it some more time, more than 4 hrs, and it could solve chess.

 

As computers get stronger the draw envelope increases. Typical is an Evans Gambit I had Deep Fritz 14 play both sides. With White Fritz won a pawn in the middlegame and a second pawn in the endgame. Unfortunately it was one of those countless Rook and pawn endings a pawn up that was unwinnable. Since Rook and pawn endings are so common it seems likely that the computer will always steer for one that is dead drawn.

 

 

vickalan
btickler wrote:

I'm just keeping it real and calling out the BS.

Most people here can read for themselves and know the difference between "will be solved" and "can be solved". But this is a good forum for you, because anyone can write whatever they want. It's no surprise you have not contributed to a math or computer science forum, where interaction with mathematicians is required. But since you've never claimed to have a meaningful conclusion, it's no surprise.🤠

pawn8888

Chess is at it's hardest with the first move, then it get's simplified because there are less correct moves to make. The more moves played the fewer moves that are available to make. The first and second moves are the most important it seems logical to say.

DiogenesDue
vickalan wrote:
btickler wrote:

I'm just keeping it real and calling out the BS.

Most people here can read for themselves and know the difference between "will be solved" and "can be solved". But this is a good forum for you, because anyone can write whatever they want. It's no surprise you have not contributed to a math or computer science forum, where interaction with mathematicians is required. But since you've never claimed to have a meaningful conclusion, it's no surprise.🤠

Still dodging.  Almost time for a graph wink.png...

Just be aware that I'm going to post your 18 years and "a couple decades" quotes for the next 200 pages of this thread whenever you deserve it...or, you could pony up, be a man, and say:

 

"Hi, my name is Vickylan.  I used to believe that chess would be solvable in 18-20 years, and I had a lot of reasons why...but I believe that they were wrong and I didn't know what I was talking about, and so I no longer believe it can reasonably happen within two decades...because I think I was wrong.  It happens.  I am far from perfect."

Or, you could say:

"My name is Vickylan and I fully stand behind my 18 years to solve chess analysis."

What you shouldn't do, is continue to evolve into a full fledged weasel by hemming and hawing and then lying when called out on it.

Flank_Attacks

 

personalincredulity.jpg

DiogenesDue

I'll just mark you in the 18 years column then, Flank.  Good luck.

Flank_Attacks

.. Hey ! .. It's Not my doing, that the following, imaged-message, cannot be conveniently seen, without clicking on it ; But since most humans, {regardless, of their dysfunctions}, tend to be curious ; It shouldn't matter, a whole lot !  [ :

 

https://steemit-production-imageproxy-upload.s3.amazonaws.com/DQmSKvduPhbQLgSPsVudhjb43rkEmXWdmsULwmLwXqgk7dA

Flank_Attacks

.. I suppose 'tickler', could've picked, another unsolved problem ; Yet, with increased chances, of Never being, definitively 'solved'; With some more investigation !

.. 1st appearing on 'YT' ..Jan. '2017'

 

Elroch

Fascinating talk!