Will computers ever solve chess?

Chess appears to be a draw at the top human level, when neither player makes a mistake.

But take the top human and pit him against a top engine, and suddenly the human's playing, which appeared to be mistake-free, is actually shown to be quite inaccurate.

There are reasons for this. The main reason is that humans play under bad conditions.

The main problem is computers can look at millions of positions in a very short period of time and humans can't.  However if we were to change this by giving humans much more time [say a week for one move or a month for one move] then the best humans would be a lot more comparable to the best computers.  Every grandmaster knows that he makes mistakes when he is subject to time restrictions.  

Which brings us to top engines, which draw a lot of games as well—but perhaps this is a similar phenomenon to top humans playing each other: their level of play is similar, so draws are more likely to occur. not really, if you have two players each rated 1000 you will find a whole lot of wins for either side. The reason is that the higher the skill level--the more draws as long as the skill level is close to equal.

That doesn't necessarily mean that "solved" chess is a draw. It just means that draws happen more when the two players are similar in strength.  But as we know this is not true.  Pit a 800 level player vs a 850 level player and you will not see many draws.

Alpha Zero seems to be the closest any chess-playing entity has come toward perfecting chess—and yet, Stockfish 8 was still able to win 4% of the time (in the 1200 bullet games the two entities played against each other).  Here you are giving a handicap by forcing the computers to make their moves fast. If you gave both computers more time you would not see the 4% rate with Black.

This means that Alpha Zero still has weaknesses in its play (else it would never have lost). It simply means if AZ is forced to play very fast it will have weaknesses in its play  just as humans make more mistakes when they play fast.

By the way, I am not saying AZ is perfect but it is getting closer. 

It also means that, quite likely, there will be an engine (or neural network) that comes along in the not-so-distant future that's strong enough to make even Alpha Zero look like a patzer.

Just like Morphy was heralded as the original grandmaster, AZ is the original self-learning chess AI. But just like has been shown in Morphy's play, AZ's play will eventually show room for improvement.  of course AZ can become stronger.

I don't know if a GM has enough knowledge to give a valuable opinions on the outcome of chess because they lose to computers. But I have given the main reason they lose to computers--the computers have an unfair advantage as they can look at millions of positions in one second. Undo this handicap and humans will play much closer to computers.

Probably create a program and ask a computer like Alpha what the answer is would be a good way. It's either a win or a draw that's for sure. I would say win because white picks the strategy he wants. Possibly when your chess knowledge increases to that of a master or grandmaster you will change your mind?

I emphasize again I never said that it would be a fact that 95% of GMs would sharePonz opinion, neither I said that it would be a fact that thousands of GMs would do so. Ifpatriot is persistantly putting words into my mouth I never said. He has now totally lost moral credibility.

Sure, you can give the human one-month-per move, and the engine one-hour-per-move. The human will certainly play at a much higher level—though they'll still lose. Actually in that situation i think there would be about 95% draws.

Even if, with such a long timer for thinking, the human somehow manages to play at a 3000 level (which is already stretching plausibility), that's still not strong enough to beat a modern engine.  the idea is to draw not to beat. Of course a human would not beat an engine in such a set up.

On an hour-per-move timer, SF9 (and AZ, for that matter) would be playing at a 3500 level or higher.

A human player, even playing at an unlikely 3000 level, would still get crushed. The human would need to reach a 3500+ level themselves to win (which simply won't happen unless the human is using engine assistance to do so). The human would not win as chess is a draw at high levels. However there would be a whole lot of draws.

Even then, that still doesn't mean that engines are close to solving chess. For all we know, the chess engines and chess-playing AIs of the future might make 3600 look weak in comparison. We currently consider 3600 or so to be approaching "perfect" chess. But it's possible that true chess mastery is closer to 4000. Or perhaps even higher. i consider 3600 level is APPROLACHING perfect chess. And also agree that perfect chess is probably around the 4000 level.

I do not think engines are close to solving chess either. Because "solving" seems to mean looking at ALL the chess moves possible and humans may die out before that happens.

However, even in his book, he could give no line where White had a win from the first move. 

 Uh...that’s because no one, master, GM, Super-GM, engine, AZ, you name it, don’t even come close to that, black blunders not included. It’s not that there’s something wrong with that particular GM, but it’s just not possible, with the actual tools we have at our disposal. 

Duh...It’s impossible for anyone, at this moment in time. Nothing to do with him. Nobody knows either way, win, loss or draw. Nobody knows best answers either way, nobody knows the theoretical result of the game either way, and nobody can use one unknown in order to prove that the other unknown can be known, so that they can prove something which they already posed as a known: the first unknown.

  They are committing the ‘sin’ of starting with the problem already solved, while pretending to try to solve it by logic. They are simply begging the question and not realizing it.

Nobody can "know" in the sense that chess is "solved". Because, of course, chess is not "solved" and may never be "solved"

However someone can be more than 99% certain that chess is a draw  because of a whole lot of evidence. Many people who are more than 99% certain of something [because of all the evidence] can say they know.

I am certain chess is a draw. I "know" chess is a draw.

And, by the way, i did not start out assuming chess is a draw. When i was 8 years old and my dad beat me 100 games in a row--i did not assume chess was a draw. It was only after i discovered a ton of evidence--did I say chess is a draw. [most of the evidence you do not even mention] 

  That is not evidence. You seem to have a reading comprehenion problem. Nothing in the above did I say was "evidence".

It is pointed out every time to you that you are clueless when it comes to best moves,Actually the truth is i have found many best moves in my life and some of the best moves were moves the best chess engines could not find.

which would indicate the perfect games. There are very probably millions of perfect games and quite often 10 or more best moves for each position.

So to prove that they’re perfect you assume two things: one, that you know that chess is a draw from other evidence that doesn’t exist.  Actually it does exist and i have given much of the evidence on these forums.

  And two, that you can detect the moves that lead to a draw and avoid a loss, which you cannot know unless you analyze all the possible moves. This is totally untrue and gives an indication that you do not understand chess--to prove this i will give a position with more than 10 best moves. And i do not need to analyze the more than 10 best moves that lead to a draw and avoid a loss.

Which you cannot because no one can, man or machine.  Really? i am going to give a position to prove  you are wrong.

 And if you cannot do that, you have no idea how the games will eventually end, which invalidates this ‘evidence’ which you bring in constantly. You have not repeated my evidence. Sometimes you give distortions such as in this post where you state i give evidence when i have not.

You never address the crookedness of this evidence, directly, but instead mention ‘other evidence’ that doesn’t exist. I have posted the other evidence so it does exist--it is just that you ignore the other evidence.

 If you don’t address the lack of logic in this ‘evidence’, please stop bringing it up. It has been refuted numerous times. no, not even once. Here is what you often do. I give 5 pieces of evidence. One piece of evidence in itself does not prove anything much. So you give one piece of my  evidence and distort it and try to say it is no relevant. and then  you ignore the other 4 pieces of evidence. and for sure you ignore the more than a dozen pieces of evidenced i a hve given.

  Bring something new, if you have.  Why should i bring something new when you have ignored most of the evidence i  have already given? 

No one’s interested in how strongly you believe that chess is a draw. And don’t equate your belief with knowing something for a fact. The word "fact" has different meanings. Many people state they know something is a "fact" when they are very confident they are correct.

Chess will probably not be solved during our lifetime: that doesn’t mean one can go on and speculate about its end result with best moves on both sides and then present that speculation as a fact. Here is the difference. I do not have 99% certainty  that chess will not be solved in the next 20 years. And my guess is that it may not be solved before and if humans die out.

However, regarding chess is a draw, to me, that is a fact as there is a ton of evidence [which  you ignore and distort] that chess is a draw.

OK here is a position where i can detect moves which lead to a draw and avoid a lost [something you said could not be done]

  But most cannot bear the unknown and jump into the ocean of speculations. It gives them a false sense of security.