Will computers ever solve chess?

  ‘ I don’t have the proof’...but ‘ I do have the proof’...

   Still dishonestly presenting his opinion as a fact.

  Of course, an engine can find a short mating net in the opening, but preceding that net, at least one side blundered, and we’re supposed to be mostly talking about perfect games from start to finish. Not blunder first, then the smarty pants finds a perfect move, or a short string of perfect moves—when everything was lost anyway. Maybe the smarty pants wants a prize as well, for finding a perfect move/string of moves after being handed the gift of a gross blunder.

Haha!

For example, lets imagine one black and one white king on a board. How many unique ways can they be placed?

There are 64 places to put the first king, and for each of those 63 ways to place the 2nd king. So I assume 64x63=4032 unique positions.

Anyway, a tablebase has every single possible position stored and solved all the way to mate (if there is a mate) and if there's not, then it's a draw.

---

"So what?" You might ask. Well for trivial positions, you're right, it doesn't matter, but if you start from just 3 pieces and work your way up, and if you literally store every single possible position, then you can end up solving some ridiculous positions.

Like this position.

White to move has a forced mate in 549 moves. An engine would never solve this.

All positions with 6 pieces or less have been completely solved (kings are counted)

Some positions with 7 pieces have been solved (or I wonder if they've finished all by now?) But with 7 we're already reaching our limits because this over 100 terribyes IIRC, and from 7 to 8 will of course be an exponential increase.

Oh, was he making fun of me?

I really thought he might not know because of some posts he'd written earlier when we were arguing. Like they might not really play perfectly.

I quoted the definition of "conjecture" exactly from the dictionary. I cannot help it if another dictionary gives something different.

An opinion can also be a fact. Per the dictionary "opinion" is a view, judgment or appraisal formed in the mind about a particular matter.

Example  Trump says he won the popular vote. It is my opinion that he did not win the popular vote. Actually, it is a fact that Trump did not win the popular vote. So what is my opinion is also a fact.

Many opinions are made without having ALL the facts. There is nothing wrong with having an opinion on something without having ALL the facts.

In fact it is often a necessity to have an opinion without having all the facts. In war most decisions are made without having all the facts and it would be harmful to do nothing because not all the facts are known. 

Many people weigh the evidence before coming to an opinion--however often they do not know ALL the evidence.

Is his point merely that:

1) Chess isn't solved yet
2) For the vast majority of positions, we can't mathematically prove which moves are best

Because those points seem trivially true.

---

In any case, it's something of a pet peeve of mine when people pretend humans know nothing. Usually it's in the context of science. They will say science only gets it 99% right in _____ problem, therefore humans have it 100% wrong (and what follows is some rubbish about the only path is through God, all knoweldge is vanity, blah blah blah).

So I'm not just being a pedantic prick for no reason (well, I'll do that too). What I want to stress is that humans have a very good understanding of chess. Not only did we find perfect play on our own (philidor endgame positions are more than just the famous rook one) but we've also built machines that can find perfect moves. We can verify this with tablebases (and when there are forced mates).

No I wasn’t. I’m really asking.

Well, I hope my answer wasn't too convoluted or rambling.

If it was then you can use google

Often when someone says "we" know little or nothing about chess--that person knows little about chess.

When a chess engine finds a perfect move IT IS NOT BY DUMB ACCIDENT.

When someone says a chess engine cannot find a perfect move except by pure accident--there is a lot he does not know about chess and also chess engines.

You would THINK it’s trivial, but we are having this lengthy debate. The only sufficient proof - to justify calling it “proof”, is a mathematical one. 

No, you explained it well 👍

Or it could just be you lack the compression ability to understand what they’re saying. 

I agree.

And @troy7915 I guess I might be a bit provocative when I say "engines can find perfect moves." Of course we can only know this is true for some positions.

Or it could be that the person making the statements above regarding humans know little or nothing about chess and/or a chess engine finds a perfect move by dumb accident   could just be wrong. 

Yeah, it's not dumb luck.

I'll say again that, elementally, checkmate is just mobility (or lack thereof).

At the foundation of human-created strategy is mobility. Hell, the relative values are based off mobility. It's common for a beginner's class to place a piece on an empty board and note how many square they control.

Queen controls the most? Worth the most. That easy.

Knight not as much as bishops, but they can hop over others and visit every square.

This is real knowledge. Piece values have been statistically derived (kauffman) and nearly match the classic 9,5,3,3,1. Recently alpha zero taught itself completely independently from humans (and to a lesser extent top engines are improved in similar ways i.e. a human isn't telling it how to play better).

And they independently discover human openings. They're not playing 1.Na3. Why? Because of mobility. Because of the center.

All this to say:  our strategic understanding is not imaginary. It would be a mistake to think we (or at least engines) don't find perfect moves much more often than we're able to prove we do.

It seems you, Troy and I are all in agreement.

I'm sure that wont stop us, we'll add another 20 pages of bickering in no time

When Troy says “dumb luck” I’m pretty sure he isn’t suggesting computers just push wood. In positions where computers just simply cannot determine a move is mathematically perfect, it may in fact pick a move (based on programming criteria) that *also* happens to be perfect (luck).

Reminds me of something I heard -- someone making the distinction between good choices and good outcomes.

You can choose to buy food at the grocery store (good choice) and it's hit by a meteor and you die (bad outcome).

You can choose to buy a lottery ticket (bad choice) and win (good outcome).

In this sense I would say engines make good choices, even if sometimes they lose.