If you give computer and best human chess player infinite time and you put them to a chess match who will be victorious...

how many yottabytes would it take to generate the whole chess game tree before move 1?

Been there done that.
en.wikipedia.org/wiki/Human–computer_chess_matches#Kasparov_–_Deep_Blue_(1996–1997)
With Machine Learning, things may be different after a few 1000 matches
So, you're saying...
the only winning move is not to play

Attached: 1514003935554.jpg (3553x1275, 1.27M)

Indeed, professor falken

Humans players may too, with infinite time.

Attached: 1532088951130.jpg (460x460, 57K)

>If you give computer and best human chess player infinite time and you put them to a chess match who will be victorious. If drawn game they keep playing new games until someone wins
The answer is unknown and most likely will never be known.

My wild guess is that perfect play draws the game, given infinite time you can figure that out easily though.

Serious question, could you make a steam computer that plays chess and how big would it have to be?

Attached: Henscel locomobile.jpg (851x554, 93K)

>could you make a steam computer that plays chess
Yes.
Mechanical computing has been done already, from there it is just a matter of scaling it.

>how big would it have to be?
Large, very large.

Depends on the computer algorithm.

Atm I'd say humans, but eventually it'll be computer (once neural networks gets good enough).

>Atm I'd say humans
Aren't computers already BTFOing humans?
Alpha zero is pretty good from what I have seen.

Those are all timed matches

>Those are all timed matches
Obviously yes, but so what.

Anyway, given infinite time both human and computer should be able to figure out perfect play.
So the question is who is the wins, assuming best play.

GM's can still beat ai given enough time?

whoever gets white :^)

>should be
Social studies student detected

>Social studies student detected
I am a student of mathematics though?

>whoever gets white :^)
That's a good guess, but obviously there is no proof.
Personally I would guess it is a draw.

What is lim(x->0) cos(1/x) ?

Given the same version of neural network and infinite time, the human would eventually learn and evolve better than the AI would. That's what the questions is about

>What is lim(x->0) cos(1/x) ?
Doesn't exist, it should be trivial to construct a sequence such that x_n to 0, but cos(1/x_n) is 1 and another one such that it is -1.
Unless you are the mentally ill /sci/ poster.

>Given the same version of neural network and infinite time, the human would eventually learn and evolve better than the AI would
That is assuming that they would never hit perfect play, which is bound to happen given perfect time.
Then it is irrelevant who is human and who is computer.

Short time - human wings, long time - computer wins, that's not even a question.

Perfect play doesn't make any sense. It implies that white always wins, or it implies that there doesn't exist a perfect play to counter the perfect play, which nullifies the concept of a perfect play, or that there needs to be a setup for the perfect play to exist, and thus the previous moves nullifies the idea of a perfect play, since it's no longer perfect.

How large though, give me a ballpark

See

>Perfect play doesn't make any sense. It implies that white always wins, or it implies that there doesn't exist a perfect play to counter the perfect play, which nullifies the concept of a perfect play, or that there needs to be a setup for the perfect play to exist, and thus the previous moves nullifies the idea of a perfect play, since it's no longer perfect.
You are dumb.
Look at other games which already have been solved like tic tac to, perfect play exists there, why not in chess??
This really makes no sense.

> It implies that white always wins, or it implies that there doesn't exist a perfect play to counter the perfect play
Not at all.
It could be the case that whatever white does black has the possibility to force a draw or a win.

Solving chess is a trivial task, given infinite time, in just go through the game tree and look at the results.

>How large though, give me a ballpark
That entirely depends on how good you want it to be.
Probably the size of a couple of football stadiums to beat a new player.

bigger than ur mom

You can still lose with perfect play, just like you said about the whites. Few games are completely fair, but the concept of perfect play doesn't depend on that.

This question is way too complicated.
Is there even such a thing as a best possible move in chess? Furthermore is there such a thing as a best possible first move?

Chess is not a solved game pea brain

>Is there even such a thing as a best possible move in chess? Furthermore is there such a thing as a best possible first move?
Both are unknown and probably will remain unknown.

Yes, but that just means that no one knows what perfect play is.

You can still play perfectly and loose, you understand that right??

Explain to me how the first move of a perfect play may exist, with some sort of logical proof, and I'll believe that it exists

but humans dont have infinite time.
even if immortal people can probably only recall about 200 years of stuff in their mind

With infinite times white wins, the best human chess player could solve chess in infinite time

No you can't, because the perfect play implies that you begin a sequence of moves which is uncounterable by the opponent, which would lead to eventual victory.

And this 'perfect play' concept would need to factor in stalemates and threefold repetition.

That's an interesting question though, could every single chessgame be brought to a draw, if one player aimed at playing it to a draw? Or are some games strictly only losable or winnable?

I think the part about a human having infinite time to reason about a problem is an interesting thought experiment. Can he come to the right conclusion?

Attached: clip.gif (388x480, 42K)

Perfect play consists of a series of moves. On of those moves is the first one. Hurr.

Attached: 545px-Tic-tac-toe-game-tree.svg.png (545x363, 16K)

Just hink about tic tac to.

You represent the game as a tree, perfect play means that you can play in such away that for every move black makes white has a move such that whatever black does white is guaranteed not to loose.
If at any point such a move DOES NOT exist then white never has had a chance and black was ALWAYS guaranteed to win.

Imagine a tree where every node represents one game state, this is some really basic stuff.

Why introduce an infinite amount of time? Giving anyone an infinite amount of time guarantees that it is at least impossible to lose or win for as long as they draw. It does not guarantee who will win first though so your question is totally irrelevant. Why have an infinite amount of time instead of just a few years unless more time guarantees a victory which it doesn't

An infinite amount of time means it's impossible for someone not to win unless any given player is not able to make a move at any given point in time or they keep getting draws, which is what happens without an infinite amount of time anyway? Your question seems irrelevant to me...unless I'm missing something here.

>because the perfect play implies that you begin a sequence of moves which is uncounterable by the opponent, which would lead to eventual victory.
No.
That is not what perfect play means.

>And this 'perfect play' concept would need to factor in stalemates and threefold repetition.
Since drawing is preferable to loosing if you can not force a win but a draw perfect play means forcing a draw.

This is some really basic game theory stuff.

>That's an interesting question though, could every single chessgame be brought to a draw, if one player aimed at playing it to a draw? Or are some games strictly only losable or winnable?
In other words, what happens assuming perfect play. That is exactly the question you are asking.

Why? There are only 20 possible first moves for white, surely the first move alone can't determine the outcome of the game (sutpid first moves that make you loose quickly, aside).

Infinite time in chess terminology means you can take however long you wish until you're satisfied with a move. If you want practical limitations, 1 month per move each.

Your image shows 3 first moves. So which one is THE BEST?

>"here, lemme just use MINE definition of a term instead of the proper one"

Attached: strawman.jpg (349x283, 27K)

Center X guarantees draw or win

Ohhhh okay I see. Well OpenAI beat the best Dota 2 players of all time. It seems to me as though the human brain cannot out do neural networks but it is an interesting question for sure.

What does perfect play mean then? I dont know anything about game theory.

I'm just assuming that the objective of the game is to win, and the perfect game is where you do everything perfectly, making it impossible to lose.

How is drawing perfect..? Everything but a win is a lose, anyone who thinks otherwise are fucking losers

>What does perfect play mean then?
Always play the move which doesn't mean you loose, regardless what your opponent plays.
If such a move doesn't exist survive as long as possible.

>I'm just assuming that the objective of the game is to win, and the perfect game is where you do everything perfectly, making it impossible to lose.
It might have been impossible from the start for you to win, unless the other player makes mistakes.
So perfect play means either forcing a draw, or if that isn't possible make as many moves as possible.

>How is drawing perfect..
It is the best thing possible, unless your opponent makes mistakes.

>Everything but a win is a lose, anyone who thinks otherwise are fucking losers
This is about game theory not your life philosophy.

Even losing can be perfect in a game rigged against you, if you minimized the losses. It depends on the rules of the game.

>infinite time
Shit would be over in under an hour, computer wins.

So, in a game theory world of chess, the perfect game is a game where both players just dodge the threefold repetition over and over again. Both agreeing that no one will ever win, and they don't need to, since they're both playing perfectly, and will do so indefinitely. Right? What's the incentive to win if you only aim to make the game theory's version of a perfect play?

[citation needed]

>It is the best thing possible, unless your opponent makes mistakes.
What is the incentive to play anything if the best thing is to draw? This game theory shit sounds like some socialist utopia... Especially since drawing doesn't actually exist in the real world no more than 1 exists (1 = 0.999...)

It reminds me of this though:
youtube.com/watch?time_continue=937&v=xOCurBYI_gY

At the end the AI is playing Tetris, and it figures out that it cannot win. So what it does is that it simply pauses the game indefinitely. Is that the perfect move, according to game theory? To stop playing?

Fucking wat m8.
X in a corner is a guaranteed win or draw

X in corner, O in center
X is fucked

How is X fucked?
At worst X draws

no u

that's so not Jow Forums

> calculate all the possible outcomes of its move, as well as any possible outcomes of the human's any move that they take.
>The computer literally won the game before the first move was even taken.
There are more chess positions than atoms in the universe bro. Computers are stupidly powerful but they can't quite calculate literally every single chess move.