[Fedoras of Salvation] - White Knights ITT

[Alex]

Talking about chess makes me want to go back to playing for money again.

[Alex]

I think I just like winning money though, so that is probably the actual reason. The chess part is probably irrelevant, other than it makes me look like Lelouch Lamperouge.

[Dechs Kaison]

Playing chess and programming an AI are two very different things.

My point is, the computer doesn't think in any way that the programmer can't. It doesn't know any moves that a human doesn't. It's not a robot that can physically beat the other player.

In my mind, the number one chess player programmed that AI.

The computer doesn't think, period. It just does what it's told to, using an amount
of raw computational power that no human mind has access to, without experiencing fatigue, stress and the likes.

The computer can beat the human player, but it's in no way "smarter" than him - it just wins on raw power.
The guy who programmed the best computer player is not "the number one chess player", he doesn't even need to be a good *player* at all, he's just very skilled at his task, and that task is not playing chess.

The computer analyzes the board, the probable next moves and responses, and decides on the best move to make. The computer isn't doing anything differently than the human player, just better. It can see more moves in advance. It can consider more options and counter options. It spends less time doing all that. It wins. In every measurable way, it absolutely is smarter than the human player.

I remember years ago when people were touting how humans were smarter than computers because they could still beat computers at strategy games like
Chess. That's not the case anymore.

And if you think the programmer for an AI like that isn't intimately familiar with the game, you're out of your mind.

[Dechs Kaison]

On an unrelated note: I'm watching professional League of Legends players on TV. There's a TV channel for this game in Korea.

[Jack]

Holy shit, we're almost at page 1000.

[Alex]

This is page 666 for me.

[RedNihilist]

Playing chess and programming an AI are two very different things.

My point is, the computer doesn't think in any way that the programmer can't. It doesn't know any moves that a human doesn't. It's not a robot that can physically beat the other player.

nIn my mind, the number one chess player programmed that AI.

The computer doesn't think, period. It just does what it's told to, using an amount of raw computational power that no human mind has access to, without experiencing fatigue, stress and the likes.

The computer can beat the human player, but it's in no way "smarter" than him - it just wins on raw power.
The guy who programmed the best computer player is not "the number one chess player", he doesn't even need to be a good *player* at all, he's just very skilled at his task, and that task is not playing chess.

The computer analyzes the board, the probable next moves and responses, and decides on the best move to make. The computer isn't doing anything differently than the human player, just better. It can see more moves in advance. It can consider more options and counter options. It spends less time doing all that. It wins. In every measurable way, it absolutely is smarter than the
human player.

I remember years ago when people were touting how humans were smarter than computers because they could still beat computers at strategy games like Chess. That's not the case anymore.

And if you think the programmer for an AI like that isn't intimately familiar with the game, you're out of your mind.

A computer doesn't think, it just computes (hence, the name).
It's not *smart* by any definition, stop claiming that - it just has more power, and mindlessly does its job as the programmer instructed it.
That's not being *smart*.

Being a good programmer is not an easy task, and programming a good IA is far from being a trivial job.
Still, the skill set involved in programming a top-level IA is quite wide, and while having a full knowledge of the many aspects of the game is certainly required, being a good chess player is far from being mandatory.

[Dechs Kaison]

Can you explain exactly what the difference is between your decision process as a human playing chess and the computer's?

[Platypus]

Khaospawn, here's a new Metallica song to put your hands in the air for:

http://www.youtube.com/watch?v=AiOtev3rEKI

[Lightning_Dolt]

This is page 666 for me.

998 for everyone else

[redthirst]

Talking about chess makes me want to go back to playing for money again.

It makes me want to continue not playing Chess.

I do need to get back into some Warmachine, though.

[Dechs Kaison]

So I'm trying this Hearthstone thing.

Any hints?

[RedNihilist]

Can you explain exactly what the difference is between your decision process as a human playing chess and the computer's?

Let's break a game of chess in three steps:
- Early game: both the players will presumably follow known openings, as soon as the human player does something out of the ordinary we enter mid game.
- Mid game: the human will try to build his board state around concepts like controlling the middle and such, his moves will inherit by his ability to read the next moves and by his inner ability of discarding stupid lines; he will try to keep his decision tree as narrow as possible by cutting useless branches, whild trying to go as deep as he can evaluating the lines his experience tolds him are the best.
The computer will
expand pretty much every branch of the decision tree, trying to calculate every possible outcome; when lines go to deep to be thoroughly calculated, the computer will estimate the last board status using a standard set of rules given to him by the programmer - as soon as this analysis is complete, it will choose the most favourable branch and move accordingly.
As soon as the computer becomes able to read every move from the current board state onward, we enter late game status.
- Late game: the computer has already read every possible board status and will always play the better move, the human can only expect to do as well as his opponent, but has lost any chance of outplaying him.

The two players play the same game in much different ways, and a skillful human player can only hope to trick his opponent by presenting him confusing lines of play that seem when not read with enough depth, but end up trapping it.
The IA, on his side, will always go for the biggest move, and will always fell for a good
enough trap.

[Dechs Kaison]

What you're describing is high level, top down. Both the human and the computer evaluate the board state, reason some moves ahead and decide the best line of play. Chess is a calculation. Whether it's a brain making the decisions or a computer, the hteory is the same.

The IA, on his side, will always go for the biggest move, and will always fell for a good enough trap.

There aren't good enough traps anymore.

[RedNihilist]

What you're describing is high level, top down. Both the human and the computer evaluate the board state, reason some moves ahead and decide the best line of play. Chess is a calculation. Whether it's a brain making the decisions or a computer, the hteory is the same.

You're deliberately missing the point.
Computers aren't smarter than regular, human players, they're just correctly instructed on how to abuse their superior computational power.
Claiming them to be is no different than saying that dogs are smarter than dolphins because the latter can't find truffles.

[quote=&
quot;RedNihilist » Thu Mar 20, 2014 8:06 am"]The IA, on his side, will always go for the biggest move, and will always fell for a good enough trap.

There aren't good enough traps anymore.[/quote]

I stopped being interested in the thing a lot of time ago, but I didn't think chess became a solved game while I wasn't watching.

[Jonnymagic]

@dechs I haven't really played it enough, but from what I've played it has been fun. Everyone talks about the druid and priest being OP, but I've had some success with all of them really. The game im most excited about being a part of is HEX though. I did the kickstarter so I'm in the alpha, and they just released the patch for booster drafts. Holy crap is it fun to draft that game. Everything about it has been awesome so far, the biggest draw to me being that they built it for virtual only, so they purposefully make cards do things that paper cards could never do -- ie make 10 copies of this trap card and shuffle them into opponents deck, and when they draw that card they take 5 damage. Another cool one is the keyword *escalation*, everytime you cast that spell the text on the card changes and becomes more powerful. Really cool mechanics, loving it so far.

[redthirst]

There aren't good enough traps anymore.

You have to give the computer a cold - a virus - a computer virus. Send Will Smith and Jeff Goldbloom into the chess computer's mothership to download a virus and blow it with a nuke.



That's how you beat the computer.

[Pendulum]

Oh, goody, the Deep Blue Debate!

Saying that a computer can be "a good Chess player" is like saying that Halo is a good representation of actual space warfare; it isn't, because the two are directly oppositional actions. Playing against a computer is simply training for the real event of playing another human player to achieve the goal of the game; ipso facto, it doesn't matter how good the program is at helping you in its designated function, even if you are completely outclassed.
That being said, there is a difference between computationally correct play and the best line of play, and as such relying too much on the training provided by a computer is, in my mind, a bad thing that will hurt your ability in the long run: overwhelming your opponent's board state, for instance, is something a computer can't handle any better than a human can, and the computer
lacks the ability to alter their play away from what it considers correct, and this can and should be utilized to the goal of victory.. but it won't work against a human player who has different options available to them.

[Dechs Kaison]

I stopped being interested in the thing a lot of time ago, but I didn't think chess became a solved game while I wasn't watching.

November 5th, 2005 was the last time a human beat the top computer.

[Platypus]

I stopped being interested in the thing a lot of time ago, but I didn't think chess became a solved game while I wasn't watching.

November 5th, 2005 was the last time a human beat the top computer.

But that doesn't make the guy that programmed the top computer the best chess player, which was your original argument.

[rcwraspy]

I don't know Chess, but aside from what the computer is capable of doing I'd assume it boils down to this:

The programmer (let's just say he worked alone although he more than likely had a whole team working with him) had hours upon hours upon hours, likely years upon years, to expand the computer's ability to read the board and make plays. They likely started by determining the most possible board states - pieces and positions.

A human player, though certainly having put in hours and years of practice, can only ever analyze a new board state in real time. And they're on a clock. With a lot on the line.

[RedNihilist]

I stopped being interested in the thing a lot of time ago, but I didn't think chess became a solved game while I wasn't watching.

November 5th, 2005 was the last time a human beat the top computer.

But that doesn't make the
guy that programmed the top computer the best chess player, which was your original argument.

...and that's not even the definition of "solved game".

Learn by my mistakes: don't waste your time.

[TubeHunter]

Chess is lame, talk about something interesting.

[RedNihilist]

Let's talk about sportsball, then.

[TubeHunter]

Sounds infinitely better.

[TubeHunter]

So U.S. members of the F.O.S, you follow the NCAA Tournement? Anyone you got to win?


Personally, I think this is Louisville's year... again.

[hamfactorial]

Football is a solved game. No harder than chess, my alarm clock could beat Louisville.

[TubeHunter]

[DarthStabber]

On the brightside we know that humans will still be better at magic than computers for many years to come, and I'm pretty sure koreans are always going to be better at starcraft.

[Pendulum]

I read an article once about the AI they wrote for some FPS a while back that abused its own glitches to cheat, like clipping through the walls and shit. I'll have to see if I can dig that up.

[Kaitscralt]

pendulum's just mad that a commodore 64 beat him at chess

[Kaitscralt]

pendulum what arcane message are you trying to communicate with that avatar

is it secret code to darcy

[hamfactorial]

The AI still can't finish E.T. on the Atari 2600

[TubeHunter]

who can finish that?

[redthirst]

I'll worry about computers when they can win at Tic Tac Toe.

[Khaospawn]

I'll worry about computers when Skynet goes online.

[rcwraspy]

what if it already has?

[Valdarith]

The AI also can't beat TMNT on the PC.

[Valdarith]

Page 1000 coming soon...

[Platypus]

Yes.