Only a human can solve this

>only a human can solve this
Never mind that, I'm more interested in knowing how you can get a third rook and the pawns in those positions?

Other urls found in this thread:

telegraph.co.uk/science/2017/03/14/can-solve-chess-problem-holds-key-human-consciousness/
mnn.com/green-tech/computers/stories/does-chess-problem-reveal-key-human-consciousness
lichess.org/GJ5cvu1h
lichess.org/Airp0Ffm
chessgames.com/perl/chessgame?gid=1497429
chessgames.com/perl/chessgame?gid=1287069
chessgames.com/perl/chessgame?gid=1069554
twitter.com/NSFWRedditGif

Also, one of those rooks should be on the lighter squares.

What do you mean only a human can solve this? The position's a draw. Maybe a computer would give a hefty advantage to black without understanding that even up all that material, they can't actually checkmate, but it's not a "solution".

>I'm more interested in knowing how you can get a third rook and the pawns in those positions?
There are only two rooks. And it would almost have to be a help position. I think someone composed one for this, actually, but I'd have to go digging through the archives for it.

>Calling a bishop a rook

Oops, sorry.

Supposedly, White can still checkmate.

White cannot checkmate. The bishops prevent any advance onto c7, and can blockade if necessary, the king is fenced out, and taking either of the rooks doesn't help white, worst comes to worst, black can sac the queen and both rooks for the pawns on b3 and c4, and still have an easy win. The position is dead drawn, it's famous because computers think black is overwhelmingly stronger.

telegraph.co.uk/science/2017/03/14/can-solve-chess-problem-holds-key-human-consciousness/
mnn.com/green-tech/computers/stories/does-chess-problem-reveal-key-human-consciousness

So long as white doesn't try to capture the rooks or move their one pawn higher up, the only thing black can move at all is the bishops. The white king can't capture them either, but he can move out to the aid of the upper pawn, allowing it to be promoted and evening the odds.

>White cannot checkmate

Except it literally can.

But unless black suffers a stroke, white will never be able to advance the pawn on c6. He has no less than 3 bishops that he can defend the square with, and can just park one of them on c7, blockading the pawn, while a second one shuffles along the d6-h2 diagonal indefinitely. The two bishops defend each other, so the white king can't take either.

Hell, even one bishop alone, as long as he's smart enough to stay on the diagonal and avoid the white king, can always sacrifice himself for the pawn as soon as it steps on c7.

White winning requires black to make an unforced blunder, which you are never supposed to assume in a puzzle.

Ok then, show me the checkmate. Because I guarantee I can hold this position with black indefinitely, no matter who you get to play for the white pieces.

lichess.org/GJ5cvu1h

>game aborted
Booooooooooooooooooooooooooooooooo

lichess.org/Airp0Ffm
Re-set. Someone joined the challenge without wanting to demonstrate white's supposed win.

>pawns prevent black from moving anything important
>bishops are all on black squares
>only a white walker can solve this.

>he doesn't play with the original elephants

What a fucking pleb. You need your babby's first chess piece to move any number of squares it wants. Give the elephants a one square move alongside the leap and it's balanced with knights perfectly.

And we got a black win. Granted, with unforced white errors, but I think it was demonstrative. You can't keep the bishops out of c7, and with that, white can't hope to make any sort of progress at all. Any white win involves black letting c7 slip for some reason, and there's no method to force your way through.

If the computer tries to delay a draw from over-repeating the board, a bishop will eventually be vulnerable.

what if we sacrifice the c pawn and move the king to d5, then c6 after the sacrifice? anything we can do?

But the pawn hasn't moved yet so it can go two spaces

It's so obvious!

The pawn is not on the starting second rank. It cannot move 2 spaces. For a white pawn to get on c6, it would need to move at least 3 times.

There you go!

Qxb5+ and black wins, not draws.

Furthermore, what happened to the pawn on c6? Black can easily stop it from ever moving by planting one of those bishops on c7 and blockading it.

Why did Black move the bishop to c7 then move it away? Leaving it there forces a draw.

Lol, looking back at my plays there's more wrong than that. I'm just retarded I guess.

holy shit
Here are the correct terms for the commonly misconceived pieces in chess:
>Rook
Actually a Castle/Tower
>Pawn
It's a peasant, alternatively you can call them fellow plebs
>Knights
Egregious mistake
They are horses

Black queen just takes the pawn then. Not like it's pinned

Quoth David Bronstein, "Pawns are excellent building material--you can make whatever you like with them."

Or, say, the Vienna Game (1. e4 e5 2. Nc3). Safe but dull opening that threatens nothing; there's a reason top-level players stopped using it after the 19th century.

please

its
>Prawns
>Forts
>Ships Figurheads

in origional chess, Ships Figurheads were all naked women

The conditions that need to be matched for the puzzle to be solved are:
White needs to either win or draw.
End of list.

The solution, then, is simple. White dicks around with the king on white squares for fifty turns. The three bishops are the only pieces the black side can move, and they are unable to eat any other pieces. Thus, after the fifty turns are done, the white player demands a draw on the grounds that in fifty turns, no pawn has moved and no piece has been eaten.

Where's my prize?

When you get down to it, GMs tend to favor the openings that give you the chance of drawing. Stuff like King's Gambit, Latvian Gambit, etc are too fucking dangerous--you either win or lose, draws are almost impossible to pull off.

Fans of weird openings/gambits are like a cult--they cannot be talked out of their insanity. No matter how many times they play a Latvian Gambit, Enguld Gambit, Damiano's Defense, etc and lose, they never learn--then they'll beat some sub-1000 player with a well-known opening trap which renews their faith in the opening.

I swear, I want to bang my head on the table when I try to get a Ruy Lopez and the retard on the other side of the board moves 2...f5.

>Enguld Gambit
What's this? I've never heard of it.

1. d4 e5

Basically reverse Scandinavian Defense, but essentially worthless and with no benefit to Black at all. He gives up a pawn for absolutely nothing, since unlike with SD, he can't immediately recapture. Almost as pointless and stupid as Damiano's Defense.

Getting a third bishop is easy. You'd just need to move a pawn to the other side of the board, and it'll get promoted to whatever piece you want. Though why you'd make it a bishop instead of a queen is beyond me.

The main reason is showing off.
chessgames.com/perl/chessgame?gid=1497429

The only conceivable reason to promote a rook or bishop is if promoting to a queen would cause stalemate.

Or that you know your opponent will take immediately anyway and it doesn't matter what you promote to.

>queen to c7
>pawn b6 to a5, promote to rook or queen

There are a lot of bizarre Ruy Lopez variants that are unsound and rarely used as well, like Bird's Defense (3...Nd4) and Schleimann's Defense (3...f5).

I apologize to anybody that plays the Schleimann Defense...but this opening is mega crap. For the life of me, I can't understand why black would rip open his king side (in many of the main lines) for insufficient compensation. I mean come on, most top GMs won't even play this opening for surprise value...and that's saying a lot. Most victories by black come when one of two things happen 1. the white player in unprepared (which makes any virtually opening sound, if one player knows what they're doing and the other does not) and 2. if white tries too hard to punish black for playing the opening.

Was trying to figure out why that move wasn't made.
Thought I was missing something.

Just a check then, and defeat to follow for white as the wall blocking everything has broken.

Some guy once drew Kasparov with a Bird's Defense so it can't be too bad. ;)

That's fucking retarded. The only reason why a computer couldn't solve this trivial puzzle is if its logic is not embodied (e.g. based in action). It's downright trivial with embodied logic -- you just look at what can do what. What black pieces can move? The bishops. But they can't threaten the king on a white space. White on the other hand can't do anything except moving their king and their one free pawn. One of white's pawns could take a piece inside the border, but then everything breaks loose and black wins.

No, that the AI given this problem failed to solve it just evidence that they were shittily programmed, not that humans are magical problem solving machines. Showing that there is a series of behavior which forces a draw is downright trivial. (Also keep in mind that that is the original goal, not showing that white can win.)

Henry Bird also tried out 2. Bb5 in the French Defense once or twice. The general rule of thumb is that if an opening or variation is named after him, don't use it.

bump

I know next to nothing about chess, how in the everliving fuck can the board look like that unless you start with it?

You basically can't wind up with that situation in a real game. Circumstances where you get an extra bishop are extremely rare to begin with, getting two is something I doubt has ever happened, and all on the same color just cuts the probability further. And that's ignoring the rest of the kookiness in that setup, which is easier to arrange, but still not something I can imagine happening in a real game.

It's obvious that black can't win if white doesn't let it, but OP is saying it's somehow possible to mate, which seems like bullshit to me.

bump

I don't see how white can win. If you're playing white all you have to do is keep the bishops in that line, and move whichever bishop the king isn't threatening back and forth. Since the king can only threaten two at a time, and capturing one is an illegal move because the other is protecting it, white can never take a bishop. Which means white can never get its pawn into play, since it'll just be taken immediately, and white can't do anything in the box or the queen gets out and starts gobbling up pawns. It's a stalemate pure and simple.

Whoops, I mean if you're playing black all you have to do is keep the bishops in a line.

>No, that the AI given this problem failed to solve it just evidence that they were shittily programmed
No, it just shows that the AI was programmed to win chess GAMES, not solve abstract chess PUZZLES.

This position isn't a puzzle. There's no solution. There's simply a bizarre and extremely unusual symmetry, where the usual brute force method of calculating lines doesn't work for the AI, because there are literally thousands of lines that all lead to the same place, and it has trouble distinguishing them.

ML nerd here, short answer is a sufficiently-trained program can solve this as well or better as any human can and if it failed to do so it was because it was either poorly constructed or insufficiently trained

>AI given this problem failed to solve it just evidence that they were shittily programmed, not that humans are magical problem solving machines.
this is correct

>No, it just shows that the AI was programmed to win chess GAMES, not solve abstract chess PUZZLES.
this most likely is not
typically an AI will learn what moves strengthen its position from a given state; it can look at any arrangement of pieces and tell you what the next strongest move will be based on a huge variety of features

an easy way to think of any given boardstate is, well, as a chess puzzle, which the AI knows how to solve quite well

>usual brute force method of calculating lines
this isn't really how that works because it's too computationally intensive, it'll look a little bit ahead but not at that many possibilities (relatively speaking), instead drawing on past experiences of similar situations and what led to the greatest success

>because there are literally thousands of lines that all lead to the same place, and it has trouble distinguishing them
most AI have efficiency checks built in to detect repeated states to speed things up, this isn't really an issue

According to the article there is a solution where white wins.

t. computer

Modern computers usually look 40-50 ply ahead, with some sort of pruning function to eliminate obviously worse lines from the computational tree. However, it's going to have trouble doing that in this position, since the plethora of bishop moves are all going to lead to identical positions, so it won't prune much.

>instead drawing on past experiences of similar situations and what led to the greatest success
The only modern chess AI that still attempts this paradigm is Junior, and it routinely gets trounced by the "dumber" AIs like Stockfish, Shredder, Rybka, and HIARCS.

>most AI have efficiency checks built in to detect repeated states to speed things up, this isn't really an issue
But that's the thing, you don't get repeated states. You get very similar states, which are nonetheless distinct. True, there's no real advantage over one bishop on c7 and the other on e5 as opposed to d6, but the computer doesn't realize that, so thrums for both.

The article is incorrect. As stated above, white can literally do nothing to stop Black from parking a bishop on c7, and then shuttling another bishop back and forth to make sure that it's constantly defended. At that point, white can move the king around (draws) or try to take one or both of the rooks with the pawns on c4 and b3 (loses).

The only way white can win is if Black makes no less than 2 unforced errors. First, he would need to get the king to a8, which should be impossible, as both b7 and b8 can be covered by the black king and bishop respectively. Then, c7 would need to remain unguarded long enough for white to push the pawn on c6 to c7, and then to c8 for a promotion to either queen or bishop which checkmates.

White can win. White does not have a win. Similarly, black can win if white makes a number of errors, namely taking anything with the pawns on b3 or c4.

Good ol' GNU Chess. The AI is none too bright though; it does at most 13 ply. I had a program called WinChess when I was a kid which had cleaner graphics, but for some reason it wouldn't let you promote pawns to anything but a queen.

Good ol' GNU Chess. The AI is none too bright though; it does at most 13 ply. I had a program called WinChess when I was a kid which had cleaner graphics, but for some reason it wouldn't let you promote pawns to anything but a queen.

I have to set the AI move time to 20 seconds or less though because it'll spend hours on a move most of the time. It's ok with tactical play, but positional games tend to trip it up; try and play through a grandmaster game and it'll make some basic, stupid capture when in the actual game, the player avoided the capture because it would have produced a losing endgame.

>The article is incorrect.
Who the hell died and made you in charge of what is and isn't possible?

Spassky turned 81 in January; he's third behind Botvinnik and Smyslov for longest-lived WC. Interesting how many of them died at an early age.

>Steinitz
64
>Lasker
71
Capablanca
53
>Alekhine
55
>Botvinnik
83
>Smyslov
89
>Tal
54
>Petrosian
54
>Fischer
66

If you look at past World Champs, the ones who lived past 70 (Lasker, Botvinnik, Smyslov, Spassky) were the ones who had a variety of interests and a life outside of chess. The WCs who died early had no life other than high level professional chess.

Lasker was a mathematician and psychologist, Botvinnik was a computer scientist, Smyslov was a singer, and I think Spassky is into opera or something.

Spassky said his last conversation with Fischer involved a debate as to whether 1. e4 or 1. d4 was a better way to open a game. They concluded that 1. d4 was better because the pawn is protected by the queen. Fischer back in the day had been known for saying 1. e4 was "best by test" and used it in about 90% of his games as White during the 60s. However, he did begin using 1. d4 openings regularly in the 70s.

Back in the 1920s-30s, 1. e4 openings of any kind were uncool and the hypermoderns ridiculed them. Richard Reti said that the Sicilian and French Defense were the only acceptable answers to 1. e4. This is generally still true since the only 1. e4 e5 opening commonly used by elite players is the Ruy Lopez.

They like to refer to 1. e4 e5 as the Open Game and 1. d4 d5 as the Closed Game because the former usually lead to tactical games and the latter to positional ones, while Semi-Open Game is 1. e4 followed by any response other than e5, and Semi-Closed is 1. d4 followed by a response other than d5.

It's not always true of course, since the Ruy Lopez typically results in a closed, positional game while queen side pawn openings can lead to open, tactical play especially if you use something like the Albin Countergambit.

Based on my experience, Ruy Lopez is the only 1. e4 e5 opening worth playing. I get pissed the fuck off when I'm trying to get a Ruy Lopez and the asshole across from me does 3. d4 or something instead.

That's because the problem was proposed by the Penrose Institute.

Penrose is an amazing mathematician and physicist - but he thinks that AI is impossible because the human mind is based on quantum processes and special physics and SHUT UP - CHINESE ROOM CHINEEEEESEEE REEEEEE-OOOM!

It's really not surprising that such a flawed dis-proof of AI would come out of it.

Korchnoi was not a world champ, but he lived to be 85 and didn't have a life outside chess.

Spassky was asked if he thought Korchnoi deserved to be WC and he said "No, absolutely not. He had no style of his own."

Fischer died of diabetes-induced kidney failure, but he refused medical treatment because he was a paranoid nutcase and thought the Jew doctors would kill him or something.

>Though why you'd make it a bishop instead of a queen is beyond me.
If promoting to queen would cause a tie.

chessgames.com/perl/chessgame?gid=1287069

Here's a very rare example of promoting to a bishop in a real game. In this case, White avoids stalemate and his new light squared bishop can simply eat the Black queen-side pawns and free up the White ones to march to the back rank.

Korchnoi was a bitter old man that was always behind. He was the Seto Kaiba of chess. (Kinda like Nakamura, actually.)

Why hasn't chess been "solved" yet? I mean, it is complicated, but you'd think it's notability alone would mean someone would've solved it by now, if only through sheer brute force alone.

Pro players are capable of games that have an exact, set movement up to like 25-30 moves because of how tried and true certain games are, like some variations of Berlin Defense. But it couldn't be extended towards a vast majority of the games. Just playing something like 1. c4 is enough to make things interesting.

That, and chess is alreayd kinda semi-solved because most of the games are drawn.

Oh yeah. Hate was what made him get up in the morning. He'd retired from competition after having a stroke in 2012, but recovered enough to take on Wolfgang Uhlmann in a 2014 match and beat him 4-2. He also beat Mark Tamianov 2-1-1 in 2015, his final match before his death. For an octogenarian in a wheelchair, his performance in these games was pretty good too.

Oddly, chessgames.com doesn't have the Tamianov match, the last Korchnoi games listed are the ones from the Uhlmann match.

>Why hasn't chess been "solved" yet? I mean, it is complicated, but you'd think it's notability alone would mean someone would've solved it by now, if only through sheer brute force alone.

I don't think you realize the magnitude of the problem of "brute force". You have a potential pool of possible games that is several orders of magnitude larger than the estimated number of atoms in the universe.

Well, that and paranoid fear. I met him once, at the Manhattan chess club, introduced by a mutual acquaintance. I tried to get his autograph, and he accused me of being a KGB agent and stormed off. Considering this was in 2002, I'm not sure why he thought that.

I don't get it - why would a Queen have led to a stalemate?

Because if he promotes to a queen, then the black bishop cannot move, as it is pinned. None of the pawns can move, they're all locked, and the black king can neither go to g7, covered by the pawn on h6 nor to g8, covered by the bishop on d5.

But a bishop promotion allows an immediate coup de grace by playing BF6#, to which black has no answer.

Ah, ok. For some reason I thought that'd be checkmate, but I guess you have to be threatening the King at the time you pin it for that to count?

Basically there's a retarded (and frankly autistic) rule that states that if no legal moves can be made while the king is not in check, it's a draw.

Honestly, it should simply be "You cannot move, you lose".

Or bare minimum expand the checkmate rule to "if you can't make a move without putting your king into check" instead of "if you can't make a move to take your king out of check." That'll leave the situations like where you both did something retarded and now you're stuck.

>Basically there's a retarded (and frankly autistic) rule that states that if no legal moves can be made while the king is not in check, it's a draw.

Eh? Stalemate is a perfectly fine rule; it's designed to punish you for making careless moves while trying to set a mating net around the enemy king.

chessgames.com/perl/chessgame?gid=1069554

See this example. Ms. Polgar had this one in the bag; she just had to play 74. Ke2. Instead she made a catastrophic blunder with 74. Kd4, stalemating Karpov and allowing him to escape with a drawn game.

Lyl Karpov also got stalemated by Korchnoi in game 4 of the 1978 WC match. The game set two records. One, it remains to date the longest game in a WC match (124 moves) and two the only stalemate in a WC game. Korchnoi said "If I could not win the game, it was my pleasure to stalemate the world champion."

The hate between Karpov and Korchnoi was legendary.

Getting a pawn to the far side of the board and not being smart enough to just get a Queen.

The thing is, this is a nigh-impossible configuration due to the pawn positions, so algorithms that deal with chess probably only account for things they'd allowed their opponent to do before finishing them off.

Even if this is technically possible, an ai would've check-mated them before they could get into this dumb-ass positioning.

AIs aren't magical problem solving machines either. I don't know a whole lot about chess engines, but it might be that they generally are incapable of that type of logic. Just because a problem trivial for the average human being to solve doesn't mean that an AI, even one which is very good at certain types of tasks which humans generally think are closely related to the one at hand, has any capacity to solve it. It doesn't mean that the AI was badly made either, just that it was made to do things which are actually significantly different from what is needed to do to solve this problem, whether it looks that way or not.

>Three rooks

>But a bishop promotion allows an immediate coup de grace by playing BF6#, to which black has no answer.

Almost. Black can prevent mate with 44...Bf7, however after 45. Bf6+ Kg8 46. Bxf7+ Kxf7, Black is left with nothing but his king and (locked) pawns against the White bishop. Then as I said, the bishop can pick off the Black pawns and liberate the White ones on their journey to the back rank.

I'd be willing to bet that Black never saw it coming either. He probably assumed his opponent would simply promote to a queen and thus instant stalemate.

The Albin Countergambit (1. d4 d5 2. c4 e5) was popular in the first two decades of the 20th century and was a favorite of Frank Marshall. It died out after WWI, but has never actually been refuted.

Frank Marshall was known as a tactical attacker in his youth, in the 1920s he followed the trends and switched to more conservative, safe playing styles centered around d4 openings (of course, so did Capablanca and Alekhine post-WWI). In his last decade, Marshall added the Caro-Kann and Indian Defenses to his opening repertoire.

A good player, although he wasn't as good as Capablanca, Alekhine, and Lasker, and scored poorly against them. He did have a good record against the hypermoderns other than Tartakower, who seemed to have him figured out.

Alekhine was undisputed WC from 1927 until his death in 1946, aside from the two years he lost the title to Max Euwe. That he died when he did was convenient and preserved his legacy, because there's little chance he could have defended his title against Mikhail Botvinnik, who had the full support of the Soviet chess machine behind him, and he was getting old anyway (there was a 20 year age difference between the two). Of course then Botvinnik won the WC against Paul Keres, who performed very poorly, at a sub-2000 level during their match. His defense of his title against David Bronstein three years later was also an exercise in slop chess and probably one of the shittiest WC matches ever.

Emanuel Lasker reportedly said of the young Paul Keres "This boy will never become World Champion." and he was right. Lasker didn't see any potential in him, although Keres was a viable contender for world champ until the mid-1960s. If you believe what Spassky said, world champions are necessarily guys who bring something unique and different to the game, which neither Korchnoi or Keres did.

Keres went to an early grave, not yet 60 when he died. He was a quite patriotic Estonian and never really accepted the Soviet occupation of his homeland.

You couldn't be more wrong, old boy!
>Rook
>Actually a Castle/Tower
It is neither of those. It's proper name is the Schlongpollip

>>Pawn
>It's a peasant, alternatively you can call them fellow plebs
I am not fellows with plebs, good sir. At any rate they are Poonswoggles.

>Knights
>Egregious mistake
>They are horses
Again, there you are being incorrect sir. They are not a mistake and they are most certainly not knights! The proper name is Sperniferous Dongwaggler you see.

Let's do that one instead and let go of our contemporary chess with its demented horse-lust.