Chess thread

[[Lo...]]

Thanks to chessplayer, my foggy brain has remembered that there are chess game databases online. Something that was obvious to me for years before my mind became a blur.

 

Long story short. I thought I'd create a chess thread just devoted to chess in general.

 

Been looking at this game. Very well played by both. A really creative win....

 

http://www.chessgames.com/perl/chessgame?gid=1106494

[Guest [ou...]]

Hi LorazepamFree2015

 

I know this is a chess thread but I've already asked chessplayer a similar question

and he hasn't replied so I might as well ask you too.

 

Are you in a any way interested in backgammon?

In any case, here's my take on the chess vs backgammon 'war'.

 

1. I don't play chess though I know the rules.

2. If you understand probability, you know that backgammon <> luck in the long run.

3. Both games are ABOUT equally old and complex.

4. My problem with chess is that its deterministic nature makes it theoretically solvable.

5. Greek (check my profile) backgammon has some interesting variants.

[[Lo...]]

Hey outis,

 

Chess kind of holds sentimental value to me since I learned to play very young, and it was always something to fall back on during different parts of my file. Sort of like an anchor. Chess can be very fiercely competitive, but there's a sense of aesthetics to it, which I was always drawn to.

 

Now, I don't know about backgammon. I've heard of the game, but never really learned it. I learned to play bridge a little bit, but it never had much appeal. Was never much into card games, so really the only board game I really cared for was chess. I guess it just came naturally to me in some way.

 

But if you want to open a backgammon thread, why not? It may be an interesting distraction....

[[Lo...]]

[[...]]

:laugh:

[[Lo...]]

:laugh:

 

:laugh:

 

There was actually a discussion on a chess forum about psych meds and chess. Pretty interesting:

 

I play chess because I am a masochist. I know that I will never be really good at chess but I keep playing anyway and not just for the fun of it but also because deep inside I like to delude myself into thinking that maybe one day something will make sense and I will become really good at it. But unfortunately that's probably never going to happen. The reason is the following: I can memorize openings, endgames, I can learn tricks, solve puzzles, read chess books, etc... but that's not enough! You know why? Because my lack of ability to clearly visualize sequences of moves makes accurate calculation nearly impossible. I can calculate using an analysis board (I've come to think of the analysis board as the brain I don't have) but without it I'm helpless. That being said, does anyone know if antidepressants or similar drugs that temporarily enhance working memory will improve my OTB chess playing ability?

 

********** #6 Sep 18, 2010

Take it from someone who has been on a wide variety of anti psychotics, mood stabilizers and anti depressants at one time or another including: Risperdal, Haldol, Fluphenazine, Seroquel, Depakote, Lithium, Ativan, etc. No, It does not increase your ability to play chess at all. In fact, it inhibited my game playing. I always felt foggy in the head and was easily "tranced" (where I would find myself staring at the floor for several minutes without realizing it) I dont take any of these meds anymore and my ability to play has taken a jump up from when I was on them. ( I am not suggesting to anyone who is on pyschological medications to quit using them without first consulting your doctor) i decided to do this as a personal decision for myself and, yes I do suffer some consequence for it. I hope this helped you in some way. Keep playing and learning bro! Peace.

[Guest [ch...]]

Side effects may include doubled rooks on the 7th rank and uncontrollable urge to sacrifice your queen. Consult your chess coach if you have a mating attack lasting more than 4 hours.

[[Lo...]]

Pretty good 

 

I wonder which one of those -pam's causes the side effect to start every chess game by playing 1.b3 followed by 2. g3. Perhaps, indecisepam?

[Guest [ch...]]

Just for fun...

 

White to move and *lose*, more specifically force Black to checkmate him in 8 moves.

 

 

[Guest [ch...]]

The solution (from problem in previous comment)

 

1.Nb1+ Kb3 2.Qd1+ Rc2 3.Bc1 axb6 4.Ra1 b5 5.Rh1 bxc4 6.Ke1 c3 7.Ng1 f3 8.Bf1 f2 mate

 

Reaching this final position:

 

[[Lo...]]

That is a really good one. The final position is amazing. Didn't think White could even theoretically lose this. My mind automatically went to 1. Nb5, Kb3, 3. Qd3++. Heck, even 1. Nb1, Kb3, 3. Qd3++ will win it

 

Very difficult problem to solve. These studies are fun. One has to uncondition from regular chess thinking

[[Lo...]]

Gotta love this one, and the dual Queen sacrifice 

 

https://chess-db.com/public/game.jsp?id=4600010.Rogoff,%2520Kenneth.10852096.9680

[Guest [ch...]]

Are you in a any way interested in backgammon?

In any case, here's my take on the chess vs backgammon 'war'.

 

1. I don't play chess though I know the rules.

2. If you understand probability, you know that backgammon <> luck in the long run.

3. Both games are ABOUT equally old and complex.

4. My problem with chess is that its deterministic nature makes it theoretically solvable.

5. Greek (check my profile) backgammon has some interesting variants.

 

I'm afraid I don't play backgammon.

 

Your point 4 is true. Solving chess (and other pure strategy games) is an interesting problem. If you simplify it by restricting the number of pieces (and thus the number of possible positions), then the following is a linear-time algorithm for solving chess. This has been used to solve all chess positions with up to 6 pieces and many with 7 pieces. E.g. white K+R+B versus black k+n+n is 6 total pieces and has been solved as a win for white (!)). Connect-4 and 4x4x4 tictactoe have been solved by retrograde analysis. Checkers is not yet solved but seems solvable in the not-too-distant future.

 

(Retrograde Analysis) Form a bitvector with a bit for every possible position. E.g. for K+Q vs k+r, you need a bit for every distinct position of the white king, white queen, black king, and black rook, which would have roughly 64**4 bits in it. Now walk through the bits and mark with a 1 each that represents an immediate win for White (checkmate position). Call that W0. Then do an "unmove" operation, which means compute all positions where White could make 1 move and reach a position marked by a 1 in W0. These are all the "mate in 1" positions. Mark all these bit positions with 1 and call that W1. Now comes the tricky part. Negate that (1 in all positions where White *cannot* mate in 1), and do an unmove, and mark all resulting positions with 1. This vector B1 is the positions from which Black can make a move resulting in a position where White cannot mate him in 1 move. So B1 are the Black-is-not-losing-in-1-move positions. OK, do another negate and unmove to reach bitvector W2 - all positions from which White can win in 2 moves. Repeat the process to form W3, W4, etc, until the bitvector stabilizes. The end result Wn has 1's in all the winning positions for White.

 

Some interesting results have been obtained from retrograde analysis of certain chess endgames. See https://timkr.home.xs4all.nl/chess/perfect.htm and http://james-plasketts-coincidence-diary.blogspot.com/2006/03/75-silly-prominences-dr-charles-hunter.html . This world championship game wound up in an 8-piece ending *almost* in the solvable realm (Kasparov consulted with computer chess experts at the time) http://www.chessgames.com/perl/chessgame?gid=1067288

 

[Guest [ch...]]

Gotta love this one, and the dual Queen sacrifice 

 

https://chess-db.com/public/game.jsp?id=4600010.Rogoff,%2520Kenneth.10852096.9680

They must have agreed to a draw in advance, played some silly moves, then went to a bar?

[Guest [ch...]]

Here's one with 2 queen sacrifices ending in a picturesque position.

 

White to move and win (hard):

 

[Guest [ch...]]

White to move and win:

 

 

Solution:

 

 

[[Lo...]]

Gotta love this one, and the dual Queen sacrifice 

 

https://chess-db.com/public/game.jsp?id=4600010.Rogoff,%2520Kenneth.10852096.9680

They must have agreed to a draw in advance, played some silly moves, then went to a bar?

 

I remember reading about that long time ago. That was actually the 3rd time this same game was played. Not sure what the circumstances were anymore. I think it may have been the last game in the tournament when it didn't matter anymore or something like that (memory fails me, read about it as a teen). I think the first game was just "white played 1. e4" and then they agreed to a draw. But that didn't fly, so they were forced to play another game.

So, the 2nd game was something like "1.Nc3  1.Nc6, 2.Nb1 2.Nb8, 3.Nc3, 3.Nc6. 4.Nb1, 4... Nb8".... Game drawn after 3 repeats. The tournament organziers got furious and got them to play a "proper" game. So, the double queen sacrifice was the proper game. Somehow, my mind dredged up that memory. I remember having a good laugh about it.

[Guest [ch...]]

I just have to write about Google's stunning AI accomplishment "Alpha Zero" https://arxiv.org/pdf/1712.01815.pdf . Alpha Zero is now the world's strongest chess-playing entity, after beating the current computer chess champion (Stockfish, rated 3400), by a score of 28 to 0 (!). Alpha Zero uses machine learning (neural nets). It went from knowing nothing but the rules of chess, to discovering all 500 years of human knowledge and then some, to become a 3400+ strength player, in just 4 hours of self-training. Its strength and games are nothing short of amazing.

 

Computers first surpassed humans at chess in 1997 with Deep Blue's victory over Kasparov.

http://www.chessgames.com/perl/chessgame?gid=1070913 . However, Deep Blue did this by brute force calculation, looking at billions of possible continuations and picking the best line using hand-crafted chess evaluation code. Successor (such as Stockfish) did the same and managed to get all the way up to 3400 strength (top humans are around 2800). These programs are strong, but they play a dry "computer chess" style. Watching them has all the excitement of watching a calculator do arithmetic.

 

Alpha Zero is different. It comes out of machine learning (AI) research. Rather than writing custom code to do chess evaluation, you train a neural network, which operates like a human brain. The predecessor to Alpha Zero was Alpha Go, which used this technology to beat the world champion Go player. To demonstrate the flexibility of machine learning, they turned to chess. Alpha Zero trained its neural network to understand chess by playing 44 million games of chess against itself in 4 hours. It started from knowing nothing except the basic rules (how the pieces move, and checkmate). Not even values of pieces. By the end of the training, it had super-grandmaster level evaluation of chess positions.

 

Let's look at a game. The Alpha Zero paper presents a number of amazing games. Let's pick #9. http://www.chessgames.com/perl/chessgame?gid=1899422 . First interesting moment:

 

 

Alpha Zero has built a nice pawn wedge on d4-e5-f4. It likes space advantage. Stockfish has just played 15...Nxd2. How to recapture? A human would play 16.Nxd2 (not 16.Qxd2 Bb4) and follow up by castling. "Castle early and keep your king safe in the corner" is one of those rules (heuristics) we have been taught. Computers such as Deep Blue would do the same, having been programmed with those same rules. Alpha Zero is not prejudiced by any preconceived ideas. It's evaluation of *this* position reveals the king is quite safe and useful behind the wedge of pawns. So it plays 16.Kxd2!, and follows up with 17.Ke3 to reach this position:

 

 

After several more moves, it has maneuvered all its pieces into great spots. But how to take advantage of this?

 

 

Showing a disregard for material, Alpha Zero sacrifices its Bishop here with 30.Bxg6!

 

 

But after the reply 30...Bxg5 31.Qxg5 fxg6 we reach this position and Alpha Zero is a piece down. How to follow up?

 

 

32.f5! is the answer.

 

 

Offering a pawn that can't be taken. If 32...gxf5 33.Qg7+, and if 32...exf5 33.Qd6! wins. But what if Stockfish just plays 32...Rg8 ? Then Alpha Zero follows up by pushing the pawn to f6:

 

 

To evaluate this position correctly, it had to understand that its dominating position with the cramping pawn on f6 is worth a bishop.

Several moves later, it becomes clear Alpha Zero was correct in its evaluation. Queens have been exchanged, it has won a couple more pawns, and its King has penetrated to e7.

 

 

Stockfish is helpless against the advance of the f and g pawns. Stockfish resigned.