Longest possible game

[sje]

What is the length of the longest possible chess game? On first guess, I'd say 5,500 moves. Does any one have an exact figure? It might be useful to construct a PGN file for this as a test for PGN scanning software.

Also, what is the average length of a game where each move is randomly selected? My Monte Carlo experiment says this figure is around 167 moves.

[Dirt]

What is the length of the longest possible chess game? On first guess, I'd say 5,500 moves. Does any one have an exact figure? It might be useful to construct a PGN file for this as a test for PGN scanning software.

Also, what is the average length of a game where each move is randomly selected? My Monte Carlo experiment says this figure is around 167 moves.

The first result I got from a Google search says 5948.

[bob]

What is the length of the longest possible chess game? On first guess, I'd say 5,500 moves. Does any one have an exact figure? It might be useful to construct a PGN file for this as a test for PGN scanning software.

Also, what is the average length of a game where each move is randomly selected? My Monte Carlo experiment says this figure is around 167 moves.

The first result I got from a Google search says 5948.

Here is my thinking on the procedure:

First make 49 knight moves, then push a pawn one square, then make 49 knight moves, then push, and 49 knight moves. Repeat a total of 4 times to get the pawn to the 6th where it can/t move without capturing. Now 49 more knight moves except black's 50th move will now be to take the pawn with a knight. that is 5 total cycles or 300 moves so far. Repeat for 7 more pawns, which gets us to 2400 total moves. Now black starts pushing pawns but each can advance one more square so that adds 2700 moves to the total, we are now at 5100, and now all we can do is go for 14 more 50 move cycles as each piece is removed to reset the counter. So we get to 5100 + 700 seems to be 5800 plus another 50 before we reach the draw. It would seem that EP is not an issue as you would not want to give up two moves at one time. Mine is probably off a dozen or so, since at times black resets the counter 1/2 move early...

I could probably come up with a more accurate estimate if I had not been working on chess all day long already...

[sje]

I searched Google too, and I asked anyway because there doesn't seem to be a definitive answer.

Seeing a sample PGN game score would help; it could be checked by a program for missed draw claims and other problems.

[sje]

Each pawn has a maximum of six moves per game. That's a maximum of 96 (= 2 * 8 * 6) counter resets. There a maximum of 30 (= 2 * (16 -1)) captures. So there are a maximum of 126 resets. With 99 ply of non-resetting moves before each resetting move, than upper limit would be 12,600 ((99 + 1) * 126) ply (or 6,300 moves). After the last man is captured, the game is forcibly drawn by insufficient mating material.

The problem here is that the pawns block one another, so some pawns must be captured early or themselves must capture. Both cases eliminate one or more potential resetting moves.

Another problem is that the sides alternate captures over the long run. Each alternation removes one ply from the total for a total 15 ply deficit.

[Zach Wegner]

I think the optimum method involves pawns moving forward, and eight of them capturing regular pieces. Capturing a pawn takes away at least two zeroing moves. I think the figure is about 5498 is accurate. White starts out by moving the a,c,e,g pawns forward to the 6th rank, then black does the b,d,f,h, then white promotes his pawns, and then black. The captures can be interleaved for whichever side is to move.

[Onno Garms]

As claiming draw on repetition or 50 moves is optional and not required, there is obviously no limit on the length of a chess game.

[Ovyron]

As claiming draw on repetition or 50 moves is optional and not required, there is obviously no limit on the length of a chess game.

1. Nf3 Nf6 2. Ng1 Ng8 3. Nf3 Nf6 4. Ng1 Ng8 5. Nf3 Nf6 6. Ng1 Ng8 7. Nf3
Nf6 8. Ng1 Ng8 9. Nf3 Nf6 10. Ng1 Ng8...

[bob]

What is the length of the longest possible chess game? On first guess, I'd say 5,500 moves. Does any one have an exact figure? It might be useful to construct a PGN file for this as a test for PGN scanning software.

Also, what is the average length of a game where each move is randomly selected? My Monte Carlo experiment says this figure is around 167 moves.

BTW, the longest possible game is infinite in length, since claiming the 50 move rule or repetition is completely arbitrary.

[sje]

I think the optimum method involves pawns moving forward, and eight of them capturing regular pieces. Capturing a pawn takes away at least two zeroing moves. I think the figure is about 5498 is accurate. White starts out by moving the a,c,e,g pawns forward to the 6th rank, then black does the b,d,f,h, then white promotes his pawns, and then black. The captures can be interleaved for whichever side is to move.

I think this is very close to the solution. Still, it would be nice to have a game score for program verification.