Chess will never be solved, here's why

@5761
I verified and the engine always plays a table base correct move as its top 1 move. Not always the fastest, but never a mistake.

And it draws from a winning position. What are you trying to say?

As I've mentioned before, you don't understand how tablebases work

Now please a KRPP vs. KRP that draws.

Would you like cherries on top? You are the one asserting your method to be correct, you should have checked these things at the outset. Why do you always expect others to fry their chips while you sit back.

As it happens I've kicked off a series of those too, because I knew you'd ask. Results in a few days.

In the meantime there's no reason for you not to apply your calculation to the KRBN vs KQB games I already posted. If they don't work there we can ignore your thesis.

[Variant "From Position"]
[FEN "2r5/8/8/2pP3K/2P5/3R2k1/8/8 b - - 0 1"]

1... Kf4 2. Kg6 Ke5 3. Kf7 Rc7+ 4. Ke8 Kd6 5. Rg3 Rc8+ 6. Kf7 Rc7+ 7. Kf6 Ra7 8. Kf5 Ra4 9. Rg6+ Ke7 10. Rg4 Kd6 11. Re4 Kd7 12. Kf4 Ra3 13. Re3 Ra4 14. Rc3 Ra1 15. Kf3 Kc7 16. Re3 Ra4 17. Re4 Kb7 18. Kf4 Kc7 19. Ke5 Ra3 20. Ke6 Rh3 21. Ke7 Rh7+ 22. Ke8 Rh8+ 23. Kf7 Kd8 24. Ke6 Re8+ 25. Kf5 Rf8+ 26. Kg4 Rg8+ 27. Kf3 Rh8 28. Ke2 Rh3 29. Re3 Rh4 30. Kd3 Rh6 31. Rf3 Kc7 32. Rf7+ Kd6 33. Rf8 Rh3+ 34. Kc2 Ke5 35. Re8+ Kd4 36. d6 Rh7 37. Re7 Rh2+ 38. Kb3 Rh3+ 39. Ka4 Kxc4 40. Re4+ Kd3 41. d7 Kxe4 42. d8=Q Rh5 43. Qg8 Rf5 44. Qe6+ Re5 45. Qc4+ Ke3 46. Kb5 Kf3 47. Kc6 Re4 48. Qc2 Rf4 49. Kd5 Ke3 50. Qc1+ Kf3 51. Qf1+ Kg3 52. Qe2 Rf8 53. Qe3+ Kg4 54. Qe4+ Rf4 55. Qg2+ Kf5 56. Qh3+ Kg5 57. Ke5 Rf8 58. Qe3+ Kg6 59. Qg1+ Kh5 60. Qxc5 Rf1 61. Qd5 Rf2 62. Qe4 Rf7 63. Ke6 Rg7 64. Qe3 Kg6 65. Qd3+ Kg5 66. Qd8+ Kh6 67. Qh8+ Kg6 68. Qh4 Ra7 69. Qe4+ Kh5 70. Qh1+ Kg4 71. Qg1+ Kf3 72. Qxa7 Ke2 73. Qd4 Kf3 74. Qd3+ Kg2 75. Qe2+ Kh1 76. Kf5 Kg1 77. Ke4 Kh1 78. Kf3 Kg1 79. Qg2#

Yes. Syzygy can achieve the correct result. SF15 can't.

The question we were looking at was will @tygxc ever be able to?

@5774

"it draws from a winning position" ++ Not on my computer. It checkmates in 79, see above.

Very similar to my short think time examples (they're not all drawn from that position). Also very inaccurate. Try running it for 2048 seconds per ply. You're not guaranteed the same game or even the same result, nor even a game with fewer blunders, but I think the same result might be likely.

"What are you trying to say?" ++ You do something wrong, or you run a bad version of SF.

Not true. You just don't know how to use the Syzygy site.

"you don't understand how tablebases work" ++ I understand that very well.

If you understood very well you wouldn't keep posting that they strongly solve 7 man positions under competition rules. They do strongly solve K v K, KB v K and KN v K, but that's all.

 "Would you like cherries on top?" ++ No cherries, just a relevant drawn FEN KRPP vs. KRP.

As I said; Coming up. Any chance of you doing some work to check your own hypotheses?

You could start by applying your calculations to the KRBNvKQB games I already posted for you.

Save everybody else a lot of work. We only need one counterexample.

"You are the one asserting your method to be correct" ++ Yes.

"you should have checked these things at the outset" ++ I did.

Obviously not.

"Why do you always expect others to fry their chips" ++ I fry a lot of chips, and do not sit back.

So here is a drawn position.

Produce a series of at least 10 games with think times up to 2048 seconds for it. We'll wait.

Or if you're going to say that position is irrelevant as soon as you can't get your calculation to work, maybe this 8 man draw that you yourself posted earlier in the thread. 

We can't check how many errors in that one but we can put a lower bound on it.

Get frying. But preferably after you've applied your calculation to the 23 games I've already posted for the purpose. If you do that first we can all stop talking to you.

"As it happens I've kicked off a series of those too, because I knew you'd ask. Results in a few days." ++ Very well, let us see. Hopefully the interface then works again too.

"SF15 can't." ++ I did it with Stockfish 14 NNUE.

Versus very weak opposition (Stockfish 14 NNUE).

I should have been more precise; SF15 can't always.

Says you. Don't think so.

...

"as the earlier version" ++ No, there are no 2 versions.

@5765
"Why is my example irrelevant and KRPP vs KRP relevant?"
++ Your first example cannot be reached from the initial position by optimal play from both sides. That is why it is irrelevant for weakly solving Chess.

For once I agree with @Optimissed.

(I'm getting worried.)

A 32-piece tablebase is a strong solution of chess. You can trivially use it to make optimal strategies for each side with merely 1 access of the database per move.

[This was a reply to a post that seems to have been deleted].

Syzygy and its 8-piece successor (under development) are state of the art and use about 1 bit per position, so no need to argue about WD/WDL.

@tygxc says  So an array of 10^44 bit: 0 = draw, 1 = no draw is enough to store a strong solution of Chess. justifying it by postulating a WD tablebase. Are you saying I should have agreed with @tygxc in preference to @Optimissed?

Is 10^44 classical bits enough to hold a strong solution of competition rules chess without holding a generation program in the probe code? Like @Optimissed I don't think so, but you can try to persuade me (granted it's a lot of bits).

Remember Syzygy is not a strong solution of any but a few endgame positions, not even the ply count 0 positions.

But I take your point about simply storing the Syzygy generation program. That brings us back to the definition of "solution".

Correct about the count. For practical purposes, it doesn't matter. It's too big to consider without a paradigm shift.

Reinstated in essence.

@tygxc is specifying a WD tablebase. That's not a solution. A WDL tablebase is also not a solution.

And 10^44 is not the number of competition rule game states, nor even Tromp's number.

I think Huffman codes come into it. Whether a solution could be stored in 10^44 is open to discussion.