mmt Vs. Ovyron (G4 D5 BG2)

[Ovyron]

34.Rfg1

1. g4 d5 2. Bg2 Bxg4 3. c4 c6 4. Qb3 e6 5. Qxb7 Nd7 6. Nc3 Ne7 7. cxd5 exd5 8. d4 Rb8 9. Qa6 Rb6 10. Qd3 Ng6 11. h3 Be6 12. Nf3 Bd6 13. h4 h5 14. b3 Nf6 15. Bg5 O-O 16. e3 Re8 17. Kf1 Bg4 18. Ne1 Bb4 19. Na4 Rb8 20. Nc2 Be7 21. f3 Be6 22. Nc5 Bc8 23. Kf2 Nd7 24. Ne6 Qa5 25. Bxe7 Rxe7 26. b4 Qb6 27. Ng5 Ba6 28. Qa3 Rbe8 29. Bf1 Bxf1 30. Raxf1 Qc7 31. Qd3 a5 32. a3 axb4 33. axb4 Qd6 34. Rfg1 Nb6

[d]4r1k1/4rpp1/1npq2n1/3p2Np/1P1P3P/3QPP2/2N2K2/6RR w - -

[jp]

I'm not comparing myself to what I could do against a group of opponents vs. what unassisted engine at 12 hours per move could do against them, but about what I could do against a 12 hours per move unassisted engine in a game against it. It wouldn't be able to surprise me with a move (so losing is out of the question)

Some time into the game against Harvey, you went into lockdown, apparently because you were surprised by his move. (Obviously Harvey is not an unassisted engine, but let's state the obvious, so the discussion doesn't go on some crazy sidetrack.) What did you discover in lockdown, what did you change because of it, etc.?

[Ovyron]

Some time into the game against Harvey, you went into lockdown, apparently because you were surprised by his move. (Obviously Harvey is not an unassisted engine, but let's state the obvious, so the discussion doesn't go on some crazy sidetrack.) What did you discover in lockdown, what did you change because of it, etc.?

I came up with an analysis method that played all of his moves in reasonable time, and used it to predict most of his moves for the rest of the game (and for the positions where he deviated, the moves I wrongly predicted would have been enough to defeat me). That's what happens when I play a superior opponent, they play a move I missed, if it's superior than the one I expected I try to come up with something that would play that move and previous moves, and then use that to predict what they would play for the rest of the game.

I have assimilated them by being able to predict them, and then I can use those methods against someone else. In this very game against mmt I played 19...Rb8 because that's the move I'd have predicted from Harvey (if I was white and was playing him I'd have predicted 19...Rb8), if the games happened in a different order and I didn't opponent model Harvey, I'd probably would have gone with 19...Rb7.

You'll see several of my opponents mention on further games that I'm playing much stronger than in our first game, because after having assimilated them I know how they play, and I'm usually able to beat them after a long stretch of games. On the ICCF one mostly plays an opponent once in a lifetime, or games happen simultaneously (one with white and one with black) so I don't get the chance for this, but otherwise I'm able to adapt against a specific opponent and exploit their weaknesses. This is a reason I often want to play black first on a series of games (I do the assimilation as black on the first game and beat them on the second with assimilation+white.)

I'm the result of assimilating all the superior opponents of the past and having analysis methods that mimic their stronger moves, regardless of hardware, so that's why I can boldly claim that my hardware suffices, I have yet to see a move that I couldn't find in reasonable time with it.

[SheikhYerbouti]

Good hardware would make the timescale minimal. I still think best optimal moves for white could hold a draw.

But further inspection would reveal that black missed a winning line on that draw, so what would be the point?

By minimal timescale I meant as long as necessary but as short as possible. So if that means a couple of hours per move so be it.

If g4 is obviously winning for black then you would expect a few lines inferring over -2.0. Anyway, over time what effect does ridiculous node numbers have on noise, does randomization reduce as the evals get more precise? Or increase as more lines converge in their rating? or stay roughly the same

[Ovyron]

By minimal timescale I meant as long as necessary but as short as possible. So if that means a couple of hours per move so be it.

The shortest is depth 1 and it'd produce complete garbage. Couple hours per move would be a waste because someone like Harvey can find better moves in a fraction of the time. Seriously the problem here is "unassisted engine." The problem with assisted engine is that an opponent is needed (otherwise you analyze 1.g4 d5 and never make a second white move.)

If g4 is obviously winning for black then you would expect a few lines inferring over -2.0.

Zenmastur has claimed that some -3.00 score is unstoppable. The real question is if a position exists where white can hold it despite that eval.

Anyway, over time what effect does ridiculous node numbers have on noise, does randomization reduce as the evals get more precise? Or increase as more lines converge in their rating? or stay roughly the same

It's impossible to add randomization without decreasing elo. Natural randomization coming from MultiCore engines still faces the problem of missing best attack/defense. The best solution is to get a skilled centaur to investigate, they can add variety on their own and not miss a key move. Both these games have shown that 1.g4 is likely lost as no blunder by white can be found after it.

[jp]

Natural randomization coming from MultiCore engines still faces the problem of missing best attack/defense. The best solution is to get a skilled centaur to investigate, they can add variety on their own and not miss a key move.

A skilled centaur also faces the same problem of possibly missing the best attack/defence.

[mmt]

1. g4 d5 2. Bg2 Bxg4 3. c4 c6 4. Qb3 e6 5. Qxb7 Nd7 6. Nc3 Ne7 7. cxd5 exd5 8. d4 Rb8 9. Qa6 Rb6 10. Qd3 Ng6 11. h3 Be6 12. Nf3 Bd6 13. h4 h5 14. b3 Nf6 15. Bg5 O-O 16. e3 Re8 17. Kf1 Bg4 18. Ne1 Bb4 19. Na4 Rb8 20. Nc2 Be7 21. f3 Be6 22. Nc5 Bc8 23. Kf2 Nd7 24. Ne6 Qa5 25. Bxe7 Rxe7 26. b4 Qb6 27. Ng5 Ba6 28. Qa3 Rbe8 29. Bf1 Bxf1 30. Raxf1 Qc7 31. Qd3 a5 32. a3 axb4 33. axb4 Qd6 34. Rfg1 Nb6 35. Qf5

[d]4r1k1/4rpp1/1npq2n1/3p1QNp/1P1P3P/4PP2/2N2K2/6RR b - - 4 3

[jp]

Komodo 13.2, depth 30: -1.36 (... Nf8).

[Uri Blass]

12 hours per move are enough to get easily the IM title so it is not clear that your methods are better than unassisted engines.

I'm not comparing myself to what I could do against a group of opponents vs. what unassisted engine at 12 hours per move could do against them, but about what I could do against a 12 hours per move unassisted engine in a game against it. It wouldn't be able to surprise me with a move (so losing is out of the question) and depending on opening I could lead it to a position where I predict it'd go into a losing variation (which is the point of being able to predict its moves.)

This would remain valid even if you give more time to the engine or faster hardware to the engine (which is equivalent) and that's why I claim I'm on a higher level than zullil if he went and relayed his unassisted engine moves to me in a game. To catch me he'd need centaur skills (and that's why a game against Zenmastur would be more interesting: he has the hardware and the skills, so he's probably at a higher level already, but is it enough?)

I do not see the reason for your confidence that unassisted engine cannot surpise you with a move and win.

[Ovyron]

I do not see the reason for your confidence that unassisted engine cannot surpise you with a move and win.

Well, let's say unassisted Stockfish. Sure, if my opponent relies on some other unassisted engine they can surprise me. I'm even fearful of high depth Leela as unassisted high depth Leela is something I can't emulate.

But Stockfish? Pfft! I can believe it'd do it if a skilled centaur provided for it a good book (leaving it into a position where it can win by itself) but otherwise you overestimate it. Unassisted Stockfish 13 will destroy Stockfish 11 with some 100 elo advantage, and corr chess controls allow one to find S13's moves already, just like I was finding Stockfish 11's moves on Stockfish 9's times, because the moves are there in the open, the moves can't be hidden from us.