|
Post by Vytron on Sept 3, 2019 18:22:37 GMT
But I'm not at the top, and I use engines incorrectly enough that I lost 2 games in the last year.
So I guess the critical question would be if you'd be interested in continuing playing games until you beat me (if I lose your claim stands) or something, or if this is the end of the line.
|
|
|
Post by Ozymandias on Sept 4, 2019 5:21:26 GMT
How many games have you lost since you were given the book you currently use? What openings did you play in those games? Do you still play those openings?
|
|
|
Post by Vytron on Sept 5, 2019 9:18:16 GMT
Haha! Do you mean the private book I was given to use for 12 +2 blitz games? While the openings on that book are excellent for quick games and to bridge the gap between my slow hardware and the top things people use today, its openings don't hold up at correspondence chess control. It's not a book that can be used against someone intelligent (unassisted engines are very stupid), and, anyway, the core of the book is from June 5 2019, as I was using it it became pretty clear that chess theory was advancing really fast, and to avoid getting the book obsoleted I had to play games daily and fix the book as lines were getting exposed and refuted.
At some point I took a 1 week break, and after I came back I don't even know if using the book was to my disadvantage, as some people were playing directly into its holes, and I was losing games playing its green moves directly, and it took me an entire week to fix it and go back to where I was before (no improvement.)
But corr chess games happen at a pace so slow that the book is just unsuitable for them. I tried it in 2 games and was in trouble, and fixing it was taking longer than not using it at all. So I haven't used the book for months and it didn't have any effect on our games whatsoever (mainly, because the book only covers the variations it plays. In our first game, it covers 4...Bb4+ and that's it. I played 4...Be7 so the book didn't have anything in it. In this game it covered 2.c4, so the book didn't have anything for 2.Bf4).
So, no, playing those opening in correspondence games would be suicide, and I haven't touched the book since July 20, so who knows if I went back to playing unassisted engine chess, if some random modern public book would perform better than it already, because book maintenance is hard work.
|
|
|
Post by Ozymandias on Sept 5, 2019 12:32:21 GMT
I suspect that book wasn't very good then. At least not for something other than what it was made for. The stats on the few corr games I've played, were obtained using public books. They're perfectly usable.
|
|
|
Post by Vytron on Sept 6, 2019 19:39:32 GMT
The main point of the book was to "outbook" the opponent to great lengths, on 12 min +2 games I would most of the time have more than twice the time than my opponents when I started thinking, and very often landed on positions that were near to the endgame, where there was no way they'd beat me even if I used a toaster engine, and thus I could draw games against Leelas on the fastest GPUs...
Corr chess is a different beast where there's no such thing as having twice the time as your opponent when out of book, because no matter what they'll have months to play the rest of the game. Where you don't want to allow them to reach an endgame position that is trivially drawn against you (specially if they have some 1800 FICGS level that would be easy to defeat with a novelty.) And where intelligent opponents can find a hole in the book and refute the line as it is being played, so it's best to play something you can't refute yourself (like the openings I played against you in our games. I still don't know how to measure their quality, but I get the vibe that I could never beat you.)
|
|
|
Post by Ozymandias on Sept 6, 2019 21:59:57 GMT
I get the vibe that I could never beat you.) If that's your impression, and I think the same about anyone at the top. And I haven't lost a single correspondence game, nor did I lose any in the last Freestyle tour. If you know why you lost the only two games you've lost in five years... what makes you think there's still room for legitimate wins against top opponents?
|
|
|
Post by Vytron on Sept 7, 2019 18:14:36 GMT
Just because I don't know how to do something, and don't see a way to do it, doesn't mean there's no way to do it. It's not like you've found the ultimate recipe to produce perfect games on the fly. I also estimate that some *me* from year 2024 could be able to defeat the *you* of today, it's just that my 2014 self didn't have any idea of what I'm doing in 2019 (the poor guy would add 7 unnecessary engines to analysis...) so who knows what I'll be doing by then (unless I quit corr chess so I'm close to my peak in strength.)
I remember believing that someone like Highendman had the power with all his resources to produce perfect chess moves, and any deviation from them would turn out to be crunched by him. It turns out his power wasn't impressive at all, and if he was as strong as I believed he was (and he was probably much weaker) I believe I surpassed him at around the time Stockfish 8 was released, and I don't see any stopping sign, so I think I'll surpass *you* in the following years (and by *you* I mean the person right now reading this, not who you'll be at that point, because if you keep improving and getting tougher to beat I might never surpass you.)
Also, I don't think I've lost two or more games the same way again, I learn from my mistakes, so it means the next game I lose will be in a new and surprising way. witchesbutt at Rybka forum has been playing games with me, and he believes his bookmaker could easily wipe the floor with me. If that's true, it'd mean he plays better than you, so who knows if he could also beat you and other top players. We'll never know unless games happen.
|
|
|
Post by Ozymandias on Sept 7, 2019 19:13:10 GMT
If you knew how to beat them, you'd be doing it already. The question wasn't "how" but rather what indicators made you think that way.
The time travel argument, about future developments giving a sufficient edge over present or past computer resources can't be directly countered, but if you're really curious, you could play some self games, having white and current resources on one side, and 5 year old SF + HW tools (for example) on the other.
Playing from opening positions already stablished 5 years ago, I'm sure you'd find that the latest advances, don't correlate with results at long time controls, specially correspondence chess like.
|
|
|
Post by Vytron on Sept 8, 2019 19:06:23 GMT
I don't need to play against myself to see that, all my analysis is still there, and I have talked about it on Rybka Forum, calling it "analysis obsolescence". So my analysis from 2016 is crap, and my analysis from 2017 is becoming crap by the minute. I guess my analysis from 2018 is still decent, but it seems some major release from Stockfish could easily make it crumble. If trends continue, the analysis I'm doing today will be crap by 2021, and I have seen acceleration on this, specially with Leela around, that even at Depth 8 is capable of busting some lines.
But this isn't about me, the ICCF allows me to download all their games from the last month, and what I've seen is that everyone is in the same boat, top players of 2016 played like crap, and I don't see any difference between GMs or 2100 rated players (pick two at random, chances are the 2100 guy might have played better than the ICCF GM for no reason.) Heck, we were talking at Talkchess recently about how players of TODAY play like crap in the ICCF, which makes my IM title meaningless.
So I'm sure the time travel argument does hold, and I'm a Top player (by your definition) of 2016 or thereabouts, because it's easy to see when today's engine play stronger moves at 12 +2 than those guys at corr chess time control. And I haven't even upgraded my hardware since 2010.
Opening positions established 5 years ago are crap, or even, from 6 months ago are crap. All you need to do to check it is downloading some top free book from January 2019, and one from August 2019, and have them play, and see the slaughter. Unless the new one is a drawbook, happy to go very deep into lines it can't lose, but that it can't win either, but those books just give a wrong impression about what is happening. The results of fast time controls do correlate to long time controls when people are playing as badly as they are on corr games. When they aren't, well, I already said I was stumped trying to beat some FICGS guys, it was you who said you'd beat them the same.
|
|
|
Post by Ozymandias on Sept 8, 2019 19:58:29 GMT
I already said I was stumped trying to beat some FICGS guys, it was you who said you'd beat them the same. Again, "the same" doesn't equate "always". Check my stats, there's no difference among servers. I also said that you wouldn't glean anything from our games, regardless the result, but it sounds like you intend on making a case.
|
|
|
Post by Vytron on Sept 10, 2019 15:32:53 GMT
I didn't expect you to "always" beat me, but you still haven't answered *how often* would you expect to beat me. Like, 1 in 10 games? 1 in 100?...
|
|
|
Post by Ozymandias on Sept 10, 2019 16:12:38 GMT
I didn't expect you to "always" beat me, but you still haven't answered *how often* would you expect to beat me. Like, 1 in 10 games? 1 in 100?... Never, the same way I don't expect you to lose any game moving forward.
|
|
|
Post by Vytron on Sept 10, 2019 23:00:11 GMT
That's incredible. I guess I'll post on here the next time I lose a game, which by my estimation would happen in about 9 months. If I don't lose a game in the next year and a half, I guess I'll concede, expecting to face people already as strong as I'll be 5 years from now, if they can't touch me, improving has lost any meaning, and I've been producing "perfect moves" for a while. We'll see...
Anyway, thanks for the games, I learned a lot, specially about my nonsense about "being able to complicate things at will", which you trivially proved wrong. Though, if I'm at the end of the line, I guess it will not matter XD
|
|
|
Post by Vytron on May 10, 2020 12:51:34 GMT
I didn't expect you to "always" beat me, but you still haven't answered *how often* would you expect to beat me. Like, 1 in 10 games? 1 in 100?... Never, the same way I don't expect you to lose any game moving forward. Well, here's a game I just lost: 1. e4 c5 2. Nf3 Nc6 3. d4 cxd4 4. Nxd4 Qc7 5. Nc3 e6 6. Be3 a6 7. Qd2 b5 8. O-O-O Nxd4 9. Bxd4 Bb7 10. f3 Bc6 11. Kb1 b4 12. Ne2 Nf6 13. c4 bxc3 14. Bxc3 Be7 15. Nd4 Bb7 16. Bb4 Qb6 17. Bxe7 Kxe7 18. Qe3 Rhb8 19. Qa3+ Ke8 20. Be2 Rc8 21. Rhe1 h5 22. Bf1 Rab8 23. Nb3 Bc6 24. Qxa6 Qg1 25. h3 Ra8 26. Qd3 Qa7 27. a3 Rcb8 28. Rd2 Bb5 29. Qc3 Bxf1 30. Rxf1 d5 31. Ka2 Rc8 32. Qd3 dxe4 33. fxe4 Qa6 34. Qxa6 Rxa6 35. Re2 Ra4 36. Nd2 Rd4 37. e5 Nd7 38. Nf3 Rd5 39. Rff2 Ke7 40. Rc2 Rxc2 41. Rxc2 g5 42. Nxg5 Rxe5 43. Nf3 Re4 44. b4 Nb6 45. Kb3 Nd5 46. Ka4 Re3 47. b5 Kd7 48. Nd2 1-0 And this was won by some 2163 rated player on FICGS! The guy ran over me, I don't even know what happened, but I'm nowhere near infallibility.
|
|
|
Post by Ozymandias on May 15, 2020 17:43:11 GMT
Let me analyze that and we'll see when (as in what move/date) you messed up.
Edit: the opening isn't exactly what I'd play, but the real problem was when you gave up the a6 pawn. Those two connected passers were pretty much game over. When did you play move 23?
|
|