When will the chess programmers write an engine that plans ?

[Tibono]

Most GMs have about 250,000 pattern chunks in their memory, and this gives them the intuition to find the right idea with very little calculation at all.

Hello guys, hi Brendan, isn't this exactly what strong AI NN achieve? LC0 + big NN is impressively strong when set to "1 node" level, meaning without any in-depth search (playing best move according to NN policy). Okay, that is still no evidence of any plan, rather playing in an smart adaptive way with regards to the current position...

[BrendanJNorman]

Hello guys, hi Brendan, isn't this exactly what strong AI NN achieve? LC0 + big NN is impressively strong when set to "1 node" level, meaning without any in-depth search (playing best move according to NN policy).

Yes. And this is why the play of NN engines always appear much more "human-like" than that of AB engines.

Thorsten is right about this IMO.

Using a NN much more closely simulates the human brain, so obviously play is going to reflect this.

Okay, that is still no evidence of any plan, rather playing in an smart adaptive way with regards to the current position...

I think if we look at a well-played Lc0 game, there is ample evidence of planning. At least what highly resembles planning...which is the best we can do until we have real, sentient machines.

[Ovyron]

If you knew how to beat 2300 players, you wouldn't be 1500. Period.

LOL you got the wrong guy, the last 2300 elo player I defeated happened in March.

So you don't even know who I am? Who's this Oly guy?

I'll stop discussing with you because you don't play the board, but the opponent, when you discuss with them, and I'd be getting completely different responses from you if I was 2400.

It was funny to see you bringing data from some random guy you thought was me, though!

I defeated a 1800 rated player 2 days ago, guess you can't find that game? You need to improve your researching plans!

[BrendanJNorman]

LOL you got the wrong guy, the last 2300 elo player I defeated happened in March.

You telling me this isn't your account?

https://lichess.org/@/Ovyron/perf/blitz

Strange, since I still have chat history in my inbox with you there.

I'll stop discussing with you because you don't play the board, but the opponent, when you discuss with them, and I'd be getting completely different responses from you if I was 2400.

Of course you would. Because if you were 2400, you would be making more sense. 2400 players know how often they need to calculate variations, for example.

It was funny to see you bringing data from some random guy you thought was me, though!

And even funnier to deny it.

[Ovyron]

Dude, I abandoned that account since ages ago, get on with the times. I'll leave you guessing how many hundred points I've improved since then, because you're incapable of holding a discussion that isn't based on the length of your... elo.

Just curious, how many hundred elo points have YOU improved since I abandoned that account? With the same rate you should be beating GMs by now!

[BrendanJNorman]

I wonder how a game between us would look like, it's possible I'd manage to reach a winning position and blew it up after missing some easy tactic, but at least it'd show I was better on the strategical department to reach it on the first place.

2 years ago this question was already answered, my friend.

[pgn][Event "Rated Rapid game"]
[Site "https://lichess.org/5bmEsYRX"]
[Date "2018.01.20"]
[Round "-"]
[White "ChessNCognac_com"]
[Black "Ovyron"]
[Result "1-0"]
[UTCDate "2018.01.20"]
[UTCTime "07:02:09"]
[WhiteElo "1500"]
[BlackElo "1619"]
[WhiteRatingDiff "+244"]
[BlackRatingDiff "-11"]
[Variant "Standard"]
[TimeControl "360+4"]
[ECO "D43"]
[Opening "Semi-Slav Defense"]
[Termination "Normal"]
[Annotator "lichess.org"]

1. Nf3 d5 2. d4 Nf6 3. c4 c6 4. Nc3 e6 { D43 Semi-Slav Defense } 5. g3 Be7 6. Bg2 O-O 7. O-O b6 8. Bf4 Bd6 9. Ne5 Qc7?! { (0.37 → 1.19) Inaccuracy. dxc4 was best. } (9... dxc4) 10. Rc1 dxc4 11. Nb5? { (1.45 → 0.30) Mistake. Bxc6 was best. } (11. Bxc6 Bxe5 12. dxe5 Qxc6 13. exf6 Nd7 14. fxg7 Kxg7 15. Qd4+ f6 16. Bd6 Re8 17. Rcd1 Kh8) 11... cxb5 12. Bxa8 Nbd7 13. Nxd7 Bxf4?? { (0.39 → 5.08) Blunder. Bxd7 was best. } (13... Bxd7 14. Bxd6 Qxd6 15. Bf3 Bc6 16. e3 Rc8 17. Re1 Bxf3 18. Qxf3 Qc7 19. a3 a5 20. Ra1) 14. Nxf6+ gxf6 15. gxf4 Qxf4 16. e3 Qg5+?! { (4.72 → 6.51) Inaccuracy. Qf5 was best. } (16... Qf5 17. Qf3 Qxf3 18. Bxf3 b4 19. Rxc4 Ba6 20. Rc7 Bxf1 21. Kxf1 a5 22. Rb7 Kg7 23. Ke2) 17. Kh1 Kh8 18. Rg1 Qh4 19. Rg4 Qxf2 20. Rg3 Qf5?! { (5.97 → 9.91) Inaccuracy. Rg8 was best. } (20... Rg8 21. Rc2 Qf5 22. Rxg8+ Kxg8 23. Rg2+ Kf8 24. Qe1 a5 25. Qg3 e5 26. dxe5 fxe5 27. Qh4) 21. Qg1 Bd7 22. Rf1 Qc2 23. Rg2 Qd3 24. Qf2 f5? { (8.98 → Mate in 12) Checkmate is now unavoidable. Qf5 was best. } (24... Qf5 25. Bf3 c3 26. Qg3 Qg5 27. Qd6 Qh6 28. Qxd7 Qxe3 29. bxc3 a6 30. Qd6 Qh6 31. Qxb6) 25. Bb7?! { (Mate in 12 → 10.95) Lost forced checkmate sequence. Rfg1 was best. } (25. Rfg1 f4 26. Qxf4 Qg6 27. Rxg6 hxg6 28. Rxg6 fxg6 29. Qxf8+ Kh7 30. Be4 Bc6 31. Bxc6 c3) 25... Rb8?! { (10.95 → Mate in 7) Checkmate is now unavoidable. Bc8 was best. } (25... Bc8 26. Bc6) 26. Rfg1 f6 27. Qh4 Rf8 28. Qxf6+ Rxf6 29. Rg8# { White wins by checkmate. } 1-0[/pgn]

[BrendanJNorman]

Dude, I abandoned that account since ages ago, get on with the times. I'll leave you guessing how many hundred points I've improved since then, because you're incapable of holding a discussion that isn't based on the length of your... elo.

Just curious, how many hundred elo points have YOU improved since I abandoned that account? With the same rate you should be beating GMs by now!

Don't talk about MASSIVE improvement unless you're willing to prove it with a link to your new account.

To not do so is childish.

It is easy to make unproven claims.

[Ovyron]

If you don't believe someone when they tell you the truth it's your problem. I'm happier knowing that you don't know, and find funny how you think I'm playing at the same level of 2 years ago

I'll be nice and let you see a small part of my new account's graph, though...



How does yours look like? Let me guess, a flat line?

[BrendanJNorman]

If you don't believe someone when they tell you the truth it's your problem. I'm happier knowing that you don't know, and find funny how you think I'm playing at the same level of 2 years ago

I'll be nice and let you see a small part of my new account's graph, though...



How does yours look like? Let me guess, a flat line?

You're just being disingenuous now Uly...Jeez bro. I really didn't want to embarrass you, but this is too much.

I found your new account on the first guess.

- Rating matches up
- Nickname matches up
- Openings match those you've been discussing here (that Italian is a win for black and this account plays Italian exclusively as black, for example)
- Date of new account matches up (or at least has no overlap)

https://lichess.org/@/uly/perf/blitz

Disappointing, man.

Honesty is always the best policy.

[Ovyron]

Nope, it's not that one!

It would have been very cool if the uly account was taken by someone with 2400, it would have blown you away. But alas, you'll continue to live in ignorance...