As a strong correspondence player, have you played against a top computer at a 24H+ time control? If not, you're the ideal candidate!
Correspondence ICCF Champion Vs Top Engine ?
Moderators: hgm, Rebel, chrisw
-
- Posts: 11589
- Joined: Thu Mar 09, 2006 12:57 am
- Location: Birmingham UK
Re: Correspondence ICCF Champion Vs Top Engine ?
As a strong correspondence player, have you played against a top computer at a 24H+ time control? If not, you're the ideal candidate!
Writing is the antidote to confusion.
It's not "how smart you are", it's "how are you smart".
Your brain doesn't work the way you want, so train it!
It's not "how smart you are", it's "how are you smart".
Your brain doesn't work the way you want, so train it!
-
- Posts: 47
- Joined: Sun May 23, 2021 6:04 pm
- Full name: Jacques Ress
Re: Correspondence ICCF Champion Vs Top Engine ?
No, i did not ! Or, if one on my opponents turned his machines for very Long Time, i did but i don't know ???
However, we speak here about engines wihout Book and, as i underlined above, such engine tend to play Always 1.e4 e6 ( perhaps disable NNUE for a little test )
Seem engines have now a little Book integrated with some moves of Berlin, Italian, Sicilian, QGD etc
Not enormous but enough for start in a solide way !!!
IM ICCF player
-
- Posts: 4190
- Joined: Wed Nov 25, 2009 1:47 am
Re: Correspondence ICCF Champion Vs Top Engine ?
You are totally misinformed. Please go and check CerebellumMerged. You'll be surprised that it's actually stronger than a book one can make from high level ICCF games. And there you can see the clear preference for both SF-NNUE and Lc0 is Spanish.jr66 wrote: ↑Fri May 28, 2021 3:46 pm However, we speak here about engines wihout Book and, as i underlined above, such engine tend to play Always 1.e4 e6 ( perhaps disable NNUE for a little test )
Seem engines have now a little Book integrated with some moves of Berlin, Italian, Sicilian, QGD etc
Not enormous but enough for start in a solide way !!!
-
- Posts: 4190
- Joined: Wed Nov 25, 2009 1:47 am
Re: Correspondence ICCF Champion Vs Top Engine ?
Btw. since it's Friday afternoon and I have a bit of time, I can demonstrate you a few things. For example your last proper (not by opponent mistyping the move) win in ICCF from more than 2 years ago (where you got your IM norm) was due to your opponent (countryman) being unable to see that 17...g6 is loosing. I doubt it was your deliberate trap, more that just by luck you stumbled into that position.
Now try analyzing it with the latest SF. You'll see that in roughly 10-20min (depending on the hardware it runs on) it switches to 17...Bxe3 and stays permanently there. So this kind of win would be impossible today if one just leaves an engine analyzing it for half an hour.
Actually the deepest "trap" I found in recent ICCF games was in one of very few WC30 decisive games that was not decided by the simplest blunder (mistyped position) where it took SFdev a depth of 65 or close to 12 hours of analysis to find that the played move is loosing.
I've never encountered a position where SFdev wouldn't discoverer it as lost after 120h of analysis. And I am using just a very slow (and ancient) 16 core machine. And bear in mind this is just a dummy analysis. One can ofc double check the things by at least exploring a bit the line after a chosen move and then it quickly becomes obvious that the chosen move is bad.
-
- Posts: 47
- Joined: Sun May 23, 2021 6:04 pm
- Full name: Jacques Ress
Re: Correspondence ICCF Champion Vs Top Engine ?
Sorry but i am afraid we don't speak about the same thing ?Milos wrote: ↑Fri May 28, 2021 5:54 pmYou are totally misinformed. Please go and check CerebellumMerged. You'll be surprised that it's actually stronger than a book one can make from high level ICCF games. And there you can see the clear preference for both SF-NNUE and Lc0 is Spanish.jr66 wrote: ↑Fri May 28, 2021 3:46 pm However, we speak here about engines wihout Book and, as i underlined above, such engine tend to play Always 1.e4 e6 ( perhaps disable NNUE for a little test )
Seem engines have now a little Book integrated with some moves of Berlin, Italian, Sicilian, QGD etc
Not enormous but enough for start in a solide way !!!
For what i understood, it seem this topic is made for a ICCF player vs engine alone wihout Book match ?
I don't search a Book at all ! I just underline above a little Book can be incorporate in the new engines and so, it would not be a no book engine match ? ( Seem already difficult with no Book because i don't use great hardware and never sleep on my laptop in waiting a move during 24h. )
IM ICCF player
-
- Posts: 4190
- Joined: Wed Nov 25, 2009 1:47 am
Re: Correspondence ICCF Champion Vs Top Engine ?
You really don't know what CerebellumMerge book is, do you? Huh.jr66 wrote: ↑Fri May 28, 2021 6:13 pmSorry but i am afraid we don't speak about the same thing ?Milos wrote: ↑Fri May 28, 2021 5:54 pmYou are totally misinformed. Please go and check CerebellumMerged. You'll be surprised that it's actually stronger than a book one can make from high level ICCF games. And there you can see the clear preference for both SF-NNUE and Lc0 is Spanish.jr66 wrote: ↑Fri May 28, 2021 3:46 pm However, we speak here about engines wihout Book and, as i underlined above, such engine tend to play Always 1.e4 e6 ( perhaps disable NNUE for a little test )
Seem engines have now a little Book integrated with some moves of Berlin, Italian, Sicilian, QGD etc
Not enormous but enough for start in a solide way !!!
For what i understood, it seem this topic is made for a ICCF player vs engine alone wihout Book match ?
I don't search a Book at all ! I just underline above a little Book can be incorporate in the new engines and so, it would not be a no book engine match ? ( Seem already difficult with no Book because i don't use great hardware and never sleep on my laptop in waiting a move during 24h. )
It's a book created with SF by exploring its tree and also to a smaller extent with Lc0. Then these two books are merged into one. The book moves reflect exactly what SF produces as the best move in each position if left to analyze long enough, but in practice couple of minutes is enough.
And there you can see that best line SF chooses is Spanish and it is exactly what SF with no book would play today. So this French misconception probably reflects how outdated your knowledge is.
-
- Posts: 47
- Joined: Sun May 23, 2021 6:04 pm
- Full name: Jacques Ress
Re: Correspondence ICCF Champion Vs Top Engine ?
As You noted, i did not play since two years and i lost in a PC all my analysis but, for what i remember, i was very happy to play 18.h4 ( i don't remember if i saw already the win or just a big advantage but i saw it quickly few moves later....Milos wrote: ↑Fri May 28, 2021 6:06 pmBtw. since it's Friday afternoon and I have a bit of time, I can demonstrate you a few things. For example your last proper (not by opponent mistyping the move) win in ICCF from more than 2 years ago (where you got your IM norm) was due to your opponent (countryman) being unable to see that 17...g6 is loosing. I doubt it was your deliberate trap, more that just by luck you stumbled into that position.
Now try analyzing it with the latest SF. You'll see that in roughly 10-20min (depending on the hardware it runs on) it switches to 17...Bxe3 and stays permanently there. So this kind of win would be impossible today if one just leaves an engine analyzing it for half an hour.
Actually the deepest "trap" I found in recent ICCF games was in one of very few WC30 decisive games that was not decided by the simplest blunder (mistyped position) where it took SFdev a depth of 65 or close to 12 hours of analysis to find that the played move is loosing.
I've never encountered a position where SFdev wouldn't discoverer it as lost after 120h of analysis. And I am using just a very slow (and ancient) 16 core machine. And bear in mind this is just a dummy analysis. One can ofc double check the things by at least exploring a bit the line after a chosen move and then it quickly becomes obvious that the chosen move is bad.
The strange thing is 17...g6 is yet the best move in a well known book i will not quote !
Perhaps a better exemple of my analysis is this draw :
[Event "FRA/C2017 (FRA)"]
[Site "ICCF"]
[Date "2017.02.01"]
[Round "?"]
[White "Ress, Jacques"]
[Black "David-Bordier, Christophe"]
[Result "1/2-1/2"]
[BlackElo "2318"]
[ECO "C12"]
[WhiteElo "2279"]
1.e4 e6 2.d4 d5 3.Nc3 Nf6 4.Bg5 Bb4 5.e5 h6 6.Bd2 Bxc3 7.bxc3 Ne4 8.Qg4 Kf8 9.Bd3 Nxd2 10.Kxd2 c5 11.h4 Nc6 12.Qf4 c4 13.Bg6 Qe7 14.Ne2 Ke8 15.Bh5 Bd7 16.a4 b6 17.Bf3 a5 18.Rhb1 Rb8 19.g4 Kd8 20.Qg3 Kc8 21.Nf4 g6 22.Kc1 g5 23.hxg5 Qxg5 24.Kb2 Na7 25.Bxd5 exd5 26.Nxd5 b5 27.Nf6 bxa4+ 28.Ka2 Nb5 29.f4 Qg6 30.f5 Bxf5 31.Rxb5 Rxb5 32.Qf3 Bd7 33.Qa8+ Kc7 34.Qa7+ Rb7 35.Nd5+ Kc8 36.Ne7+ Kd8 37.Nxg6 Rxa7 38.Nxh8 Bxg4 39.d5 Ke8 40.d6 h5 41.Rf1 Bc8 42.Rf4 Be6 43.Rf6 Bd5 44.Rh6 Rb7 45.Rxh5 Kf8 46.Rg5 Be6 47.Ka3 Rb5 48.Kxa4 1/2-1/2
I was not happy with my game and searched to draw ! Then, i saw i could play 24.Kb2 because i had calculated the draw after the feared 24...Na7 with the bishop sacrifice !
I admit it is not always Sunday for find a such move in 2017 but it happen...
I am not simply an engine BUT i admit also engines are now so strong what wins in CC are very very difficult and, most of time, simply an opponent big mistake ???
Regards
Jacques who hide nothing as you see
IM ICCF player
-
- Posts: 47
- Joined: Sun May 23, 2021 6:04 pm
- Full name: Jacques Ress
Re: Correspondence ICCF Champion Vs Top Engine ?
In fact, it is not important for us if a book is in SF or not because we don't use it ! It is why both i am not interested by that and the reason i did not doubt you knew much better engines than me !Milos wrote: ↑Fri May 28, 2021 6:56 pmYou really don't know what CerebellumMerge book is, do you? Huh.jr66 wrote: ↑Fri May 28, 2021 6:13 pmSorry but i am afraid we don't speak about the same thing ?Milos wrote: ↑Fri May 28, 2021 5:54 pmYou are totally misinformed. Please go and check CerebellumMerged. You'll be surprised that it's actually stronger than a book one can make from high level ICCF games. And there you can see the clear preference for both SF-NNUE and Lc0 is Spanish.jr66 wrote: ↑Fri May 28, 2021 3:46 pm However, we speak here about engines wihout Book and, as i underlined above, such engine tend to play Always 1.e4 e6 ( perhaps disable NNUE for a little test )
Seem engines have now a little Book integrated with some moves of Berlin, Italian, Sicilian, QGD etc
Not enormous but enough for start in a solide way !!!
For what i understood, it seem this topic is made for a ICCF player vs engine alone wihout Book match ?
I don't search a Book at all ! I just underline above a little Book can be incorporate in the new engines and so, it would not be a no book engine match ? ( Seem already difficult with no Book because i don't use great hardware and never sleep on my laptop in waiting a move during 24h. )
It's a book created with SF by exploring its tree and also to a smaller extent with Lc0. Then these two books are merged into one. The book moves reflect exactly what SF produces as the best move in each position if left to analyze long enough, but in practice couple of minutes is enough.
And there you can see that best line SF chooses is Spanish and it is exactly what SF with no book would play today. So this French misconception probably reflects how outdated your knowledge is.
I just did not accept your affirmation we observe only engines wihout calculations...
I asked for the book included because i think i will cannot beat an engine working 24/24 if he has this advantage too....just because win possibility decrease strongly if openings are good !!!
IM ICCF player
-
- Posts: 47
- Joined: Sun May 23, 2021 6:04 pm
- Full name: Jacques Ress
Re: Correspondence ICCF Champion Vs Top Engine ?
A last note : you are off course right about Spanish preference for SF NNUE but i made a little test by disabling NNUE and i found again the French on mono PV!
Can you check for confirm please ?
Can you check for confirm please ?
IM ICCF player
-
- Posts: 4190
- Joined: Wed Nov 25, 2009 1:47 am
Re: Correspondence ICCF Champion Vs Top Engine ?
Correct, SF classical still prefers French, simply because SF classical didn't change almost at all since NNUE was included. But NNUE is a huge improvement particularly on the opening. With SF classic one would probably need some small book to avoid some rare losses due to bad play in the opening. With SF NNUE that is not the case any more.