Usually, in chess, professionals will write about trends and novelties in a chess opening, and for that, they use GM games and maybe sometimes corr games.
However, since about 2004 (or thereabouts), many or even a majority of the important developments in the Najdorf have their origin in computer chess related activities: The engine room in playchess, engine tournaments, engine books, freestyles.
But nobody has recorded these developments, even though many people must know about these things. It's a bit sad, since the accomplishments by a whole community of Najdorf freaks thus never get any recognition.
So let me start here with a few questions:
1) Who are the main "players" in the engine room? I know Eros must be a super expert. Are there others like him, but more under the radar? Who contribute to theory? How do they do it?
2) What are some major events, new developments and trends in recent years? For instance, we have the Rybka 3 and Rybka 4 books that no doubt had a big impact. What else? We can add more and more stuff, down to specific novelties.
I am not asking people to reveal cutting edge stuff (or even unplayed improvements). I am more interested in recording stuff after the event. At any rate, new stuff only stays new and super valuable for a brief time.
1) Who pioneered the 6. Be3 e5 7. Nb3 Be6 8. h3 + 9. f4 system?
2) What about the 6. h3 system, was it Jiri who popularized this?
3) The whole line 6. Be3 e6 was at some point super popular. Then white found out how effective the unforcing a3-move was. When did that become a trend?
There are so many things worthy of notes. Who, and when, and how?
In PP, was it you or Jeroen who pioneered 10.e5? (Or someone else?)
I think it was you who pioneered the h6+g5 defence, was it not? I remember Arno being quite derogatory about it when kibitzing a Freestyle, six I think
All credits have to go to Dagh, no doubt about it! Sometimes he would sent me files with analysis on this line and I could only stare at my screen in amazement. By the way, it appeared we both got interested in the 10.e5 line after seeing Motylev-Anand from Wijk aan Zee, 2008. Although Motylev lost the game, his position was much better than most people realised. So there was room for improvements. In the Freestyle tournament played in April 2008 I was lucky to get the Poisoned Pawn twice and I could use one of Dagh's super inventions. I had an IM at home assisting me and the guy was totally impressed by this incredible preparation by Dagh :-).
> we both got interested in the 10.e5 line after seeing Motylev-Anand from Wijk aan Zee
That's an interesting tidbit! http://www.chessgames.com/perl/chessgame?gid=1444572 Who'd have thought that game would have led to the consumption of so much electricity. Are humans still leading the way then?
We should totally make a commented list of notable games in the 10. e5 system from those years. It all happened so fast.
10...h6 42.5%, 57.5% draws - played ~68%
10...dxe5 30.9%, 34.7% draws - played ~30%
all others 33.3%, 41.9% draws - played ~2%
Following the main line, here's what you get at move 12:
12...g5 43.6%, 58.7% draws - played ~80%
12...Nd5 43.7%, 65.4% draws - played ~8%
12...Nfd7 23.2%, 33.1% draws - played ~12%
Assuming black wants to smother the game and just draw, 12...Nd5 is the best choice.
I agree that maybe 12...Nd5 is the most drawish like move, and maybe the most suitable for easy prep for a human. I think that a lot of the losses in the 12...g5 line are due to white "tricks" that people were not booked against for a while in the engine room. Once you do the leg-work, with access to a batch of representative games (so you don't miss any tricks), 12...g5 seems a pretty safe book/engine draw. I don't think there are any tricks left for white there, but such sentiments have been refuted before, of course.
I think I was the first one to play the h6 + g5 defence, in our Flying Fatman team in Freestyle 6. I think Arno just said something like the move didn't impress, as in, he just considered it dubious at first glance. Nothing wrong with that, though it would also have been nice to show some respect for the move and possible effort behind it.
So, my story behind h6 + g5:
1) Jeroen and me were investigating the 10. e5 system, sharing knowledge and ideas. At some point, one of us mentioned the 10...h6 move, can't remember who, but I think it was Jeroen. I began to look into it, with a view to refuting it, I think this was mainly in order to prepare the Rybka tournament book since an unaided engine might prefer the move. But also just for Freestyle and for completeness.
2) It turned out there was no immediate clear-cut refutation, so I looked more and more into it, it became almost an obsession. Either the move was good, or I should know the refutation. So I made an endless game-file with my analysis. Note, it is not clear at first sight what moves are the best candidates, for white as well as for black.
3) Shortly before Freestyle 6, I sent this file to you and Majd, to show you some example work. We had great fun in those days, planning our Freestyle weekend. I guess I also shared this stuff with Jeroen, just to keep him updated while he looked at other ideas.
4) We played it in game 8 against Kreutæasjdæakjsdfj (name impossible to spell). The whole draw in that game was in my prep.
5) I think the game and following developments show something typical: Once a comfortable line/move-order for black has been found, people will adopt it, and then alternatives will not be given much notice. So maybe it is not apparent how much work was behind the line we played, with Qa3-Qa5-Qg5. The thing is, there are many, many permutations of the moves Qa5, Nd7, h3 and others, but only one move order works well for black. However, the refutations of black's "wrong" move orders have not really been shown (afaik).
6) An amusing anecdote: Anand faced the move as white in one of its first high-level outings, a rapid game, and lost.
7) Eventually, GMs, corr players and not least playchess people adopted it as the main defense against 10. e5.
8) Within the next 2-3 or so years, the line was probably played to death (draw) in that all white "tricks" and black defenses were found and shared. So 10. e5 with the new ideas gave 6. Bg5 a big revival, but after h6 + g5, 6. Bg5 is losing some of its popularity once again.
Once again you have to get the credits Dagh, as you came up with the 10... h6 idea! Not me :-). I still remember the lines looked just incredible to me, but no refutation could be found. So IMO the line should be named the 'Dagh Nielsen' line! Today it is still not refuted :-).
Reads like "Kreutzfahrtschiff", the nick used also here in the forum by Herbert Kruse (FICGS).
Edit: Corrected, sorry Herbert.
Hartwig is an amateur :)
Maybe someday when I retire from computer chess you will be my successor and inherit this treasure house. Warning, though: it is a serious lifestyle decision, kind of like taking vows of celibacy and dressing in a simple monk's clothing.
In fact, it is a LOT like that.
Pie Iesu domine,
dona eis requiem
while hitting yourself on the head with a plank...
Among computers, the first instances I have are from November 2006 when it was seen five times. Eddief-Frayer appear to be the pioneers on 27.11.2006. But it still didn't catch on: no games in 2007, only two in 2008, none in 2009.
There were two more games in early 2010, but the breakout event was without question Intagrand (Anson Williams) playing the opening 11 times with mixed results on 16.10.2010. Then KERCH followed the next month with eight games, winning four, establishing the opening in the repertoire.
Why would Anson play a good and unknown system 11 times in a single day?
Vas, Larry and I talked a bit about Freestyle chess in Mexico. Larry did not know much about Freestyle yet, and I think Vas and I tried to convey that there really was something very special that Anson added to the game/analysis, even though he was low-level in OTB chess.
It would be great if Anson at some point explained some of his decision processes from his games. I analysed several of his/yours games and was awestruck. Not just by the general "efficiency level", but by the width of chess concepts employed. And these concepts certainly looked like "real" chess concepts. But how did Anson arrive at them? I am wondering about the balance of logic, pragmatism, decision making efficiency, but also wondering if Anson has cultivated a high level chess understanding that complements engines well (like in, engines do the bean-counting, while Anson understands, say, if an exchange sacrifice offers good winning chances, or if lines of certain types are worth extending because he smells a win or a draw based on the positions in the line).
To put it in a more general, provocative way: Can you, in combination with an engine, develop a high level chess understanding (Rxc3 is promising!, since eval is 0.xx and decreasing + this and that feature on the board) even if you couldn't solve a mate in two if your life depended on it?
Please note that, in the above, I don't try to convey any view about Anson's set of chess abilities. I simply don't know and have often wondered much about is, which is exactly why I am now taking the liberty of asking you!
The first thing is that he is another person about whom it can be said that "his ways are not our ways". He's sui generis. What can you really say when you are a pretty high-IQ person relative to the general population, but you encounter somebody who is incomparably brighter, not just in one topic but almost across the board? A true polymath? That's him, and he's very unassuming, modest, laconic, even shy.
He's a terrific computer programmer. He kart-races and dominates the racetrack. He plays J.S. Bach with astonishing virtuosity with no sheet music in front of him. He is murderous in Freestyle chess. He may have other hobbies and interests that I don't even know about. He's very devout and observes the Sabbath to the greatest extent he can.
I met Anson in the Playchess room in October 2004. At the time I was showing early but half-baked interest in kibitzing games. I'm not sure how we actually connected, but I remember that in our first conversation I was skeptical of the things he was saying. He had a kibitzer, he said. Really? That was practically unheard-of at the time. I asked him to prove his bona fides. He sent me one day's output. It was staggering. Our relationship rapidly became an exclusive partnership. We exchanged a lot of ideas; in those days I was like an uncapped fire hydrant, writing down lengthy ideas daily. Most Anson would ignore--it turns out, most of them were non-essential tangents. A few intrigued him and he programmed utilities and GUIs and stuff that formed the foundation of what we still do to this day.
Anson played quite a bit of Playchess during 2004 and the Freestyle years. Less now. We spent almost zero time on bookmaking in the traditional sense. We just collected and processed games and put the ones we wanted into an opening book, which we further embellished with evaluations, which then resulted in move-rankings. In essence we were doing some of the basic Aquarium things in 2005.
Our concept in competition was pretty straightforward: let the book get us to a safe position, or a winning position if the opponent blundered, and then let our decision-support system do the rest. What was our decision-support system? It varied from tournament to tournament. Anson might have seven computers grinding away on four or six engines, and his girlfriend was assisting in a very real way, calling out things she was seeing and exploring. You had two skilled people working the search using different engines, and both of them were very, very fast in terms of digital dexterity and constantly looking far ahead. But there was a secret sauce, too, and that was Anson himself. The guy had an almost supernatural ability to detect danger, and if he sensed it, he would just take his time as cool as can be and find a way to avoid losing and maybe turn the game around. I'll never forget that C11 game against the Czechs where we were dead lost and he took 31 minutes (!) to figure out a way to draw, a way which was subsequently busted, but beyond the ability of the Czechs to figure out the bust in the time allotted. Can you imagine what a total panic attack I was having on my end, not knowing what was going on in London? That kind of thing happened often! I would think we were in a bad spot, there'd be a long pause, and wham--he'd win! Let me tell you, for me, a hypercompetitive guy with no actual talent for competition, this was ecstasy. Some of the biggest thrills of my lifetime.
I think your questions about how Anson thinks are very intriguing but beyond my ability to describe them. I mean, he wasn't all that different from you; the big differences were a) our book, b) our general strategy, and c) divine inspiration--or something.
"Anson played quite a bit of Playchess during 2004 and the Freestyle years. Less now. We spent almost zero time on bookmaking in the traditional sense. We just collected and processed games and put the ones we wanted into an opening book, which we further embellished with evaluations, which then resulted in move-rankings. In essence we were doing some of the basic Aquarium things in 2005."
I had a guess that you had been doing exactly that due to some things you mentioned. I remember you saying that you also made a survey of how closely engine evals correlated to results (very well, obviously, if I remember correctly, but quantifying is interesting). The thing I wondered about was how far you could take it without server farms, or if you just did it to shallow depths and/or in selected lines. The CAP of Convecta is interesting, but not really precise enough to base decisions on. Of course, if you couple it with game results, you can do some stuff. For instance, create an aggregate score and min-max it.
"We exchanged a lot of ideas; in those days I was like an uncapped fire hydrant, writing down lengthy ideas daily."
I think many of us have shared that excitement. I was for a few years regularly spending time thinking and fantasizing about creating this unbeatable book/chess entity, with hundreds and thousands people using thousands and thousands of cores + human input to create it, and I would ponder many different approaches and obstacles. Of course, it's a totally silly fantasy ("chess!?!?"), but it is intellectually stimulating for me. It is also a standard question from newcomers to chess forums, like in, "have computers solved chess???". Maybe it is an unrealistic target, but once we wise up, we may already be too engrossed in it all.
Note that it is not apparent that it is a hard thing to accomplish. The 1st April fool joke about Vas and the King's Gambit was slashdotted, and most of the early posters on Slashdot took it at face value, discussing it in earnest. These people are computer geeks!
Going out on a tangent:
If persons A and B said to you:
1) Checkers is solved, Go is too hard for now.
2) Go is solved, Checkers is too hard for now.
How the **** would you know who was right and why, if not having studied the subject?
Consider the following simplified poker game ("slightly" more complicated that the normally discussed AKQ game, but infinitely closer to the real thing):
X is out of position.
Y is in position.
X and Y has either A, K, or Q with equal frequency.
Pot is 1.
Min-bet is M.
Pot-stack-ratio is P.
One street of normal no limit betting, so X can check-raise etc. Highest card wins.
Has this been solved? What is the optimal strategy?
What is the expected value of X and Y? In other words, what is the advantage of being in position here, if any?
I searched around, but could not find anything. People were writing articles and theses on the much simpler game where the bet size is fixed, and X can only bet or check, while Y can only call or fold, so action is closed. With this severe limitation, you can just do straight forward (but somewhat lengthy) calculations to arrive at the answer. But in the rules posted above, the betting space is practically "infinite" (no fixed bet size), even more so depending on the min-bet and the pot-stack-ratio and the option of re-raises. I think it is very hard to figure out the equilibrium, if possible (probably you can do simulations or whatever). And this is just with AKQ!
As said, going out on a tangent. Maybe someone will have something to offer about that poker stuff.
At any rate, we can probably be excused for indulging in "crazy" chess dreams.
Can you post that game with Anson you mentioned? (and critical juncture)
> I was for a few years regularly spending time thinking and fantasizing about creating this unbeatable book/chess entity, with hundreds and thousands people using thousands and thousands of cores + human input to create it, and I would ponder many different approaches and obstacles. Of course, it's a totally silly fantasy ("chess!?!?"), but it is intellectually stimulating for me.
Is this a Jedi mind trick or a John Kerry-style gambit of some kind? Totally silly fantasy? What are you talking about, it's totally feasible as long as you scale back the vision a bit and size the project appropriately. The big issue has always been money. You need strong hardware and you need total control over it, i.e. ownership. Over time, miracles could be accomplished with just a couple of 32-core AMD servers (cost: about $20,000 for both). The basic framework for a massive book already exists. Now it's just a matter of kicking off a funded project and seeing it through. I've considered www.kickstarter.com but Anson doesn't seem interested.
> I've considered www.kickstarter.com
Wondered what this was so checked it out and found this.
I think I really understand the personality you are describing here.
I think the best way to understand him is to watch/listen to him play Bach. He sent me a video of himself playing several Goldberg Variations and all I can say is that I was moved. No, there are definitely things you or I will never understand.
However, some of these characteristics are part of certain and very well known profiles of personality. I have the chance to know some people who can also be somehow classified into these profiles, at least partly... Even sometimes I get myself trapped into some of them...
Among engines you see the move being played hundreds of times in 2000-2004 in matches including lots of engines of that period. Looks to me like one of the popular opening books of that time must have had it as a featured move.
By the time Jiri came along the move had been seen over a thousand times in different places.
Among humans the move dates to the 1960s but it is seen rarely and its proponents have never been very strong. I can't find any GM-level games with this move.
8.a3 62.2%, 36.0% draws
9.a3 58.8%, 37.9% draws
10.a3 74.9%, 26.3% draws
9...b4 has to be played instead of the more-common 9...Nbd7. There is a performance gulf there.
And does 6...e6 fare much worse than ...e5?
2. 6...e5 45.0%, 43.4% drawn; adjusted for ELO 45%
6...e6 42.3%, 32.6% drawn; adjusted for ELO 44%
So, looking at it with a big panoramic view and completely missing all the granular detail, it looks like 6...e5 is more drawish and slightly better overall, while 6...e6 carries more risk but also a better probability of a decisive result.
> So, looking at it with a big panoramic view and completely missing all the granular detail, it looks like 6...e5 is more drawish and slightly better overall, while 6...e6 carries more risk but also a better probability of a decisive result.
Couldn't have been said any better.
I like the idea behind this thread, by the way (even though I find the obsession of large parts of the computer chess community with the Najdorf hard to understand).
There have been tremendous developments in this opening due to a huge number of games played in the engine room. What seems to be missing, though, is a systematic account of these developments which links them with 'human theory' and its history.
> I was under the impression that everyone is playing ...e6 now.
Hardly. Among 2013 engine games (Playchess, plus everything else I round up from numerous sources) 6...e5 is 20x more common. Among human players about 2.5x more common. Anand, Topalov, Kajarkin, Gelfand seem to prefer e5 in recent years.
> What seems to be missing, though, is a systematic account of these developments which links them with 'human theory' and its history.
It's one of the tragedies of computer chess that I am sitting on all this opening theory and I don't actually know the game at all. You see a position and see piece development, light squares, dark squares, pawn chains, king threats. I see numbers. Just about anybody on this site could do a better job of realizing your idea than I could if they had the same information I have.
>> I was under the impression that everyone is playing ...e6 now.
Sorry, this was a typo. I meant to say e5.
> It's one of the tragedies of computer chess that I am sitting on all this opening theory and I don't actually know the game at all. You see a position and see piece development, light squares, dark squares, pawn chains,
> king threats. I see numbers. Just about anybody on this site could do a better job of realizing your idea than I could if they had the same information I have.
Maybe you should look for a co-author and write a book.
Chapter 6, page 12, recommend a full reading if topic is of interest.
Powered by mwForum 2.27.4 © 1999-2012 Markus Wichitill