In accelerated dragon, if white wants to go to Yogoslav attack against classical dragon, can black prevent it?
King's Indian and Accelerated Dragon
Sort:
Essentially yes, that's the advantage of the Accelerated Dragon over the Dragon.
The main line of the Yugoslav Dragon is 1. e4 c5 2. Nf3 d6 3. d4 cxd4 4. Nxd4 Nf6 5. Nc3 g6 6. Be3 Bg7 7. f3 O-O 8. Qd2 Nc6:
The best move here is 9. O-O-O, then black's best response is 9. .. d5. The old main line is 9.Bc4 Bd7 10. O-O-O Rc8 11. Bb3 Ne5 12. h4 h5 but then black equalizes.
If white tries to follow this in the Accelerated Dragon, he'll start with 1. e4 c5 2. Nf3 Nc6 3. d4 cxd4 4. Nxd4 g6 5. Nc3 Bg7 6. Be3 Nf6:
Now 7. f3 will be met with 7. .. O-O 8. Qd2 d5 where black has saved a tempo by playing .. d7-d5 in one move rather than in two moves as in the Dragon (ie. 2. .. d7-d6 and then later 9. .. d6-d5). This is the same position as the Yugoslav after 9. O-O-O d5 except white hasn't castled yet.
White's best continuation here is 7. Bc4 to stop .. d7-d5, then 7. .. O-O 8. Bb3 (not 8. f3 Qb6) d6 9. f3 Bd7:
For a long time everyone played 10. Qd2 here, then black exploits white's slightly premature Bb3 by playing 10. .. Nxd4 11. Bxd4 b5. We actually reviewed this yesterday, Vasik's round 7 opponent has played it twice with white.
Recently, players have started to play 10. h4. If this happens Vasik will transpose to the old Yugoslav Dragon main line by playing 10. .. h5 11. Qd2 Ne5 12. O-O-O Rc8 since that is equal. Black does have some extra options after 10. h4 which he doesn't have in the Yugoslav Dragon with 9. Bc4 but the main thing which the Accelerated Dragon move order accomplishes is avoid the Yugoslav Dragon with 9. O-O-O.
Also, we've chosen the Gruenfeld. I'll learn it properly before we play the second league, if we make it there. Vasik will learn it with his trainers and on his own until then. We may still play the KID on certain occasions or in some move orders.
The Gruenfeld is actually a nice dynamic option. It's a "dark square" opening, like the KID, Dragon, etc, in that black plays on the dark squares and his dark squared bishop plays a key role. The Nimzo and Queen's Indian are "light square" openings, similar to the French, Caro-Kann, QGD, etc, where black plays on the light squares, and the role of his dark squared bishop is to help control the light squares (by pinning a kniight on c3 or breaking a pin against the knight on f6). Dark square openings are (somehow) more dynamic than light square openings.
We looked at a few Gruenfeld games yesterday, there is a lot going on, a lot of sacrifices, etc.
I see. 
So an Accelerated Dragon player must also know the classical dragon, in case white chooses to enter the classical dragon by allowing black an extra tempo (d7-d5)?
Black should understand those types of positions. In most of the Classical Dragon lines, the Accelerated Dragon version is up a tempo. There is one direct transposition, 1. e4 c5 2. Nf3 Nc6 3. d4 cxd4 4. Nxd4 g6 5. Nc3 Bg7 6. Nb3:
This prevents .. d7-d5 for good so after finishing his kingside development black will play 8. .. d6, which will directly transpose into the classical Dragon with Nb3.
Vasik got another KID in round 7:
The opponent never played 3. g3 with white. 80% of his games were 1. e4 and the rest were 1. Nf3 Nf6 2. c4 g6 3. b3 (although only two black players played 2. .. g6 so it's not clear what exactly his intentions were in those 20% of his games).
Vasik wasn't ready with the Gruefeld and decided to play his old defense against the g2-g3 KID. He had only one game in the Polish databases with that defense, from early 2023. Since then he's been playing .. e7-e5 in those positions, with three games in the databases. Unfortunately that one game was enough for the opponent to prepare it pretty well. The trainer must have figured, correctly, that Vasik could go back to that when surprised like this, to try to get out of the preparation.
The opponent blitzed until 12. Qc1, then Vasik finally wiggled out with 12. .. Nfd7, which isn't one of the top five engine moves but looks perfectly playable. A typical KID fight followed. Black sacked a pawn for compensation. The engine continued to prefer white more than me. When I thought it was equal the engine said +1, when I thought black was better the engine said +0. The opponent eventually got low one time, Vasik still had 25 minutes and was trying to win, but actually Vasik made the first blunder and at one point white had a win. Two moves later it was over.
Rip it! Nice finish though.
The Grunfeld is a tough opening for both sides so I think it makes sense for an aspiring player, and it is likely for there to be mistakes he can take advantage of. The problem is learning the basic ways both sides can go wrong.
I think the Grunfeld fits Vasik because he seems to be a concrete tactical player who is comfortable with material imbalances. And possibly nonconcrete evaluations like compensation for a pawn or exchange.
One possible downside is the array of strong attacking schemes. Vasik seems to be able to bounce back in the next game, which might unfortunately happen. At least, from the games posted, it doesn't seem likely he'll face too many wild attackers. More likely get a g3 Grunfeld.
Right. Today's players are by our standards more positional and more materialistic. Vasik doesn't fit this pattern. He didn't really work alone with engines until recently. We'll see. Of course there is a lot to learn in the Gruenfeld.
Here is round 8, with another material imbalance:
Position on move 14, black to move:
We prepared 14. .. e4 15. Nc4 Ne5. Opponent went with 14. .. Nf6 to hit d5. In this kind of position black attacks on the kingside while white pushes on the queenside, which is the opposite of how Vasik's games normally look. He quickly flipped that upside down, unsoundly.
Position on move 26, black to move:
This is more of a typical Vasik position. Computer gives -2. The problem isn't actually the material deficit, it's white's king. If the white king were on h1 instead of f1 it would be equal. The opponent got scared and returned the exchange.
Round 9 is underway:
Vasik got the old Dragon main line, via Accelerated Dragon move order, as I wrote about in my reply to Meng. Position on move 13, white to move:
This is all equal. 13. g4, 13. Bg5, 13. Bh6, 13. Kb1 are all a huge mess, but equal. We looked at it a few days ago, preparing for another game.
Back in 1990 this was iffy for black, wasn't it?
Saw the game. Didn't like what happened, especially 16...Rfc8 17 Bxf6 exf6 At that point I'd be tempted to bail with 17...Bxf6 18 Nd5 Qxd2 19 Nxf6+ Kg7 20 Nxh5+ gxh5 21 Rxd2, which poses some technical difficulties for white (maybe). Is black supposed to play 16...Kh8? or 16...Rb4?
The Krzysztof game was interesting because it shows the possibilities when the opponent just doesn't know for sure, in this case what happens if he doesn't give back the exchange. Also, I would have tried winning at the end with black, although maybe I am playing too much blitz.
Yes, I think it was seen as a bit iffy. It was played mostly by second-tier GMs. Kasparov did play it twice against Anand in their WC match but never played it again.
I liked Vasik's 17. .. exf6 and the engine does too. He knows the 19. Nxf6+ Kg7 idea, after 20. Nxh5+ black can also play 20. .. Kh6, but those endgames are clearly better for white. 17. .. exf6 is +0.0 and from a human point of view I prefer black.
21. .. Rxc3 was just terrible, position:
I was scratching my head about what Vasik missed, settled on 22. Qxd7 Rxc2 23. Qxa4 (ie. that the queen was hanging). The reality was much worse. Vasik just thought that black is winning after 22. Qxd7 Qa3. He calculated this way back when playing 18. .. Rxc3 and didn't stop to review it once he got here. He's insanely optimistic sometimes. I was too but this is crazy even for me. It's a tough loss which cost the club quite a few "sporting points" which determine club rankings and funding, but it's also a good lesson.
Anyway, here is the full last-round game:
Agree about the Urych game (round 8), I would also keep playing with black, as would Vasik. The chance of a win is low, and the chance of a loss isn't zero, but still, black is surely a bit better. Of course the engine gives +0.0.
(By the way, the official scores are "Last-Name, First-Name", so the kid's first name is Krzysztof, ie. Christopher, and his last name is Urych.)
Not much consolation, but I had the same hallucination as Vasik about ...Qa3 when I was playing over the game.
I wonder about positions like the end of the Urych game. It seems like in my blitz games people at my rating will lose even simpler positions. Of course, time is a factor, and so is the better side's risk tolerance.
People lose drawn rook endings all the time. Since I've started playing again, I've played 20 games and lost 4. Two of the losses were drawn rook endings. I posted both of them here, the first was the 1. b3 + 3. g3 game from the second league, the other was the game against Lewandowski with .. Qxa2 from last year's third league. Vasik has also lost several, here is another from last year's junior first league:
The Urych one was a little bit easier than these but it also seems losable.
Good to hear your results are ok. Losing only 4 is pretty good even if they are lower rated players.
These positions from rook endgames are probably only the most obvious of the situation where the theoretical result is clearly drawn, but one side is much easier than the other side. In other positions we have to trust the computer's evaluation, e.g. when it seems white is strategically lost in a KID or dragon with a +0.6 score. It's definitely murkier.
As we saw in our last vote chess game, computers often give scores near zero when there is no risk for an engine to lose it, even if in a human game one side has all the chances. You'll often see this with simplified (but not tablebase) pawn-down rook endgames.
The chess level has gone way up relative to FIDE rating since we played. If you teleported a 2100 FIDE from today into the mid-1990s, he'd be something like 2300 FIDE or 2400 USCF. The players from our generation who kept playing have dropped 200-300 Elo. In the match against Piescirogi, their 3rd board was a guy named Artur Grzelak who I've known for 25 years. He used to be 2350 FIDE and played in the first league. Now he's 2050 FIDE and plays third board in the third league. He's even been the weak link on their team this year, scoring 2.5/5 on 3rd board while his teammates have been cleaning up. I've dropped 40 Elo in my 20 games and have been a little bit lucky, as you've seen. This effect might vary from country to country though, not sure. Also, the improvement in chess level isn't even across the board, for example endgame play hasn't risen to the same extent and may even have fallen.
Maybe that makes you the wily veteran.
About quayling, I sometimes wonder if it is general aging, or some specific medical thing like ministrokes, or the change in chess culture. I think a lot of my moves are based on intuition and automatic reflexes, and it used to be more concrete calculations.
That's interesting about 200-300 Elo. So a 2400 today could go back in time and be competitive with K+K. Not sure how I'd fare OTB today. It seems from the Lichess where the titles are sometimes shown, that I could be in a similar place, just below the FM level.
About the rook endgames, the material is often not a big indicator. Although it is hard to predict quayling, that is how it happens. Your position, and Vasik's position, I think were a little rare where the defense required knowing the right position to play precise, as well as advanced knowledge for what to do. It can also be tricky in the tablebase positions.
It is also common for people to mess up K+P positions. I do too, but less often than my opponents. This of course is in blitz, but it is evidence for your last statement about the possibility that endgame play has fallen. It might also go with the increasing disregard for conventional wisdom (which might have begun with QSPM), in particular, cautions about K+P endgames.
I guess to be consistent it would need to be a 2500 FIDE to be competitive with K+K. It's hard for me to judge that. Getting FM or higher nowadays seems pretty hard but those titles don't go away so you might be playing FMs from our generation or even older. Online ratings are also different of course, Vasik is 2000 FIDE and 2200 on here. It's just a question of how the numbers are calibrated.
It's also possible that I am a bit weaker now than before due to brain decline. I don't feel that way but it's hard to know for sure. The only way I've experienced this kind of aging is that I have to be more careful about sleeping correctly, eating correctly, taking breaks, etc. I've been taking naps between rounds in the third league this year, after a poor performance last year in the afternoon games. 30 years ago I would have laughed at something like that.
You mentioned that you might play again at some point, so you can try it for yourself. 
Re. those endgames, normally we'd say that the stronger side is "better". The problem with engines is that they will give +0.0 if the advantage has no chance to translate into a win in engine-vs-engine play. At least, this applies to engines which are based on minimax. Eventually, such engines will give +0.0 for every white move in the starting position, then +0.0 for every black reply other than the really big blunders, etc.
It's also true that people seem to rely less on heuristics nowadays. One potential explanation is that heuristics are the result of puzzling over positions you don't quite understand. Probably engine scores have displaced heuristics to some degree. You still need heuristics to play as a human but the being-puzzled stage is shorter. Not sure about this one.
I did a lot of thinking about this yesterday, and talking to Vasik.
We had kind of put the King's Indian and Accelerated Dragon into the same category, as old school openings which computers dislike but where arguably black had good practical chances. Actually, they are very different from each other. The Accelerated Dragon is perfectly playable even today, while the King's Indian really isn't, at least not in the old-school way we've been approaching it.
First, engine scores. Just running the engine on the basic King's Indian position is already +0.6. The Accelerated Dragon is +0.4. To keep the scores that low black has to play all kinds of weird stuff in the King's Indian, while in the Accelerated Dragon those scores are reached through natural moves.
Second, my results have been quite different. Since I started playing again in 2023, I've had 10 blacks, which break down as follows:
1x 1. g3 & 3. b3: equalized early
1x London: equalized early
4x Accelerated Dragon: 3x equalized by move 10, 1x equalized on move 17 and won on move 23
4x King's Indian: was never equal in any of them until move 27; 2x scores near +2, 2x scores near +1
Third, the number of white options. In the Accelerated Dragon, white's only option to get any kind of advantage is the Maroczy, and then black is the one with the next few options. Around 50% of players play 5. Nc3 (without c2-c4) when the score is already below +0.2. The computer would rather have white against the Najdorf (the "best" Sicilian) than 5. Nc3 against the Accelerated Dragon. Other deviations such as 1. e4 c5 2. Nf3 g6 3. c3 and 3. d4 cxd4 4. Qxd4 also score below +0.2. In the King's Indian, everything is +0.7 or so, including all kinds of systems I've never seen before. In the third league I twice faced pet KID variations I'd never seen before, both were +0.8 and quickly got to +1.5 after my "mistakes".
Fourth, we have some new ideas in the Accelerated Dragon. In studying for Vasik's fifth round opponent, we realized that his Accelerated Dragon move order is really cool and forces white into an early hard decision. (I'll skip the details for now, feel free to ask.)
Conclusions:
1. I'm keeping the Accelerated Dragon. No Berlin.
2. King's Indian, I am not sure yet, but something needs to change. There are three options:
a. Nimzo/Queens-Indian.
b. Gruenfeld (playing 2. .. g6 allows a bit more reuse of knowledge against side variations).
c. "Funky" KID variations, preferred by engines to keep the scores low (of course). When top players play the KID these days, usually in blitz and rapid, they also play these.
For example, against the system Vasik faced in round 6, the engine wants to play 1. d4 Nf6 2. c4 g6 3. Nc3 Bg7 4. e4 d6 5. h3 Nbd7 6. Be3 e5:
The point of delaying .. O-O is to meet 7. d5 with 7. .. h5. This stops g2-g4 of course, and after an eventual f2-f3 black will play .. h5-h4. The engine scores after 7. .. h5 are only +0.45.
These types of "funky" computer variations exist throughout the KID.