Tuesday, November 27, 2012

Non-Random Chess960 Trial Game 6: SP408

The PGN database for these trials is here.

This time we are going to work through the first ten moves of some of the SP's Mark found in this post to be possibly dangerous for black. First up is SP408 = RBQNBNKR.

This SP is really "diabolical" to quote Mark. White's first move 1.d4 has so many good things about it including:
  1. Develops two pieces at once
  2. Claims territory in the center
  3. Threatens d5 which closes down black development options
  4. Although undefended, black cannot attack it with without a compromised development
  5. The pawn does not hinder any of white's other pieces in the slightest
Black on the other hand has no such luxury. As Mark put it:
"First, the minor pieces don't coordinate well. The Knights are competing for the same square on the e-file and the Bishops have no obvious optimal deployment. Second, the weak e-Pawn, which is not easily defended, is easily attacked by the Knights. Third, the castling option is limited; the King has to stay on the h-side."

The question is what can black do about 1.d4? At first we followed the CCRL database (see "cool links" top right). One possibility is for black to build a dynamic safety net around his king thus:

SP408 "The Diabolical SP" Hypothetical: 1.d4...c5!?

The variation above is now scored at almost 0.00 by Houdini-2c. This is where it get's interesting. Mark has decided to forget about the CCRL database altogether and play by gut feel! So we have the situation 1.d4 d5!? a symmetrical opening and not one engine in the CCRL database even considers this move. Houdini2c scores white's lead at almost half a pawn.

After the almost automatically good move for white 2.c4 Mark finds what I think is the needle in the haystack 2....Ned6!? that saves blacks position at least for longer. Studying the continuations after this move, I found this surprising line that actually does produce the possibility of long side castling for both colours!

SP408 Hypothetical: Long side casting both colours is possible!

Now at move 7 in the real game, clearly white has all the initiative (see below). I played 6.e3 which is strictly not a developing move but it sure is a positional move. This actually has given black a free tempo and here at move 7 Mark actually uses it!

SP408: 7...b6, what does it achieve?
The move is sure got me both interested and a bit stumped for now. 7...b6 seems not to aid development and the c5 pawn doesn't need protecting yet. Also, a tempo is lost if later ...b5 is necessary. Perhaps it's a case of black not having a better move? Here is Mark's thoughts on it:

...A useful move that prepares various a-side deployments, strengthens the c-Pawn, and leaves Black's options open.

After this move I played 8.Qc2 a very simple developing move that also waits to see what black does. White doesn't want to castle yet in case the h-file opens and doesn't want to move the e1 bishop until the c3 square is cleared. Notice the important feature that white has, a free clearance square or intermediate square on e4 to park a knight so that the c3 square is cleared. This is typical Chess960 stuff where we need a free square so that our position can be improved.

So now we have:
7....b6 8.Qc2 Nd6

Again another interesting move by black. I think if black would have played 8....Nxg3 the opening of the h-file would have been a fairly serious problem for black since white is just continuing development and pressuring the king side.

However now after 8...Nd6 I'm studying the crazy possibility black has of playing ...h5/h4/h3!? and potentially developing the rook to h5 as well as disturbing white's kingside! Take note that the edge pawn rush is a very common feature of Chess960 when a knight sits on Nb3/Nb6/Ng3/Ng6 and get biffed by the edge pawn. The pawn then marches on to undermine b2/b7/g2/g7 in concert with a bishop or queen on the long diagonal! But that is not the reality in this SP.

Here is a special milestone reached in this trial game, black has played ...f5!?, a really interesting move to say the least. How can such a move work with black's b8 bishop blocked in while all white's minor pieces are aiming at the king.

SP408 black continues to play actively!

Here is one way that ...f5!? proceeds:

SP408 white insists on attacking g7

After 12...f5 I castled g-side. Mark's immediate comment was "...I was surprised to see White castle. I expected h2-h4 at some time, using the h-pawn to disrupt the h-side". I then tried the idea of a4/b3/Bb2 to keep long term pressure on the long diagonal.

So here is the final position reached in this trial:

SP408 End of trial at move 16

My own thoughts are that black has defended amazingly well. It is remarkable that both black's rooks have had to transpose to the 7th rank so early in the game. I think the difficulty is that although black's piece position is compromised, white has only created one weakness when two are needed. Mark's comment at the end of the trial were:
"...The Bc3 is certainly a better piece, but the difference between the two positions is probably nothing more than the difference between these two bishops"

Lastly, please see the comments to this blog for the details of what we were thinking move by move in this trial run at the very difficult SP408 where white calls all the shots and black is very much in the hot-seat.


  1. I like you will study the CCRL databases for this SP. The move played is also by far the most popular both for computers and I think for humans!


    ...developing two pieces in one move, blocking no other pieces for the forseeable future, and setting up the center beautifully. The ideal first move!

  2. Good choice! From CCRL I'm seeing 1.d4 with a WLD score of +44-12=8. The two most important variations are 1...c5 with +22-9=6 and 1...Ng6 with +20-2=0. Those are terrible stats for Black. Now I have to set the engines to work. - Mark

  3. There is at least one other feature about 1.d4 in that black has no direct way to attack the pawn and develop using minor pieces....

    When you look at the CCRL you will notice something straight away if you study Houdini-3. I'm pretty sure it's got a bug in the program. It will play 1.d4...c5 2.d5 then ...c4?! moving the same pawn twice as black!? It did that with both it's games against the other top engines and lost (games 111931 and game 113233). When it played as white, it could only manage a draw against Critter (game 97607).

    Houdini-2 does not play like that and neither does Critter. Considering that Houdini 2->3 is really only incrementally better, combined with other 960 games where I think Houdini-3 wants to move the same piece twice in the opening and looses as well, there is something wrong with it in 960 mode.

    For me the useful engines that I have seen are few. Basically Houdini-2, Critter 1.2 and perhaps Rybka 4.1. Stockfish 2.3.1 is a wonderful engine but it's weakness is it has trouble developing the queen when she is in the corner.

    It all comes down to how much effort the programmers have put into Chess960.


  4. I'm not ready to move yet, but I have some observations to share. It's truly a diabolical SP.

    First, the minor pieces don't coordinate well. The Knights are competing for the same square on the e-file and the Bishops have no obvious optimal deployment. Second, the weak e-Pawn, which is not easily defended, is easily attacked by the Knights. Third, the castling option is limited; the King has to stay on the h-side.

    After 1.d4 c5 2.d5, the most popular move is 2...c4. It's not a bug. The engine is trying to prevent c2-c4, which creates a strong center for White. Note that White has also moved the same Pawn twice, but the second move is very strong because it limits the movement of the Black Knights. I don't think 1...c5 is playable.

    On top of that, I don't like 1...Ng6. It commits the Knight to a less than optimal square and leaves White a free hand in the center. I appreciate that it prepares ...O-O and guards the weak e-Pawn, but neither of these objectives is a priority.

    I'm looking at other moves instead. I have a preferred move, but White's response is obvious and I'm not sure the tactics work for Black. I'll make another comment when I'm ready to move. - Mark

  5. Hi Mark. Thanks for the update. I'm enjoying not having to do the hard work here! What about the line 1. d4 c5 2. d5 e6 3. c4 f5 4. Bc3 Nf7 5. b3 Ng6 6. h4 h5 *

    It's a truly amazing defense if nothing else!

  6. This SP reminded me why I stopped looking at CCRL or engine suggestions for the openings of my own games...

    1.d4 d5

    ...Better to work out the issues myself rather than rely on engines which have no understanding of positional play. Even if I lose the game, I've learned something. - Mark

  7. Once again you have completely surprised me. It is you who are a creative player! I'll spend a bit of time thinking about this since the CCRL database is now completely useless. Check out the amazing 1.d4 c5 variation I found and have put on the main blog.

    Catch you in a day or two while I sort out what the problem is if any, with the symmetrical start.


  8. Well you knew I was going to play it. Just needed a bit of time to work out why exactly it is so strong.

    1.d4 d5 2. c4

    ...black seems unable to stop white gaining control of the center, combined with development tempo and an attack on black's stranded king.

    The primary moves that I think are important for white down the track can be (e4) which white can afford to play because of the spare tempo. If this move is played, amazingly white's position transposes into an almost classical like position but with a lot of extra tempo and a simple plan to attack black's king.

    I spent a fair bit of time analyzing what happens if black draws white's queen out after:
    2.c4 dxc4 3.Qxc4 b5!? 4.Qc3 a5!? 5. Ng3 b4!? (followed by ...Nb7!?)

    A possibly amazing solution for black because it biffs white's queen and controls c3 as well as exploiting white's exposed queen.

    No hurry Mark. This is an absolute mine field for black.

    In the future, the openings of some SP's for black are going to have to be committed to memory. Ironically I think it is possible that the repertoire top players will need for Chess960 will be no less extensive and no less theoretical than what players have to memorize today.


  9. You wrote, 'In the future, the openings of some SP's for black are going to have to be committed to memory. Ironically I think it is possible that the repertoire top players will need for Chess960 will be no less extensive and no less theoretical than what players have to memorize today.'

    This might be true if you believe that the players of White play like engines and the players of Black don't. In reality, both players are looking at the given start position for the first or second time, without the help of an engine. They are no more likely to find the most critical variations than 17th century players were to discover the Marshall Gambit in the Spanish Game.

    The problem with traditional chess is that *both* players have studied deeply the same variations using engines. Both have memorized the same tactics and both have weighed the same positional tradeoffs. The opening becomes a battle to see who has better prepared the variation that appears on the board.

    It is extremely unlikely that, with 960 different positions to cover, both players will have studied the same SPs to more than the first move or two for each side. Even if they have, the variations branch so rapidly that the players will quickly be in uncharted territory.

    You also appear to believe that there is some inherent flaw in SP408 that gives White an easy advantage. I see another possibility: that the engines have some inherent flaw that guides them all down the same path, blind to alternatives. After one move each, we already have a position that none of the engines has tried. I hope to discover that this variation is no worse, perhaps even better, than the variations that the engines have played many times, with dismal results for Black. - Mark

  10. I wouldn't have played 1...d5 if I were forced to answer 2.c4 with 2...dxc4. There are more interesting possibilities, such as...

    1.d4 d5 2.c4 Nde6

    ...I read with interest your paragraph in the post that started, 'White's first move 1.d4 has so many good things about it'. Everything you list also applies to 1.e4 in the traditional setup, and yet Black survives!

    I also liked the next paragraph, 'Black on the other hand has no such luxury' with my list of features in our SP. Again, these apply to White as much as to Black. It seems that Black, moving second, is confronted with the defensive aspects of the position before White, who has been able to cherry pick from the offensive aspects. Have we stumbled onto a new principle of chess? - Mark

  11. Great thought. That exactly resolves what I feel when I say "I have the initiative!" It is that I think I have more offensive options available than defensive and then there is a positive feeling surrounding having it. A rush.

    However that is not all initiative is. Initiative is also a sense of harmony that the risks of offensive options are manageable. If I have offensive options but there is a great sense of risk, I feel like I do not really have the initiative.

    So initiative could be is defined as offensive options + manageable sense of risk. If this is true, initiative is unique to each player because each player weighs up offense and risk differently.

    I'll go one step further with this idea. I think that humans are resistant to changing things even if they could do with changing, because although they have the offense (knowledge and skills), they still have great doubts, and thus do not have the initiative to make the change. This is one reason why Chess960 is slow on the uptake.

    Note that initiative is not the same as motivation. You must be motivated in all cases before initiative has any meaning.


  12. Hi Mark. Your last move was wonderful; possibly the needle in the haystack. Instead of me spamming you with emails, check out my blog from time to time about what I am thinking in this position. I'll probably make a move today or tomorrow. On my blog I even found a way for both colours to castle to opposite wings.

  13. In our first game in this series, 'Fischer-Bronstein Non-Random Chess960 Trial', I wrote, 'I don't like to comment on games on progress, but I'll make an exception here since this is so new.' I still think it's a bad habit, for several reasons. Let's not overdo it. - Mark

  14. Fair enough. If I'm inspired by particular lines I find, I'll post them in the main blog. Just check it from time to time in case I have gone quiet.

    1.d4 d5 2.c4 Nde6 3.cxd5 if Nxd4 4.Nc3

    ...trying to gain in relative development while black reorganizes the position of the knight. The center pawn appears to have staying power on the 5th rank.

  15. The d5-pawn is not as worrisome as in the 1...c5 variation, because it no longer dominates the Knight on d8...

    2.c4 Nde6 3.cxd5 Nxd4 4.Nc3 c5

    ...The main alternative was 4...c6, but the Pawn will eventually continue to c5. Better to put it there in one move. - Mark

  16. 3.cxd5 Nxd4 4.Nc3 c5 5.Ng3

    ...continuing development and trying to get an edge in better piece placement. What has seriously struck me this trial, is how the most commonsense of moves are also so difficult for the engine to see.

  17. I've said many times on my blog that the engines, when left on their own, don't play the openings particularly well. It has to do with their inability to understand schematic thinking...

    4.Nc3 c5 5.Ng3 Bd7

    ...Develops a piece and protects f5 for the eventual retreat of the Nd4. - Mark

  18. Thanks on the "schematic thinking" concept. It really is a useful technique with 960. The next section of play is actually where the engine really helps.

    5.Ng3 Bd7 6.e3 if Nf5 7.Bd3 if Ng6 8.Qc2

    ...a positional move that is neutral in tempo but will allow the schematic arrangement Bd3(without being threatened by Ng6-e5)/Qc2/Ne4/Bc3/Nf5(an outpost possible because of d5) aiming at the king and heavily restricting black's pieces at the same time....I think. The interesting square for me is e4 because it seems to be an intermediate square for a knight to sit that will clear c3 for Bc3.

    Phew, a big paragraph there! Lots to think about.

  19. I don't understand the reason for the second 'if' move. It's not forced or especially strong. Since Black has a number of attractive moves, I'm going to take more time to look at the position. - Mark

  20. No problems Mark. Since this is non-competitive research I just put in the second "if" just in case it is useful to you.

  21. Non-competitive research? You aren't trying to win?...

    5.Ng3 Bd7 6.e3 Nf5 7.Bd3 b6

    ...A useful move that prepares various a-side deployments, strengthens the c-Pawn, and leaves Black's options open. Re 'if' moves, here's one of your compatriots in action...

    C.J.S. Purdy's Correspondence Skullduggery

    ...Beware an Aussie bearing 'if' moves! - Mark

  22. Thanks for the link. I wish I had your confidence that 960 is balanced where black can put up a pretty good fight across all 960 SP's. That is my hope that one day 960 will be up there with Chess in interest. In this SP you are showing me that black is surviving much longer than I initially thought which is already good news.

    5.Ng3 Bd7 6.e3 Nf5 7.Bd3 b6 8.Qc2

    ...beginning the process of connecting the rooks, strengthening the b1-h7 diagonal, delaying moving the e1 bishop until c3 is clear for it, and delaying castling in case the h-file opens.

  23. Re the fairness of all SP408 positions, I'm not as confident as you might believe. The CCRL results need to be explained...

    7.Bd3 b6 8.Qc2 Nd6

    ...The only real alternative was 8...g6, but that move has positional drawbacks. - Mark

  24. I took my time here Mark because again the computers don't seem to be that useful and there is some really weird stuff going on in my mind if you respond with ...h5

    My thoughts on this SP so far are that black is having a hard time, but if elite 960 players were to memorize the basic ideas, white's win is by no way guaranteed and if it comes, it comes long enough down the track that no-one can say that white's win is trivial! It produces a lot of interesting chess where white needs to think hard. The main issue is that because white has all the easy moves and initiative, black's clock would die a horrible death unless the basic ideas are committed to memory.

    7.Bd3 b6 8.Qc2 Nd6 9.Nce4

    ...routinely proceeding with the schematic plan of clearing c3 for the dark bishop.

  25. Why would Black play 9...h5? It's just an unforced weakening of the h-side with no visible compensation...

    8.Qc2 Nd6 9.Nce4 Ng6

    ...The main alternative was 9...Nxe4; it cedes the Bishop pair, but I see no need for such drastic action. As for 9...Ng6, I can't see the Knight going to any other square in the near future, so might as well bring it out now.

    Re your comment that 'Black is having a hard time', I couldn't disagree more. I don't understand why the engines are consistently overvaluing the position for White. This seems also to have been the basis for the strange CCRL results. There is something about this SP that the engines don't understand, but what exactly? Maybe the next SP will offer a clue. - Mark

  26. ...h5 was pretty much my engines top choice! :-) It wants to play ...h5/h4/Rh5 attacking the d5 pawn. Perhaps the problem with the engine is how it evaluates king safety. That would be my guess.

    8.Qc2 Nd6 9.Nce4 Ng6 10.Rd1

    ...continuing development and delaying non-essential moves.

    That's 10 moves and a successful result from my perspective in terms of at least knowing that black has fighting chances which is what we want to know.

    Chess960 is starting to remind me a bit of the old grass court tennis where the serving side has the advantage and is trying to "hold serve". So in Chess960, white is the server and black is the receiver and whoever is serving is trying to win the game, while black is trying to draw but pounce on any counter attack...just like in tennis.

    Let me know if you want to move onto the next SP or if you feel it is important to continue playing. Initially when we started I felt that 15 moves would be a good goal for each SP but there are a lot of SP's to work through!


  27. Yes, King safety undoubtedly plays a role, as does piece activity. The engines don't like the Rook in the corner and, since the King doesn't move in O-O, probably don't detect a big difference between Rh8 and Rf8. On h6, however, the Rook looks ready to rumble. As for me, I prefer more classical development...

    9.Nce4 Ng6 10.Rd1 O-O

    ...One of the choices for next SP is SP749 RKNBNQBR, the twin of the current SP. Since there are no additional issues about castling on the first move, I suspect that the engines will calculate similar variations to what we saw in the current game. I propose that we continue the current game for a few moves to make sure Black continues to survive, and to give me time to look at the CCRL archive for SP749. Then we can switch to a new SP.

    My 'Yellow Flag' post also identified SP868 QBBRKRNN as a problem SP. What do you think about trying that one? - Mark

  28. I do enjoy playing SP's that are similar to each other. SP749 would be good next. The similarity helps me understand more of what is going on. Some might say that it merely the difference between peanuts and cashews, but I am not one of those!

    9.Nce4 Ng6 10.Rd1 O-O 11.Nxd6

    ...a positional move that allows Bc3 and Nf5 and restores white's initiative that seemed ominous from the first move. I sincerely hope that black can find some resource here to quell white's energy.

  29. Sometimes the most difficult choices are where you have exactly two moves branching in entirely different directions...

    10.Rd1 O-O 11.Nxd6 exd6

    ...Black's Bishop is blocked by its own Pawns, but I'm confident this is only temporary. - Mark

  30. What happens to black's dark bishop is going to be fascinating seriously.

    10.Rd1 O-O 11.Nxd6 exd6 12.Bc3

    ...a painfully simple developing move for white that could be played by a 5 year old, in contrast to black's very complex world!

    PS) I haven't looked at SP868 yet but will try to remind myself.

  31. Painfully simple developing move? Black's very complex world? What on earth are you talking about?...

    11.Nxd6 exd6 12.Bc3 f5

    ...The alternative was 12...b5, but I prefer the direct approach here.

    I looked at CCRL's SP749 and discovered that the themes are similar to the current game. The mirror of the line you thought was a blunder is the most popular variation for the engines. I wasn't too keen about playing a similar SP, but now I agree to play it. I also request to play Black again. This will give both of us the opportunity to build on previous experience. - Mark

  32. Probably a good idea for you to play black again. I doubt I have the positional understanding to play this SP as black.

    11.Nxd6 exd6 12.Bc3 f5 13.O-O

    ...white completes development and connects the rooks. I thought about Nh5/b3/Qb2 which is immediately eye catching and was one of the big plans I had, but your knight is an incredible defender on e5 and if I prise it off with f4, you have an incredibly active defense attacking my e3 weakness which is good enough for a draw *I think*

  33. You think I'm playing for a draw? All three results are still possible here....

    12.Bc3 f5 13.O-O Rf7

    ...I was surprised to see White castle. I expected h2-h4 at some time, using the h-pawn to disrupt the h-side.

    Black's last move guarantees sufficient protection of the g-Pawn for the next few moves. How about we play through move 15, then tackle the next SP? - Mark

  34. Yes h4 would have been logical.

    12.Bc3 f5 13.O-O Rf7 14.a4

    I guess we could keep playing until black finishes their development. When is that?

  35. Development? All in good time...

    13.O-O Rf7 14.a4 a6

    ...Prepares ...b5, opens a second option for the Bb8, and clears Black's 2nd rank for Rook maneuvers. - Mark

  36. The Christmas crazy's. Have a peaceful and enjoyable Christmas Mark and sincerely thanks for sharing your knowledge with us again this year for Chess960!

    Our main Facebook and Youtube Chess960 enthusiast Piewalkermatt is still keen and hopefully will bring more Chess960 videos for us to enjoy in 2013 as well.

    13.O-O Rf7 14.a4 a6 15.Ra1

    ...tries to stop black's c-side campaign using the undefended a8 square, by keeping black's position cramped. If the position opens, white should get some nice lines of attack in compensation for the isolated edge pawn.

    If you move the b8 bishop, as far as I'm concerned that is technically the end of the development phase?

  37. Merry Christmas, Harry! This is my last move before the festivities start...

    14.a4 a6 15.Ra1 Ra7

    ...The toughest move so far. Black has lots of alternatives, but which is best? - Mark

  38. Honestly, I am having trouble with this position optically. Your two rooks will connect on the 7th rank and what is your dark bishop doing!? The bizarre world of Chess960! If black can develop the dark bishop to f6 via c7/d8 that would be pretty special.

    14.a4 a6 15.Ra1 Ra7 16.b3

    ...the fianchetto bishop has no future on c3 and so it will need a new home on b2. Whites plan to keep attacking g7 while weakening black in other places is starting to fall apart.

  39. The Bb8 is doing a lot. On top of watching d6/e5, it protects the Ra7, making my next move possible...

    15.Ra1 Ra7 16.b3 b5

    ...The Bc3 is certainly a better piece, but the difference between the two positions is probably nothing more than the difference between these two Bishops. - Mark (suffering from a flu, so a little slow to move)

  40. I think you have made a great case to start the next trial! Thanks for your comments and inputs. We can continue this game, but how about we tackle the inverse of this tricky SP perhaps first thing next year?

    Until then, all the best for 2013 and hope that the flu will be fixed with a good dash of Honey Whiskey for New Years Eve as an alternative to champagne :-)

  41. You've been calling the shots in this game until now, so if you're willing to concede that Black is still standing, we can move on to the next adventure. When you set up a new thread, give me a shout via this one and we'll have another bash at each other. In the meantime, have a Happy New Year! - Mark

  42. Hi Mark hope the flu is getting better.

    White was calling the shots but only managed to create one weakness and couldn't find a second serious enough one.

    I've started the next trial.