Wednesday, August 29, 2012

Non-Random Chess960 Trial Game 4: SP959

The PGN database for these trials is here.

Mark and I have finished up our non-random trial of the 'barbecue' BBQ SP's with SP959. Depending on how you number the first position in Chess960, the very first position in all of the 960 starts is a BBQ or 'barbecue SP', but is also the very last! Now in retrospect having studied them in detail, I think they have their wonderful place in the Chess960 repertoire for sure! I first assumed that there would be no way that black would have time to castle at all, that his king would be stuck in the corner, but I was wrong. I also thought that the game would have too few options, but essentially I was wrong about that as well, if you take out the first few moves as almost "inevitable". Note that I have played the 'barbecue' gambit 2.b4 with a reasonable game in our previous trials.

In the sample opening we played through below, the feature is that not only do we have a broadside bishop attack on the king, but the king is on the c-side and thus castling long could leave the kings safety even more exposed than in SP000 (SP960). That said, Mark made the classic mistake that all Chess960 players of this age including myself make, and this is to confuse in one's mind, the c-side from the g-side. Mark based all his early thinking on the plan to castle short, but that just so happens to be on the extreme opposite side of the board! Amazingly, it turned out fine in any case.

SP959 Nf3 variation: What are white's castling intentions?

Here are my thoughts on SP959:
These 'barbecue' SP's are actually surprisingly stealthy start positions because it turns out that all three castling options are possible, castling c-side, g-side and not castling at all, stealthy because neither side want's to reveal what their castling intentions will be until the last minute, for obvious strategic reasons. We also noticed that with some solid principled play, black maintains strong fighting chances. That said, the pressure is definitely on black to play solidly! This confirms an educated "guess" about Chess960, that there are roughly 50 start positions that have to be studied in detail and memorized so that black specifically knows how to play them.

Here are Mark's comments playing as black:
"..The pieces from the d- to h-files develop naturally. The big question is what to do with the RKR.....Good observation that many chess960 SPs offer three castling options for both sides. SP518 (the traditional start), with the King starting in the center, offers only two, with a few exceptional variations where a King doesn't castle. If I wanted to play it safe, I would move 10.Bf4, but I'm ready for a fight...Move 10.Qf2 prepares O-O and defends against 10...Re8 with a tactical trick. That's ten moves for me, but the game is just getting started. White can't claim any advantage and it will be a tough battle ahead."

I will leave it to someone else to work out what Mark meant by "10.Qf2 defends against 10...Re8 with a tactical trick". In our trial, Mark and I accidentally had been playing by book up to move 6 from the CCRL 40/4 FRC database where Mark deviates from Baron170 by exchanging knights. The game that we were accidentally following includes a stunning but curious migration of white's king and a double rook lift to the second rank! What was the computer thinking?

SP959: Baron170 migrates the king from g1 to b1 starting move 22

Finally, here is a further sample of a hypothetical line in this opening. In some Chess960 start positions, the game proceeds almost as if it is playing itself (where the moves are so natural as to be inevitable). Normally, once this inevitable position is reached, then things get interesting. To complement the above opening system where in the resulting game the king manually migrates laterally across the board, here is a hypothetical variation where white castles kingside?!

SP959: White castles kingside?!


  1. First, to make sure we are on the same page, the position that I am playing is SP959 RKRNNQBB, the twin (or mirror) of the previous game SP000 BBQNNRKR. That game demonstrated that there is no forced path to an overwhelming advantage, although Black faces multiple problems and has to play carefully...


    ... I'll tray a different approach. The move 1.Nd3 would have repeated the opening of the previous game. My move grabs a corner of the center, opens a diagonal for the Bishop, makes some space for the Queen, and offers the Knight a good square on f3. - Mark

  2. Thanks, I've added the comment above to the write up on the last SP.

    1. f4 Nd6

    ...if white's doesn't play the knight then black will! *I think* it is possible for three reasons:

    1. Because the knight stops 2.e4
    2. If 2.b4 black has a direct way to extinguish the rapid gain of space on the queenside (which could be pointing to another universal principal of Chess?)
    3. The threat on d2 appears to be serious enough to prevent white from being able to make rapid positional gains.

  3. By coincidence, I ran into this today...

    Fischer Random Chess position of a game I recently won.

    ...As for our game, I'll continue classically...

    1.f4 Nd6 2.g3

    ...Opens the second diagonal for the adjacent Bishops while waiting to see Black's continuation. - Mark

  4. It feels like we are getting the hang of the BBQ SP's? They aren't that difficult if you know what you are doing and have a bit of experience with them.

    I'm trying to think of ways that we can continue to cut through this massive 960 cake with our tiny little knife in ways that quickly build more knowledge! It's a massive undertaking but there would be some shortcuts surely.

    1.f4 Nd6 2.g3 f5

    ...develops a bishop and prevents e4

  5. I was half expecting 2...g5 with another look at the BBQ gambit...

    2.g3 f5 3.Nf3

    ...The pieces from the d- to h-files develop naturally. The big question is what to do with the RKR. - Mark

  6. Good point on the RKR problem. That really is a big theme. g5 was my computer's favourite move for quite some time?

    2.g3 f5 3.Nf3 g6

    ...put's the future direction of the game into white's hands.

  7. My engine wants to play a4, but there are other ideas to try first...

    3.Nf3 g6 4.d3

    ...Symmetry is not possible. How will the two armies challenge each other? - Mark

  8. I honestly liked your previous move 3.Nf3 because it appears to free up the pawns in the center and kingside and it's not an impediment at this stage that the corner bishop bishop is blocked. I also like your idea here to break symmetry a bit later than when I broke it with the barbecue gambit at move 2. The big question is whether it has to be broken at all?

    d3 feels intimidating especially because it feels like you are priming to make pawn break and also that it restricts my d6 knight, but black is still able to find some strong moves...

    3.Nf3 g6 4.d3 Nc6

    ...Black takes the lead in development and actually has more control of the center than white. After all black can play ...e5 but white cannot play e4. If you are planning to develop via a4, if you can make it work, I would love to see how.

  9. Symmetry will probably turn out to be one of the most important challenges in chess960. The SP518 doesn't allow much symmetry, but some chess960 SPs can go for many moves with Black copying White...

    4.d3 Nc6 5.c4

    ...There are a number of ways White can untangle the Kingside pieces. The last move preserves the options. - Mark

  10. Mark I was wondering if you would like to stick a coin into the Chess960 jukebox of your favourite "Top 960 Chess Starts Of All Time", and take SP468 for a spin next trial?

    SP468 = RBBNNKRQ

    I was thinking that we might be best sticking to the bishop pairs united, considering that almost half of all Chess960 starts feature them.

    In SP468 you get a beautifully tricky positional start which the computers are terrible at (they cannot handle queens in the corner very well).

    The patriarch bishops bear down their combined authority on not just one major piece, but all of the major pieces KRQ!

    The tricky thing with SP468 is that the queen has no easy escape because she cannot easily exit via the h-file and if she travels laterally, the tension is to castle queenside, but by doing that the king's protection is probably going to be compromised. At the same time, the military knights dead in the middle have a potential to block up the whole works but at the same time they have to be developed....

    Note that the KRQ combination is pretty special, because you cannot get any other combination of major pieces with the queen in the corner.

    Back to our game!

    4.d3 Nc6 5.c4 e5

    ...I don't consider the c4 pawn break a serious enough threat because my knight can actually retreat to f7! and any further advance of the c-pawn block's your own bishop. So I take a page out of your book and simply say that black must play actively.

  11. SP468 RBBNNKRQ is fine with me. You'll be up for White. As for the current game...

    5.c4 e5 6.fxe5; if 6...Nxe5 7.Nxe5 Bxe5 8.Nc3

    ...White continues to develop the Kingside. - Mark

  12. Make that 'White continues to develop the Queenside'. I made the same mistake on my previous move. Argh! - Mark

  13. Really nice opening design Mark. I'm now really nervous about the a-pawn break. We've been playing by book up to move 6! Check my main page where I've published the other game from this opening.

    7.Nxe5 Bxe5 8.Nc3 c6 has an amazing queenside exit plan, but first c6 is critical tactically. Superficially it restricts white's bishop and knight. One of the variations looks like white has castled kingside but is still on the queenside! It get's very confusing :)

    SP959 is a beautiful SP, both simple and really difficult at the same time :)

  14. My confusion of Kingside and Queenside on the previous moves was more than a semantic problem. Since the first few moves of this game I've been laboring under the mistaken impression that castling a-side meant leaving the King on b1 and moving the Rook to c1. Likewise, I thought castling h-side meant moving the King to c1 and the Rook to d1. My plan until now has been to play 8.Rc2 followed by 9.O-O-O, leaving Kb1/Rc1. I think I made this elementary mistake because we just played the twin start position, SP000, and I simply flipped all my ideas in my head without readjusting for the new castling situation.

    Now I need to come up with a new plan. I could still continue 8.Rc2 followed by 9.O-O-O, leaving Kc1 & Rd1, but that position is not as compact as I would like. Furthermore, 8...Qh6 prevents 9.O-O-O, when I need to find yet another plan.

    Since I find myself in troubled waters, I'm going to take more time for the next move. There are many plans to consider and I'm not sure which is best.

    This little episode has taught me a few more things about chess960! - Mark

  15. I did it again! Where I wrote 'Likewise, I thought castling h-side meant moving the King to c1 and the Rook to d1', I should have said 'moving the King to f1 and the Rook to e1'. It's not the first time that my confusion of Kingside and Queenside extends to notation. - Mark

  16. That would not have been a pleasant realisation! If it's any consolation, it's a very easy mistake to make and I've made it probably dozens of times!

    Someone on your blog a couple of years ago suggested instead of a-side we say "c-side". Instead of h-side, we say "g-side".

    This would have the dual benefit of distinguishing the left and right halves of the board, as well as clarifying exactly where the king lands.

    Unfortunately, I think "a-side" and "h-side" are becoming the norm thanks to a few players that are doing Chess960 video's on Youtube, but are not thinking through the implications fully.


  17. Even simpler would be just to put a spot on the chessboard to represent where the king lands. The "spot" would be the same as the spot that is used to mark the 7th and 12th frets of a guitar which is just as confusing as Chess960!

    The thing is that the difficulty of castling in Chess960 is simply that we are the first generation that are playing Chess960. We have the traditional game hardwired into our memory and it interferes with our thinking.

    Future generations will not have that difficulty. Until then, a couple of spots on the board. One on the g-square and on the c-square.


  18. Castling, like the other moves of the pieces, is handled by my subconscious mind. Calculation of variations is done by my conscious mind. In this game the subconscious mind gave the conscious mind incorrect information. It was some kind of a disconnect.

    The designations Kingside / Queenside, a-side / h-side, etc. are notational conveniences for communication and I don't believe they are the cause of the breakdown. The cause was flipping the board in my mind from SP000 to SP959 while subconsciously maintaining certain assumptions that no longer applied. At least, that's how I'm understanding it. - Mark

  19. Back to the game: I looked at a lot of different plans and finally narrowed it down to 9.a4 and the move played...

    7.Nxe5 Bxe5 8.Nc3 c6 9.Be3

    ...It keeps the Queen off h6, stops ...f4, prepares Bf4, and comes one step closer to castling both O-O & O-O-O. The tension on the e-file is manageable. - Mark

  20. 2nd game in the barbecue where the bishop sits in front of the pawn! It's highly logical and flexible as well.

    The castling situation is fascinating. Three viable options including moving the king to the 2nd rank and connecting the rooks. Your camouflage on what your castling intentions are is very cool. After all there is a possibility that you could castle g-side with an attack on the semi open f-file?

    Knowing what each other's castling intentions are is pretty important to framing the strategy. I was not expecting that this SP would have that many castling options. Most pleasantly surprised!

    7.Nxe5 Bxe5 8.Nc3 c6 9.Be3 Qe7 {mirroring a similar move I played in both our other barbecues as white} plan is to occupy the e-file with a rook, but I must not reveal my castling intentions yet with ...Re8, and so change my move order to hide it for a bit longer :) I was going to castle O-O, but your Be3 move has sent me into disarray because I again was not expecting it.

  21. Good observation that many chess960 SPs offer three castling options for both sides. SP518, with the King starting in the center, offers only two, with a few exceptional variations where a King doesn't castle

    If I wanted to play it safe, I would move 10.Bf4, but I'm ready for a fight...

    9.Be3 Qe7 10.Qf2

    ...Prepares O-O and defends against 10...Re8 with a tactical trick. That's ten moves for me, but the game is just getting started. White can't claim any advantage and it will be a tough battle ahead. - Mark

  22. Thanks Mark, I've played my move and done a write up. I've also left a placeholder for our next trial. Cheers