Monday, November 5, 2012

A Bust to the King's Bishop's Gambit, Part 2

So here is my proposed "bust" to the King's Bishop's Gambit, with best play by White (main line).

1.e4 e5 2.f4 exf4 3.Bc4 Qh4+ 4.Kf1 Nf6 5.Nf3 Qh6

The "theory" move here is 5...Qh5 instead, presumably to prevent the perceived threat of 6.Ne5.  But it turns out to be no threat: 5...Qh6 6.Ne5 d5 7.exd5 Bd6, or 7.Bxd5 Nxd5 8.exd5 Be7, both leave Black a solid pawn up.

The key features of 5...Qh6 are that it 1) protects f4 immediately, leaving Black other, better, options than the weakening g5 in response to a White d3 or d4, and 2) protects e3, at least temporarily, after the Black maneuver N-g4-e3+ Bxe3 fxe3, again giving Black time to gain compensation if White goes after e3.

White's best try here is to accept a slight positional advantage for the gambit pawn:


Other tries all have good answers by Black:  6.d4 Bb4 (6...Nxd4? 7.Qe2=) 7.Ne5 O-O 8.Nd3 (8.Qf3 Qh4) Nc6.  Or 6.Qe2 Be7 7.e5 Ng4.  Or 6.Kg1 d6.  Or 6.Ne5 covered above.  The Ng4 idea comes into play if White plays the weak d3, e.g.:  6.d3 Ng4 7.Nc3 c6 8.Qe1 Bb4 9.Qh4 Qxh4 10.Nxh4 Ne3+ 11.Bxe3 fxe3 12.Re1

6...d6 7.d4 c6 8.Kg1 Nh5 9.Ne1 Nd7 10.Be2 Ndf6

Or 8.Qd2 Nh5 9.Kf2 g5 10.h4 (10.Re1 Bg7;10.Rd1 g4) gxh4 11.Rxh4 Rg8 12.e5 d5 13.Bd3 Be7 14.Rh2 Bg4 15.Ne2 Qg7

Sunday, September 23, 2012

Fried Liver, Fegatello, Chess a theoretical draw

I know I still have a loose end in the blog to close -- the King's Bishop Gambit main line, 5.Nf3 with my preferred reply of 5...Qh6 (instead of book 5...Qh5).  As per my Chess Theory post, it too leads to a virtually equal position with best play.

Which leads me to the topic of this post, another draw-by-repetition end to a theoretical opening line.  I was interested to see GM Andy Soltis's article in the August 2012 Chess Life ( on that very topic (pp. 12-13).  (Was he one of the 2 pageviewers of my Chess Theory post? :-) )  Andy points out main-line draws-by-repetition for Najdorf Poisoned Pawn Accepted, Muzio Gambit Accepted, Rice Gambit, Belgrade Gambit, Richter-Rauzer 7.Be3, a Queen's Gambit declined line, and a couple of Modern Defense lines.

And I believe I have found another draw-by-repetition with best play, in the Fried Liver:

1.e4 e5 2.Nf3 Nc6 3.Bc4 Nf6 4.Ng5 d5 5.exd5 Nxd5 6.Nxf7 Kxf7 7.Qf3+ Ke6 8.Nc3 Ncb4 9.a3 Nxc2+ 10.Kd1 Nxa1

I call this the Full Fried Liver, as opposed to 10...Nd4 (the Weak Fried Liver?) which also seems to be a dead-even game in the proposed main line 11.Bxd5+ Kd6 12.Qf7 Qe7 13.Ne4+ Kd7 14.Nc5+ Kd6 15.Nxb7+ Bxb7 16.Qxe7+ Bxe7 17.Bxb7 Raf8 18.Rf1.

11.Nxd5 c6

Perhaps surprisingly, the popular line 11...Kd6 loses here to 12.d4:  11...Kd6 12.d4 c6 13.Bf4! exf4 14.Qxf4+ Kd7 15.Re1 and it's mate in 7 due to the Qf5+ threat (15...cxd5?? 16.Bb5#).  I recently won an over-the-board game in this line after 12...Kd7? 13.Re1 Qh4? 14.Qf5+ Kc6 15.Qxe5 Bd6 (protecting against 16.Qxc7#, but:) 16.Bb5+ Kxb5 17.Nc3+ Kc4 18.Qd5+ Kd3 19.Re3#.

Black's only reasonable try after 11...Kd6 12.d4 seems to be 12...Be6 but it also seems to lose, to 13.Re1 b5 14.Nb4 bxc4 15.Qc6+ Ke7 16.Bg5+ Kf7 17.Bxd8 Rxd8 18.Qxc7 Rd7 19.Qxe5.  (Or 13...c6 is even worse: 14.Rxe5 Bxd5 15.Rxd5+ cxd5 16.Qxd5+ Kc7 17.Bf4+.)

Update.  There are improvements on both sides leading to a different main line after 11...Kd6 12.d4 Be6 13.Re1.  After 13...b5 14.Nb4, 14...Bxc4! is a big improvement for Black because it allows 15.Qc6+ Ke7 16.Bg5+ Kf7 17.Bxd8 Rxd8 18.Qxc7 Be7!, and then also 19.Nc6 Rxd4+ 20.Nxd4 Rd8 21.Qxe5 Bf6! with minimal White advantage.  But White has an even bigger improvement in 11...Kd6 12.d4 Be6 13.Re1 b5 14.Rxe5!, when 14...bxc4 15.Nf4 is decisive for White due to the control of e6.

So 13...b5 is busted for Black.  The main line seems to be instead 13...h5 (threatening a royal skewer (ok actually a royal pin, but skewer sounds better)) 14.Qe4 and White cleans up on the e file: 14...Kd7 Qxe5 15...Bg4+ 16.f3 Bd6 17.Qxg7+ Kc6 18.fxg4 b5 19.Bd3 with threat Be4.

12.Nc7+ Kd6 13.Nxa8 Qh4 14.Qd3+ Qd4 15.Qxd4+ exd4 16.d3 Bf5 17.Re1 Kd7 18.Re5 g6 19.Ra5 a6 20.Nb6+ Kc7 21.Na8+ Kd7 draw.

If anyone thinks they have an improvement for either side, please post a comment.  Thanks.

Sunday, July 22, 2012

A Bust to the King's Bishop's Gambit

Fischer's "A Bust to the King's Gambit" was really only a "bust" of the King's Knight's Gambit.  As I mentioned in the Bishop's Gambit game post, I have developed a response, or "bust", to the Bishop's Gambit.  So I should flesh that out for any readers.  If I am successful, you can file this under Fischer's concluding statement to the effect that White can play different moves (in this case 3.Bc4) but will just lose in a different way.

This post is dedicated to "zavenc" on ICC, a Turkish player who likes marathons, and who likes to accommodate me in ICC marathons by often playing the King's Gambit as White, both Knight's and Bishop's!  :-)

1.e4 e5 2.f4  King's Gambit.

2...exf4  Accepted.

3.Bc4  King's Bishop's Gambit

3...Qh4+ 4.Kf1 Nf6  This is my proposed "bust".
As elaborated in previous KG posts, the knight has a free path to g4, and then to e3 after White plays d4.  In addition to immediately attacking e4 of course.

Theory only considers the obvious 5.Nf3 at this point, which I'll cover in a later post.  But in honor of zavenc's preferred line, I'll cover it first:

5.Nc3  which protects e4.  But because the queen has not been ejected, it allows Black's plan with a mate threat which must be answered:

5...Ng4 6.Nh3  This is zavenc's preferred answer, guarding both the mate threat at f2 and the capture threat on h2.  6.Qf3 is also possible and may be better:
Here Black's most direct response is 6...Qxh2 7.Qh3 Qg3 8.Qxg3 fxg3 9.Nge2 c6 10.Nxg3 d6; Black has exchanged queens and exchanged her f pawn for White's h pawn.

The alternative capture 6...Nxh2+ is also possible.  It gets Black the exchange in addition, but gives White a massive development advantage as compensation: 7.Rxh2 Qxh2 8.d4 Qg3 9.Bxf4 Qxf3 10.Nxf3 d6.

Black's best response to the 6.Qf3 alternative however, is probably an "attacking retreat" making use of the available queen-bishop fork:  6...Ne5 7.Qe2 d6 8.Nd5 (or 8.Nf3 Nxf3 9.Qxf3 c6) 8...Qd8 9.Nf3 Nxc4 10.Qxc4 c6 11.Nxf4 Nd7.

6...c6  This is needed to neutralize the threat of the Nc3 hitting c7.
From this point on, Black holds the threat of d6 followed by Ne3 or Ne5 hitting the knight on h3.

7.d4 g5  Hitting f4, protecting f4.

8.e5  Ready to exchange on d6 or f6, exposing Black's king.

In our last game, zavenc played 8.Qe1 to eject the black queen.  I don't know if this is in his repertoire or just an over-the-board choice, but it doesn't seem best.  In fairness, in the game I responded badly also, pulling the trigger with Ne3+ and thus throwing the f pawn; and zavenc did win the game in the end.  But the correct response to 8.Qe1 is 8...Qxe1+ 9.Kxe1 h6, when White has nothing better than 10.Nf2 d6 11.Nxg4 Bxg4 12.Be2 f5 13.Bxg4 fxg4, Black simply holding the solid extra pawn.

8...Ne3+ 9.Bxe3 fxe3  Black carries through on her plan.
10.Qe2  White has to find this move; nothing else works.  It hits e3 and continues the lineup against the king.  But White also has to protect against developing threats at f2 and c4 simultaneously.

10...b5!  An important in-between move for Black.  Now White has to choose between her bishop and her pawn center.

11.Bd3  The more likely move psychologically, saving the bishop.  11.Bb3 fails to the combo threat b4 and Ba6.  But White can also sac the bishop for e3 plus counterplay against the king:  11.Qxe3 bxc4 12.Ne4 h6 13.Re1 Kd8 14.Nhf2 Qf4 15.Qc3 Na6 16.a3 Bb7; Black will lose c4 but otherwise is secure.

11...Qxd4 12.Nxg5 Qxe5 13.Nf3 Qf4  and Black will lose e3 but still holds a solid extra pawn.
For example:

14.Nd1 Bc5 15.Nxe3 Kd8 16.Nf5 Re8 17.Qd2 Qxd2 18.Nxd2 d5
Or 14.g3 Qh6 15.Nd1 d5 16.Qxe3+ Qxe3 17.Nxe3 Bg7 18.c3 O-O

So that's my first "bust" of the Bishop's Gambit.  Comments, criticisms, improvements, corrections, suggestions all welcome.  Thanks for reading.

Sunday, June 10, 2012

Chess Study and Training

One purpose of making the 3 previous posts was to improve my chess game.  In a roundabout way it has, I think.  While any type of chess study is said to improve one's game, at my age I'm looking for the most time-efficient method.  Those posts took a long time to put together, but somehow their lessons on my (poor) time usage during games led me to a resource that had almost certainly helped me in the past, but which I had since neglected.  In one over-the-board tournament a couple years back, in the first round as I recall, I had managed to reach a king+rook vs. king endgame but with very little clock time left.  I am convinced that it was the KRk bot training on ICC that I had included as part of my tourney prep that had converted a loss (on time) into a win in that case.

So the first step in my thinking was:  I need to find some chess training, as opposed to just study.  And the next step was going back to ICC's KRk and developing a ladder goal of completing three sequential all-best-move mates with at least 3:00 remaining on my clock (and no takebacks), with the next step in the ladder to be KQk.  I'm still working on the KRk.  Some of those first moves are really hard to know/guess.

I've recently gone to the nice online tablebase site to get a better overview some of the trickier first moves, e.g. when the black king starts in the center.

What I've found from just doing this training, is that it seems to have expanded way beyond just one set of mate patterns, into a better/quicker understanding in my mind of how rooks "work" in general.  I recently dispatched an opponent fairly handily with low time on my clock and only up the exchange or so in a busy position, with two rooks working together.

After KQk, I'm not sure how my ladder will go; the sequence in ICC seems to be KBBk, KBNk, KQkn, KRkn, KQkr.

Comments/suggestions welcome.

Friday, May 18, 2012

King's Gambit (King's Bishop's Gambit), Internet Chess Club

OK, on to the other King's Gambit game I got in last weekend's ICC marathon.  This one was a King's Bishop's Gambit.  (See previous posts for the King's Knight's Gambit game.)

This is zavenc (1725) - jientho (1292).

1.e4 e5 2.f4 exf4 3.Bc4  This is the King's Bishop's Gambit.  After Fischer "busted" the King's Knight's Gambit (3.Nf3 d6), word is that he played and won three tournament games as White using the Bishop's Gambit; here are two:,

3...Qh4+  I always take this check.  I think it messes up White more than Black.  And I have a simple system I developed using the computer.  The other common responses to the Bishop's Gambit are 3...Nf6 and 3...d5.

4.Kf1  Forced.  As I covered in the Knight's Gambit Part 1 post, 4.g3 loses a piece:  4...fxg3 5.Bxf7+ Kxf7 6.Qf3 Nf6 7.hxg3 Qxe4+ 8.Qxe4 Nxe4, or 5.Kf1 g2+ 6.Kxg2 Qxe4+ 7.Nf3 Qxc4.  If 4.Ke2, easiest is to snag an additional pawn via 4...Qg4+ 5.Ke1 Qxg2 6.Qf3 Qxf3 7.Nxf3 Nf6.

This sets up my simple system targeting Ng4 and Ne3+, before or after Nf3 Qh6.

5.Nc3 Ng4  Before.  And now mate and Nxh2+ are both threatened.

6.Nh3  7.Qf3 is the only other option and it runs into 7...Ne5 8.Qe2 Nbc6.

6...c6  Keeping the knight out of d5 and b5.  Also preparing b5 and perhaps d5.

7.d4 g5  Standard call and response, black holding f4.

8.Bd2  Trying to develop. Also maybe eying Be1 to chase away my queen.

8...d5?  I don't know why I just throw away this pawn.  I think I just wanted to get my c8 bishop out with tempo.  8...d6 is much better.

9.dxe5  Now I notice my king has been exposed and so I had better pull the trigger:

9...Ne3+ 10.Bxe3 fxe3  Obviously I can't hold that pawn, but White has to find the right way to hit it.

Apparently trying to hit (and pin) e3 and to expel the queen on h4 (to stop the Bxh3 threat) simultaneously.  But he apparently missed that it gives me a really good reply.  White had to find 11.Qd3.

11...Qxd4  And I find the right reply.  Gain a pawn, defend e3, and hit the undefended c4.

12.Bb3 Bc5?  I guess I was lining up on f2, and also looking to castle quick, but I needed to see that g5 was hanging and find g4.  Sadly, this non-optimal move took my longest think of the game, 3:15.

13.Rd1  Again trying to expel my queen and again apparently missing that I have a good reply.

13...Qf6+ 14.Nf2  Yes he has to give up the knight because Kg1 loses the d1 rook to the discovery e2+.  But now it's my turn to mis-calculate.
14...O-O  I must have thought this was necessary for getting the knight, or maybe just a stronger way to take it, or I was still obsessed with getting quickly castled.  Boy was that wrong.  Now it becomes just a piece trade, because I failed to see my opponent's fork threat.  Lesson:  consider all ways to capture, and don't be spooked by a pin that will dissolve.

15.Nce4  Fork!  Lesson: always look at opponent's possible attacks, not just your own.

15...Qxb2  I saw this as an exploit of the knight move.  But again, I must have missed Nd3 by which the knight escapes.  But my opponent chooses to liquidate:

16.Nxc5 exf2 17.Qxf2 Rd8 18.d6
I thought I had a nice pin, but of course the pawn is not pinned from moving forward, silly.  And now I have to deal with f7.  I should mention another factor at this point:  I now have 1:10 on the clock while my opponent has 10:14.

18...Qg7  Best, not hard to find.

19.d7?  Now it's my opponent's turn to throw away a pawn needlessly; it also allows me to unwind my cramped pieces.  He must not have seen the knight (3rd piece) covering d7.  But more likely he is just running down my clock.

19...Nxd7 20.Nxd7?  Much better would have been to just leave my knight pinned and my bishop thus blocked in.  Again, just a clock move.  I am at a sad 0:09.

20...Bxd7 21.Bxf7?  0:08

21...Qxf7 22.Qxf7 Kxf7  0:04

23.Kf2 Rf8  0:02

24.Rxd7+?  And I forfeit on time.

Well the obvious lesson here:  Learn how to use my time better.

Thanks for reading.  Comments welcome.

Wednesday, May 16, 2012

Part 2: King's Gambit (King's Knight's Gambit), Internet Chess Club

This is the second part of a post about a game I played last weekend on ICC.  See Part 1 for the opening:

One-DarkKnight (1644) - jientho (1277)

I've completely got this game in the bag, right?  Materially up rook, knight, and 2 pawns.  No obvious threats by White.  At best White gains the pawn by 13.Bxe5 Bxe5 14.Qxf6 Nxf6 15.Rxe5+ Kf8 right?  So how did I blow it?  Let's see:

13.Qg3  All this seems to do is to load up further on e5 and get away from a queen swap.  Yes it takes away one additional flight square from my queen (g7) while the queen attack Ne4 is in the air, but my queen still has plenty of options especially g6.  And all I need to do is to break the pin on e5 and the bishop on f4 is immediately attacked.  As I recall, the only thing above that I didn't see during the game was the Ne4 attack.

13...Be6  I'm pretty sure that my thinking during the game was, first, try to trade off pieces (because I'm way ahead material).  Second, I also saw this as a double attack on c4 and f4 (because it breaks the pin on e5).  I was also vaguely aware that I could let the bishop on a1 drop if it meant I could develop other pieces and that the White rook would have to leave the action to get it.

14.Ne4?  There it is.  A good shot against a lower-rated player.  Because I hadn't seen it in advance and probably also because I was aware of a significant difference in remaining time budgets (6:27 for me, 9:59 for White at this point), I started to go into panic mode.  Always a bad thing for a chessplayer, especially a "novice", because it narrows the imagination down even further than it already is.  If I could wave a magic wand, panic is the one thing I would eliminate from my chessplaying.  Because even if I can't find the absolute best moves, at least I want to "see everything" (at a basic level) that's going on on the chessboard.

14...Qg6?  Not horrible.  It doesn't lose the game or any material.  But oh what I missed.  I was so focused on my threats of exf4 and Bxc4 and probably in trying to trade queens that I missed Qxf4, according to the computer.  OK I shouldn't be so hard on myself because, looking again, I must have seen White's Qg7 threat as well.  And it's not at all obvious how to handle that or whether Qxf4 is worth it.  Even now with plenty of time, I am not seeing it.
OK, I "cheated" to learn what I was missing.  I did not see any way to both protect the knight on g8 and counter the threat of Nf6+, thus thinking I would have to lose both the rook and the knight (with check) for the two bishops.  For some reason I thought the king was frozen out of f8 even after Qxh8.  Lesson to learn: consider all moves which can defend a piece after an otherwise favorable liquidation; also always know which side is "on move" during a trade-down (I assumed Nf6 would necessarily be a check, forking king and pinned knight; not so if I have an extra move to counter that).  The line:  14...Qxf4 15.Qg7 Bxc4 16.Qxh8 Kf8.

15.Bxe6 Qxe6?  Complete blunder.  Here again panic must have been a factor.  I just completely lost the previous threads of:  1) Qg7 threat (thought it was permanently addressed already or something?) and 2) trading queens (maybe I thought I was in check?).  Also possible paranoia that my opponent was trying to "trick" me into removing the support of f7 for my queen and doubling up my pawns.  Why oh why does my brain sometimes turn into spaghetti? :-|

16...Nd7  Somehow I stumble into a good move.  Unfortunately I took 2:21 to come up with it, leaving me 3:06 on the clock.  This and my next move were my plan for saving the g8 knight with development.  No way was I going to venture exf4 now, even if I had seen the bishop on a1 hitting g7, because I definitely saw the coming rook pin against e6 with Nf6+ or Nd6+.

17.Qxh8 Ke7?  This is much worse than the other way to get the a8 rook to defend g8.  Right idea, wrong execution.  What do I need to do, besides ignoring the clock, to get myself to double-check for an improved way to accomplish something?  Just because we're on move 17 doesn't mean O-O-O is not available.  I think it is a matter of keeping more possibilities in my consciousness at the same time.  There is just no excuse for a chessplayer not at least keeping all legal single moves in mind.

18.Bg5+  Of course, removing the bishop from attack with check.

18...f6 19.Qxh7+ Qf7  I do some pretty good defending here, for a while.

20.Qh3 fxg5 21.Nxg5 Qxa2  Inviting a king chase.
Qg6 was better, hitting the knight while still stopping nasty queen checks at e6 and h7.  I am now down to 0:47 on my clock while opponent has 5:06.  With no forced mating plan in place, it's all but hopeless.

22.Qh7+ Kd6 23. Rd1+  My thoughts at this point are to try to run the king toward a3 whenever I'm checked, and otherwise to try to catch my opponent unawares somehow with a QxR#.

23...Kc5?  I am not seeing that the knight on d7 is hanging.  Bd4 is the move.

24.Ne6+  Haha.  Opponent does not see the knight hanging either, but instead moves his own knight en prise.  Cute.  I guess he's just in run-down-opponents-clock mode.  He has 4:02.

24...Kb4  He was right that I wouldn't see Qxe6.

25.Qxd7  Now he does see my knight hanging.  I forfeit on time here.

Very discouraging, especially considering how some folks play successful 1-minute chess and with 1 minute left I still had a 4.56 advantage by computer eval.  So another lesson learned: learn how to optimize my own play at different time budgets.

And of course the overarching lesson: don't lose won games.

Thanks for reading.  Comments are welcome.

Tuesday, May 15, 2012

Part 1: King's Gambit (King's Knight's Gambit), Internet Chess Club

I played in the ICC Marathon tournament May 12-13, 2012.  ICC is the Internet Chess Club (  Every 2nd month they have a 24-hour marathon of tournaments noon Saturday to noon Sunday (  I do relatively poorly, playing in the "standard" section (each player having a 15-minute budget per game), but I usually can get some membership extension prize by being in the top six in total number of games played.  For example this time I got 8 points out of 32 games (and that includes some "unearned" full-point byes, which happen sometimes when there are an odd number of players and the "lowest man" is thus not paired to play).  By comparison, the player with the highest number of games played won something like 20 points by playing 45 games; and the player with the highest score won something like 27 points from playing 33 games.  (The marathon consisted of 47 rounds total).

But I want to do better, and have more fun, so I want to publicly analyze one of my games here in great detail, trying to recall my thought process, and being brutally critical.  Maybe it can also help other lower-rated players like myself.

I have chosen a game whose opening I love to play as Black, the King's Gambit.  I got one each of the King's Knight's Gambit and King's Bishop's Gambit this marathon, and I'll do the Knight's game in this first post.  The Bishop's in a later post, if this one proves useful to me or to others ("on request"), also time permitting.  I am "jientho" on ICC.

So, this is the game One-DarkKnight (1644) - jientho (1277).

1.e4  Something like 43% of games start with e4, 40% with d4, and 18% other (mostly Nf3 and c4).
1...e5  Of the games where White starts 1.e4, Black's replies break down approximately as follows:
  1...c5, 47% (Sicilian Defense)
  1...e5, 26% (Open Game)
  1...e6, 12% (French Defense)
  1...c6, 7% (Caro-Kann Defense)
  1...d6, 3% (Philidor Defense, Pirc Defense, or Hedgehog Defense)
  1...g6, 2% (Modern Defense or Pirc Defense)
  1...Nf6, 1% (Alekhine's Defense)
  1...d5, 1% (Scandinavian Defense)
  1...other, 1%
Combining percentages, of all games, about 11% start with 1.e4 e5.

2.f4  And here is the King's Gambit.  White has not defended the f4 square before sending the f pawn there.  As such, the full intent of this move is to distract Black's e5 pawn or to force her into awkwardly trying to defend it.  Rather than wait until d3 or d4 is played first, when the f4 square would be defended by the bishop on c1, White holds the move d4 "in reserve" to attack the black pawn on f4 if Black accepts the gambit.  Another note here -- whenever the opponent moves f3, f4, or g4 early in the game, I keep in mind that that is half of the Fool's Mate: just in case the other shoe drops, I don't want to miss a mate-in-one.
2...exf4  I always take the offered material both because consensus is that this is best for Black and also because it leads to more interesting, attacking, dynamic, open positions.  This is called the King's Gambit Accepted or "KGA".  I'll mention one version of the King's Gambit Declined or "KGD" because it sets a trap for White, although I never play it because the White players who opt for the the King's Gambit surely know about it.  Called the Classical KGD, it's 2...Bc5, which "looks through" White's gap on f2 and prevents White castling while apparently not defending the e5 pawn at all.  But White loses the pawn on e4 and the rook on h1, if not the game, if she takes on e5 now, because of two sequential Queen forks, first of e1 with e4, then of e1 with h1:  2...Bc5 3.fxe5 Qh4+ 4.g3 (if 4.Ke2 (the only other move), then 4...Qxe4 is checkmate!) 4...Qxe4+ 5.Qe2 Qxh1.  So if White wants actually to threaten the e5 pawn in the Classical KGD, she needs first either to keep the queen off of h4 (3.Nf3) or to protect e4 (3.Nc3 or 3.d3).

3.Nf3  This is the King's Knight's Gambit.  The main/immediate purpose is prevention of the queen check on h4, which was made available to Black by 2.f4.  Normally g3 would be an effective block to that check (if e.g. Black tries the immediate 2...Qh4+).  But once Black has her pawn on f4 from 2...exf4, then g3 becomes ineffective as a block because it can be captured with advantage.  From the King's Bishop's Gambit: after 3.Bc4 Qh4+ 4.g3? fxg3, the recapture loses because the h pawn is pinned (5.hxg3 Qxh1); meanwhile the discovered g2+ is threatened (winning the rook with promotion), as well as Qxe4+ winning the undefended bishop on c4.  White's only correct response after 3.Bc4 Qh4+ is 4.Kf1.

I wasn't going to do the Bishop's Gambit in this post, but in case I don't get to it in a separate post, I want to make a first confession.  I did not know White's best response to 4...fxg3 above and I did not know how Black should handle that response, in case White really does try 4.g3 to test Black's preparation.  White's best is not 5.Kf1 getting out of the discovered check, because 5...g2+, although it doesn't get the rook, still does the job (gains the bishop plus another pawn): 6.Kxg2 Qxe4+ 7.Nf3 Qxc4.  (5...d5 may be slightly better but is much more complicated).  White's best is actually to throw the bishop in a different way: 5.Bxf7+ Kxf7 6.Qf3+, both checking and guarding h1 so h2 is unpinned.  Black now must see that 6...Nf6 is best, allowing her to answer 7.hxg3 with 7...Qxe4+ and taking the queens off the board.  (If White doesn't take g3, then g2+ is still in the wind.)  If confronted with 6.Qf3+, I almost certainly would have panicked (in my novice way) that 6...Nf6 would be met with 7.e5 attacking the pinned knight.  But in that case, 7...g2+ is absolutely devastating: 8.Kd1 Qe4 and White must allow gxh1Q or give up her own queen to prevent it.

The lesson to learn is, always know your own leverage (the power of your threats) on the board, so you are never needlessly cowed by an opponent's weaker threats.  In the above case, even if White tries to hold everything but the rook by answering 8...Qe4 with 9.Qg3, and then answers 9...gxh1Q by taking the knight as feared, 10.exf6, the blocking move 10...Qeg2 is devastating, but 10...d5 is even better, a mating trap.  (For heavens sake just from a material standpoint, you've turned a pawn into a 2nd queen plus traded the knight for a rook.)  Although there are slightly better defenses after 10...d5, White would almost surely try to get Black to blunder into her own mate (or think she might get a perpetual) with 11.Qxc7+, but alas that is a road to a cute mate-in-4:  11...Kg6 (11...Ke8?? or 11...Kg8?? 12.f7#) 12.Qg3+ Bg4+ 13.Qxg4+ Qxg4+ 14.Ke1 Qhxg1#.

3...d6  The Fischer Defense.  The story goes that Fischer was so devastated by a 1960 loss to Spassky when using the Classical Variation 3...g5 ( that he developed the little-used 3...d6 into a full-blown answer (he called it a "bust") to the King's Knight's Gambit.  Modern analysis shows that Fischer was actually ahead in that game until move 25 or 26, although I suppose it's possible that the 17-year-old Fischer, in pre-computer days, could have exaggerated the responsibility of the opening in his own mind.  Then again he was one of the top players in the world even then and probably knew better.  Regardless, he did show that the ideas used by White after 3...g5 don't work after 3...d6.  Most directly, after 4.d4 g5 5.h4 g4, White no longer has the Kieseritzky Gambit idea, 6. Ne5, available to her; and neither is the alternative Allgaier Gambit idea, 6.Ng5, workable because the bishop on c8 now holds the g4 and h3 squares for Black's g pawn, giving Black this resource: 6...f6 7.Nh3 gxh3 8.Qh5+ Kd7 9.Bxf4 Qe8 10.Qf3 Kd8.  So now the main line after 5...g4 is 6.Ng1 Bh6 and Black holds on to the extra pawn, with development.

Fischer may have busted the old ideas, but White has since found that 3...d6 opens the way for a new idea to help in relocating her knight if Black tries to kick it with g4, namely 4.Bc4.  After 4...g5 5.h4 g4 6.Ng5, now threatened is 7.Nf7 ouch, forcing Black to prevent first with 6...Nh6, and afterward Black's 7...f6 kick is not even a threat because the knight then has 8.Ne6.  So Black needs to answer 4.Bc4 with 4...h6.  And the King's Knight's Gambit itself is not busted.  Fortunately my opponent in this game didn't play 4.Bc4, but unfortunately I soon miss a correct continuation.

4.d4 g5  This is a standard "call and response" in the KGA.  White pulls the d4 move "out of reserve" attacking f4; Black tries to hold on to the extra pawn in the most economical way, also opening g7 (or sometimes h6) for her bishop.

5.Bc4  As shown by Fischer (see above under 3...d6) 5.h4 does not really work here.  But neither does anything else particularly.  5.h3 is a try.
White has waited too long to play Bc4: now that Black has played g5, the threat of g4 is too strong.  Unfortunately I miss it.

5...Bg7  Not a particularly bad move, g7 being, usually, a good strong outpost for the bishop in the Knight's Gambit, hitting d4 and also making h6 now a good response to h4 since the rook h8 is protected.  It's just that 5...g4 would have been much stronger: if the knight moves, Qh4+ is strong, e.g. 6.Ng1 Qh4+ 7.Kf1 f3 8.gxf3 Nf6; and otherwise the knight is lost, e.g. 6.O-O gxf3 7.Qxf3 Qf6 with the threat of Qxd4+ more than countering Black threats against f4 and f7.  This is like a Muzio Gambit except with d4 and d6 having been played.  The main line from here seems to be 8.e5 dxe5 9.dxe5 Qg6.

So what is the lesson?  When White has not played h4 nor castled, g4 is a strong move for Black.  Or more generally, look for simple attacks on pieces that have limited/bad escape squares and/or which are guarding squares you'd like to move to (Qh4 in this case).

5.O-O  Castling out of all the danger.  Best move.  Although I note the possibility of Bxd4+ later.

6...g4  A day late and a dollar short?  White now has the good safe e1 for the knight (and no more Qh4+ threat).

7.Bxf4 gxf3 8.Qxf3  In true gambit fashion, my opponent opts for sacrificing the knight anyway!  In return for big pressure toward f7, Muzio-style.

8...Bxd4+  Of course.

9.Kh1  Best.  In a previous over-the-board game in a similar setup my opponent opted for 9.Be3 perhaps thinking of the immediate exposure of f7 to his queen and rook, but of course 9...Bxe3+ being a check either gives Black time to cover up f7 or draws his queen off of f7.

9...Qf6  This always seems to be a necessary response to a Muzio setup.  Black must be in a position to trade off queens if/when White moves the bishop.  We've also entered a zone of maximum complexity in the position, i.e. the middlegame.  It's hard for either player to know for certain how to proceed.
Materially Black is a whole knight up, but she is at least 3 moves behind in development and still facing a barrage on f7. Black's immediate threat is of course Bxb2.  The straightforward 10.c3 Be5 followed by liquidation and gaining of the f7 pawn doesn't really give White enough back for the knight:  11.Bxe5 Qxf3 12.Rxf3 dxe5 13.Bxf7+ Ke7.  White's strongest seems to be 10.Qb3 which accomplishes 3 things at once:  attacks f7 albeit now from behind the c4 bishop, attacks b7 preventing Black from defending f7 with Be6, and removes the queen from the line of sight of Black's queen leaving the attack as rook vs. queen instead.  Black has to either defend with Nh6 (probably best), or else pull the trigger on b2 apparently resulting in a liquidation of 2 rooks for her queen and ultimately a likely draw by perpetual.  This would need to be further examined to be sure, but here is the line:  10.Qb3 Bxb2 11.Bxd6! Qxf1+ 12.Bxf1 Bxa1 13.Bxc7.

10.e5  A clear blunder.  Although e5 is "thematic" in the normal Muzio, my opponent must have overlooked that the d4 pawn is missing or thought that I couldn't retake with the d6 pawn for some reason...

10...dxe5 11.Re1  ...or that this rook pin would be sufficient to win the pawn back.  Of course, the pawn is now also protected by the bishop on d4.

11...Bxb2  The obvious attack.  White has some countering possibilities that I did not see, namely 12.Nc3 regaining control of e5 and thus getting my h8 rook in trade via Bxe5, and the immediate 12.Bxe5 forcing 12...Bxe5 and a trade of bishops, due to the pin, instead of rooks.  But trades are better for the one who is already up material.

12.Nc2  Neither did my opponent see those counterplays.

12...Bxa1  And now I am completely dominating.  How in the world did I lose this game?  Complications, my friend.  Stay tuned to find out in Part 2.  <cliffhanger/>

This is getting much longer than I thought it would, so I have decided to serialize it.  Until then...

Monday, May 7, 2012

Smith-Morra Gambit (or Morra-Smith or Mora)

I want a main line for chess's Smith-Morra Gambit: 1.e4 c5 2.d4 cxd4 3.c3

Granted, from what I said in the previous post (Chess Theory), the concept of a main line in chess is theoretically incoherent.  Because all playable lines lead to draws, there is no such thing as "best play", at least not in any single-line sense.  (I am assuming the Smith-Morra is playable.)  All roads lead to Rome.

On the other hand, it is certainly possible, based on computer and human evaluation, to find what appears to be a "best line of play", or at most a handful of such lines.  That is what openings books are all about.  From a theoretical perspective, this is not an exercise in finding the best line(s), but rather in finding the longest line(s) before the inevitable draw.  The most scenic routes to Rome.  This is directly due to the horizon effect: the only reason humans and computers don't see all reasonable moves as leading to a draw is because they cannot look far enough (and thoroughly enough) ahead.  But scenic is good too.

So here is what I currently consider to be the main line of Smith-Morra:

3...dxc3 4.Nxc3 Nc6 5.Nf3 d6 6.Bc4 a6 7.O-O Nf6 8.Bf4 Bg4 9. h3 Bh5 10.Qb3 Qc7 11.Ng5 e6 12.Bxe6 h6 13.Nd5 Nxd5 14.exd5 hxg5 15.dxc6 fxe6 16.cxb7 Rb8 17.Qxe6+ Be7 18.Rac1 Qxb7 19.Bxd6 Qd7 20.Qxd7+ Kxd7 21.Bxb8 Rxb8 22.b3

if 8...e6 then 9.Qe2 Qc7 10.Rfd1 Nh4 11.Be3 Be7 12.Rac1 O-O 13.Na4 Nf6 14.Bb6 Qb8 15.Bb3 Nd7 16.Qe3 Nf6 17.Nc3 Nd7 18.Na4 draw by repetition.

If Black wants to decline the gambit, this is what I consider to be the main line of the best try:

3...d5 4.exd5 Qxd5 5.cxd4 e5 6.Nf3 exd4 7.Qxd4 Qxd4 8.Nxd4 Nf6 9.Bb5+ Bd7 10.Bxd7 Nbxd7 11.O-O

Can anyone find anything better/longer?

Thursday, April 26, 2012

Chess Theory

Standard chess is, beyond any reasonable doubt, a draw with best play, i.e. a theoretical draw.  That means, from a theory perspective, that all positions in all "playable" openings are of exactly equal evaluation (0.00 in chess computer terminology, "=" in opening book terminology).  So what are all the other evaluations appearing in openings books and on computer chess programs?  From a theory perspective, those are all just probabilistic estimates as to a position's true theoretic value, which can only be one of ("white wins", "draw", "black wins"), or in computer terminology ("+#", 0.00, "-#"), or in opening book terminology ("+-", "=", "-+").

The interesting and wide-open questions are then:
  1. Where exactly is the line between "playable" and "unplayable" in the opening?  Or, how big is the gray area between "playable" and "unplayable", can it be reduced, and if so how and how far?
  2. What is the function from computer evaluation value to numeric probability?  Or is this an apples-to-oranges comparison because of the inherent uncertainty in the value (dependent as it is on how much, how deeply, and which parts of the move tree the computer has searched, let alone the computer's underlying base (0-depth) evaluation function)?

Wednesday, April 25, 2012

Corbomite Details

So, what is Corbomite?  Well, according to Kirk (Star Trek, Original Series, "The Corbomite Maneuver"), it is "a substance", "a material and a device which prevents attack on us. If any destructive energy touches our vessel, a reverse reaction of equal strength is created, destroying ... the attacker."  Some kind of substance rolled into some kind of material rolled into some kind of device.  Huh?  Did Shatner flub his line or did the author (Jerry Sohl) really write this?  Maybe the idea is that Kirk, clever lad, already had suspicions that Balok was bluffing, and only needed a vague and hardly-believable bluff of his own to probe his antagonist in turn.  Of course the suspense of the episode depends on that not being the case, that it was true desperation/bravado on Kirk's part.  Interesting that the intent of the ambiguous bluff is itself ambiguous. #multilevelwriting

Also somewhat interesting is that this is a fiction (bluff) within a fictional story.  That's hardly uncommon though, deception being a frequent plot device.

But to make it the title of the episode??  Good lord, that is like titling "A Piece of the Action" as "The Fizzbin Maneuver" instead.  How about "Space Fear" or "Face of the Ultimate Unknown" or something.  The portrayal of Dave Bailey was actually quite good in this episode, especially in his relationship to his captain.

There is one other reference to Corbomite in the series.  Again Kirk is using it as a bluff, but this time he describes it as a self-destruct device, in "The Deadly Years".  Come on Kirk, make up your mind.  :-)

Monday, April 23, 2012


Well, nothing like a blog for dredging up memories and thoughts.  This started as a questioning thought: what is the actual scientific meaning of the prefix corbo- as used in "Corbomite" from Star Trek (Original Series episode "The Corbomite Maneuver")?  Wouldn't that make a good trivia question?  Feel free.

Sounds like a variant of carbo- doesn't it.  Carbohydrate, carborundum.  Not so; it appears to be completely made up.

Here is a complete list of words in my Webster's Unabridged starting with corb-, categorized:

Rooted in Latin corbis (basket):
  corb, a basket used in coal mines
  corbeil, a basket (of dirt or flowers) on a wall
  Corbicula, a genus of mollusks
  corbiculum, pollen basket of a bee

From Hebrew korban (offering) from karab (to offer):
  corban, an offering to God
  corbana, a church treasury

From Latin corvus (raven; shape of its beak):
  corbe, curved
  corbel, (to form or add a) curved architectural bracket
  corbel(l)ing, making of or series of corbels
  corbel step, a corbiestep
  corbel table, any piece of architecture using corbels
  corbet, a corbel
  corbie/corby, Scottish for crow, raven
  corbie crow, the carrion crow
  corbie gable, a gable with corbiesteps
  corbiestep, a step forming part of a roof

As for the -ite suffix, possibly relevant entries:
  (f) salt or ester of an acid whose name ends in -ous (corbomous acid?)
  (g) mineral or rock (discovered by a person named Corbom?)

Oh well, I guess Balok was not familiar with the Federation linguistic banks.  So how did he speak English so well?  :-)

This appears to be the closest real-world inspiration for the term: