Wednesday, 25 December 2019

058. Ulvestad's Second Variation

White: AndyAndyO - thematic tournament,, 2019

Olaf Ulvestad was a US master and openings theoretician, now known primarily for his variation of the Two Knights Defence: 4 Ng5 d5 5 exd5 b5!?. Ulvestad wrote enthusiastically about his discovery in the 1941 issues of Chess Review: “It stands out head and shoulders above all the other moves as the strongest, sharpest and best.” Okay, he's getting carried away there perhaps, but 5...b5 certainly very interesting.

The critical reply – indeed, the only good reply – is 6 Bf1!. Against this, Ulvestad originally proposed 6...h6 but later discarded it, albeit for the wrong reasons (7 Nxf7! Kxf7 8 dxc6 Bc5 9 Be2! is why this isn't any good). 6...Nd4 7 c3 Nxd5, transposing to the Fritz Variation (5...Nd4), has its drawbacks too (such as 8 cxd4! Qxg5 9 Bxb5+ Kd8 10 0-0).

Ulvestad's second idea, 6...Nxd5!? 7 Bxb5 Bb7, looks rather more promising. For example, 8 d4 exd4 9 Qxd4? Qe7+ 10 Qe4 Qxe4+ 11 Nxe4 0-0-0 12 Bxc6 Bxc6 (Ulvestad) is good for Black, since 13 Ng5 f6! 14 Nf7 loses quickly to 14...Bb4+ and ...Rhe8.

Taking the d4-pawn is clearly premature here. White should just castle: 9 0-0 “etc” (Tartakower). This is generally regarded as a refutation of 6...Nxd5, following analysis by GM Reuben Fine: 9...Be7 10 Qh5 g6 11 Qh6 Qd7

12 Qg7 0-0-0 13 Qxf7 (Fine), when White has regained, and appears to have consolidated, the extra pawn: 14 Ne6 is a threat, or if 13...Rhf8 then 14 Qe6 swaps the queens off. So it is somewhat surprising to discover that the engines think that Black is better, casually throwing out the disruptive 14...Ne3!. Well, splendid!

White is not obliged to play 12 Qg7?!. Instead, D.Pena-C.Fonseca, Pamplona 2012, saw 12 Re1 0-0-0 13 Nxf7 Bb4 14 Nxd8 Bxe1, and here 15 Nxb7 Kxb7 16 Bd2 would have been very good for White.

But once again the engines are ready with an improvement: the nonchalant 12...a6!, intending 13 Ba4 Qf5. I think it might be difficult to decide upon this over the board. The main point is that 12...Qf5? gets hit by 13 Bd3, but that's impossible once the bishop has retreated to a4.

Many possibilities arise from White's options at moves 13 and 14 (and move 17 too), but in all of them the activity of Black's pieces provides excellent compensation for the pawn, even into the ending. The game below is a case in point, where I forced the queens off (with 18...Bf8) just to see how easily it might be defended. As it turned out, I didn't have much trouble at all.

I should also add that my opponent also managed to draw as Black, this time after 6...Qxd5!? 7 Nc3 Qc5. It seems there are still plenty of secrets to be uncovered in these old Open Games.

Thursday, 31 October 2019

057. Not the Halloween Gambit

White: A. Faller - BCCL Premier Division, 1991

On October 31st I should be posting a game with the Halloween Gambit: 1 e4 e5 2 Nf3 Nc6 3 Nc3 Nf6 4 Nxe5. There are two reasons why that isn't happening:

1. It's rubbish. After 4...Nxe5 5 d4 Ng6 6 e5 Ng8 7 Bc4, GM Bologan's straightforward 7...d5! 8 Bxd5 c6 9 Bb3 Be6 is just good for Black.

2. Although not deterred by rubbish (as you'll see below), I haven't actually played any games with the Halloween Gambit, from either side.

So, instead, here's a Belgrade Gambit: 4 d4 exd4 5 Nd5!?. My reply 5...Nb4!? was unusual at the time, and the follow-up 6 Bc4 Nfxd5!? 7 exd5 Qe7+ even more so.

My plan was to win a pawn after ...Qc5, which I duly achieved. Unfortunately, White's lead in development more than made up for it. (Bologan, in 2014, rightly preferred 6...Nbxd5 7 exd5 Bb4+!.)

Inevitably, Stockfish puts up a better defence than I managed – first of all with 8...d3! (which I saw, but it wasn't my plan) and then 12...Qd6! (which I never considered at all).
A few moves later, my opponent went very wrong himself with 16 Rh5?? (the preliminary 16 Bxh7+! wins), when I might have escaped with 16...g6 but rejected this because of 17 Rxh7 Kxh7 18 Qh5+, overlooking 17...Bf6!.

And following that fine display I soon obtained the CC-IM title. Perhaps I should have turned it down.

Sunday, 29 September 2019

056. Computer Says Yes

Black: K.J. Bowyer - C&DCCC Ward-Higgs Trophy, 2000

Another main line Ruy Lopez, another sideline defence by Black, this time 10...Qd7!? in the Zaitsev Variation.

Searching MegaBase shows that Averbakh played this first (in 1962) and won, and Geller later won twice, albeit all against weaker opposition. More recently, GM Igor Kovalenko has tried it a couple of times, both winning and losing.

Objectively, it probably isn't very good. I say that because, facing this line as White, I just played natural, obvious moves and won very smoothly indeed. For instance: 15 d5, 17 a4, 19 Bf4, 20 e5, 26 Nc6, 27 a5, etc. I'm sure Ken saw all these coming and yet was unable to do anything about them.

Which is perhaps the definition of a dubious opening. You play something "unusual", your opponent replies with natural, obvious moves, and you lose without ever having any chances at all. All very depressing.

But what does Stockfish have to say about it?

Computer says yes.

And that's a total yes. Stockfish would have played every single one of my moves. It did stall for a moment over two or three but quickly came round. In other words, the engine says it was a perfect game.

Well, that's really quite gratifying :)

Monday, 26 August 2019

055. Computer Says No

White: M. Poppe - 1st North Sea Team Tournament, 1998

Searching my games by ECO code, looking for one that hasn't featured here before, brought up (along with numerous others) three examples of C96. This code covers lesser lines of the Closed Ruy Lopez, Chigorin System (9...Na5). All three games continued 10 Bc2 c5 11 d4 Bb7 12 Nbd2 cxd4 13 cxd4 exd4 14 Nxd4 Re8.

Yes, it's a bit loose for Black in that the d6-pawn is left as a weakness, backward and isolated. On the other hand, there's little prospect of being cramped under the Spanish Torture. Black has open lines and active pieces and can look to strike back in the centre with a timely ...d6-d5.

GM Oleg Romanishin (see Game 29) played this way on 24 occasions in 1988-97, for a plus three score. My own score was plus one: two draws and the win below.

It seems I was pleased with the win, since exclamation marks are strewn enthusiastically through my notes: 19...d5 “!”, 20...Bc5 “!”, 21...Nh5 “!”. Okay, I'll just check those with Stockfish now...

Tap, tap, tap.

Computer says no.

And that's no to all of them. In particular, the engine bangs out 21 Ngf5! instantaneously, assessing the position as winning for White; e.g. 21...gxf5 22 Nxf5 Bxf2+ 23 Kh2 Bxe1 24 Qxe1 and Black's extra rook is useless against the white queen coming to the kingside. Yurk. My notes don't mention 21 Ngf5 at all.

To be fair (to me) 23...Rxe2 “!” was good, and the idea of trapping the h6-bishop with 28...g5 “!” was good too. Computer says yes. But as for my earlier moves...

Computer says no :/

Saturday, 10 August 2019

054. Playing On ’Til Checkmate

White: rgs56 - all-play-all tournament,, 2018

There are always a few players who don't or won't resign in utterly lost positions but carry on ’til checkmate. Or, very often, the move just before checkmate. And that's fine.

In correspondence chess, back when we played by post, it was annoying because you were forced to waste money on stamps sending unnecessary moves back and forth. Your opponents might thus consider themselves slightly revenged on having been beaten, albeit at the cost of their own stamps and reputation.

In online server games pecuniary penalties do not apply, so the only effect of playing on is to postpone an inevitable result for as long as possible – a strategy which can, if desired, be further extended by not replying until the very end of the time limit for each move.

Take the game below, for instance. At move 22, Black's extra bishop and knight are better than White's extra rook and pawn. At move 46, I assessed my position as winning. And by move 65 I thought it was time for Black to resign. My opponent opted to play on for another nineteen moves, ’til just before checkmate.

No worries. I got to look at my happily won position for significantly longer, make an aesthetically pleasing bishop retreat to the far corner, with zugzwang, and deliver mate with a pawn. Or nearly.

And sometimes I'll carry on in losing positions myself – if I want the game to resolve in a particular way to my own satisfaction. And I like my opponents to do likewise. I prefer some things to be played out rather than "left in the notes". Sometimes I'll play on ’til mate too – if it's going to be a nice one, say, and will happen soon. Such as in Game 35. Anyone can play any position on if they want to, and for whatever reason.

As it happens, another of my games (against the same opponent) is going the same way. I'm currently a rook up with a passed f-pawn one square from promotion. At one move per week (I usually reply straight away), if the game runs its full course we'll be enjoying it for three months yet. One of us will anyway.

Sunday, 4 August 2019

053. Grandmaster

White: M.L. Nicholson - Koshnitsky Memorial, 2002


I'm not a CC grandmaster. The best I can say – and indeed have already said – is that I once got a GM norm. This came from a joint second place (on 10½/14) in the CCLA's Gary Koshnitsky Memorial.

I also managed to win against a proper, OTB GM (Colin McNab) by copying the moves of another, stronger GM (Evgeny Gleizerov). Colin's improvement, when it came, wasn't much of one and I won relatively easily. But this was a Modern Defence and hence not blog relevant.

Instead, here's another King's Gambit. For some reason I discarded my usual 3...h5!? as Black and opted for 3...g5 “!” and a Kieseritzky, the only one I've ever played in a serious offline game. The subsequent 6 d4 and 9 Be2 was analysed by GM Joe Gallagher in Winning with the King's Gambit (Batsford 1992), with the open-ended conclusion that “practical tests are awaited”.

Practical tests duly came and went and the line was more or less abandoned. In particular, Gallagher's 9...Nc6 10 c3 Bf5 11 d5 Nb8 12 0-0 Qxh4 13 Nd2 g3 14 Nf3 Qh5 15 Qa4+ Nd7 16 Rae1 is well met by 16...Bg7! 17 Bc1 0-0 18 Nf4 Qg4, as in C.Santagata-S.Sabaev, ICCF EM/M/A071 1999, when White is struggling to show the slightest compensation.

My game saw 12 Na3 Bg7 13 Nc4 and then 13...Bxc3+. That was the threat behind ...Bg7, with the idea 14 bxc3? Nxc3 15 Qd2 Nxe2 16 Qxe2 Bxd3 and wins, so I went ahead and played it. My various engines (Stockfish, Houdini, Deep Fritz, Deep Rybka) now all go for 13...h5 “-+”. I guess I was dubious about the significance of Black's extra f7-pawn in the typical Kieseritzky structure.

It didn't matter. It doesn't matter. 6 Bc4 is regarded as the critical continuation nowadays, though Black has a plus score there too.

Wednesday, 17 July 2019

052. Senior International Master

Black: Bj. Laursen - 6th European Team Ch., Preliminaries, 1999

The Senior International Master title was, if I remember correctly, created so that CC-IMs would have something else to play for, at a time when there were limited opportunities to make GM norms. I was a beneficiary of this myself. Even if, as with the IM title, I can't remember exactly how I qualified. Searching my databases I see I did have three further decent results in the years 1998-2002.

These were:
For a total score of +12 from 31 games, and without a single loss.

Previous posts already feature the North Sea Tournament (Game 28 and Game 31) and the Olympiad (Game 4), so here's a game from the European Teams. It has me on the White side of a Petroff.

I can't remember anything about this game. My notes have various “+= Nunn” assessments in the opening, from where I'm not sure. Possibly NCO, if it was out by then. There's also an obscure reference to E.Mnatsakanian-V.Maslov, USSR Spartakiad, Moscow 1963.

The natural 11 h5 was a novelty, against which my opponent enterprisingly castled short. It's not actually as bad as it looks, at least not with Stockfish in charge of the black pieces. Fortunately, Bjørn didn't have a rating of 3400+, and his queenside counterattack was ultimately unsuccessful.

Playing it through again now prompts no recollection either. My memory is not what it was. I had a game in the Sheffield Summer League a week or so ago which followed the same path as one against the same opponent three weeks earlier. Presumably he'd prepared something for me. If so, he was probably disappointed when I deviated early. But that wasn't due to any foresight or caution on my part. I'd simply forgotten what we'd played before and inadvertently went a different way on move eight.

Amnesia does have some advantages. Regrets, I've had a few, but then again...
I've forgotten what they were.

Monday, 1 July 2019

051. International Master

White: P. Sváĉek - 21st World Championship, semi-final 2, 1995

I can't remember what the criteria were for the CC-IM title back in 1995, but it was a performance of 8/14 in WC21/SF02 that clinched it for me.

Really? 8/14? Is that all? And in the Open Games I only managed 2/6, with four draws and two losses. It was 3/3 with 1 e4 c5 2 Nf3 d6 3 Bb5+! that got me over the line.

The game below is a Scotch Four Knights with the slightly unusual 12 c4!?.

This seems to have been a pet line of Czech correspondence players around that time, including my opponent, who has another three games with it in the databases. It's not particularly testing, and I made an easy draw. In fact, Stockfish gives me the edge in the final position. Magnus Carlsen would probably play this on with the two bishops.

My best was against his homophonous namesake, CC-GM Anders Ingvar Carlsson. Everything just worked for me, and the way I was able to creep into his position on three half-lines (b3-b6, c3-f6, g3-g6) was very enjoyable. I think this is an IM-worthy game.

Sunday, 9 June 2019

050. The Transvestite Attack

Black: Jarolim - Unrated game,, 2004

If you've ever clicked on my profile, you'll have read the words: “genderqueer femme”. In short, that means my gender is “queer” in some unspecified way, qualified by the placement “femme” within the butch/femme spectrum. To explain that in long would mean a conversation and likely some more reading on your part.

And what does it have to do with chess anyway? Not very much. Neither gender, nor sex, is a serious indicator of inherent chess ability. (Nuts to Nigel Short.)

But there is one instance where chess and non-normative gender collide, and that's the Transvestite Attack. This was an invention of US player Jack Young, and involves the moves 1 e3, 2 Ke2, 3 Qe1, 4 Kd1, whereupon White's king and queen are on each other's squares, wearing each other's clothes, as it were.

For instance: 1 e3 e5 2 Ke2 d5 3 Qe1 Bc5 4 Kd1 f5 5 Nf3 e4 6 Ng1 Nf6 7 b3 0-0 8 Bb2 c6 9 Ne2 Nbd7 10 f3 Qe7 11 Qh4 Bd6 12 h3 Be5 13 Nbc3 a6 14 f4 Bd6 15 g4 Nc5 16 gxf5 Bxf5 17 Nd4 Qd7 18 Nxf5 Qxf5 19 Be2 Ne6 20 Bg4 Nxg4 21 hxg4 1-0 was J.Young-D.Sarkisiam, USA 1988. This appeared in Rainer Schlenker's offbeat openings magazine Rand Springer, issue #46 (1989).

Obviously White's set-up has no merit whatsoever. On completing the manoeuvre White is clearly worse, having wasted three tempi with the royalty, and is now unable to castle. But it seems Jack's thing was to see what he could get away with. Hence a few characteristically silly moves, given a name for posterity, and start the game from there. It helped that he was (is?) quite a decent player, rated USCF 2261 in 1988.

I've never tried 1 e3 e5 2 Ke2 myself. Even if I had, it wouldn't be blog-relevant, and trying a similar thing in an Open Game (e.g. 1 e4 e5 2 Qe2 Nc6 3 Kd1 Nf6 4 Qe1) would probably lose by force. However, there is one opening in which it can arise naturally: 1 e4 e5 2 f4 exf4 3 Nc3 (the Mason Gambit) and now 3...Qh4+ 4 Ke2.

In this sequence the white king has to go to e2, and is going to have to move again; challenging queens with Qe1 is then often useful for White; and retreating the king to d1 frees the light-squared bishop. This situation has occurred several times in my praxis. Each time, Qe1 and Kd1 were actually correct and led to an advantage for me (even if I didn't always follow them up correctly). The game below is one example. (I've included another, more recent one in the notes.)

Here 9 Kd1 has uncovered an attack on the a6-knight, which is defending the c7-pawn. A further threat is 10 d4, regaining the f4-pawn with advantage. Black has no satisfactory way to solve these problems. Captain Transvestite strikes again!

Sunday, 2 June 2019

049. Queen vs. Rook and Pawn

White: Ares777 - all-play-all tournament,, 2018

An opening sacrifice (8...0-0!?) – first seen in G.Sanakoev-J.Boey, 10th World Correspondence Championship 1978 – gave me good play for a pawn in the early middlegame, which developed successively into a strong initiative, a material advantage, and (after 47...h4!) a winning endgame with queen vs. rook and pawn.

The Shredder endgame database gives this as mate in 50. I spent a long time trying to understand why this is so, searching MegaBase for similar positions, and getting my endgame books down from the shelf.

GM Karsten Müller & IM Frank Lamprecht (Fundamental Chess Endings, Gambit 2001) write: “If the pawn is on its original square or if it is a knight's pawn the position is drawn, as long as the attacking king is cut off and can't get behind the pawn.”

GM Efstratios Grivas (Practical Endgame Play, Everyman 2008) writes more expansively:
“(...) provided that the defending king and rook are close to the pawn and that the opponent's king cannot attack from behind. The basic rules are:
1. With a central (c, d, e and f) pawn, it's a draw if the pawn is on the second, sixth or seventh ranks, but otherwise the queen wins.
2. With a b- or g-pawn, it's always a draw.
3. With a rook pawn, it's a draw if the pawn is on the third or seventh rank, but otherwise the queen wins.”

And yet Shredder says I'm winning. Well, the authors are not wrong in what they say about the g-pawn, since they give the necessary proviso: if the attacking king cannot get behind the pawn. But it certainly doesn't look as if it can from the diagram.

The crucial stages are as follows:
  • First of all, 48...Qf2+! 49 Kh3 Qg1 threatens mate, so the rook has to drive the black king out and across the f-file (50 Rg4+ Kf5 51 Rf4+ Ke5).
  • Then (after 52 Kh4 Qh2+ 53 Kg4) the passing move 53...Qh1! creates a sort of zugzwang, in that White has to weaken his fortress slightly. In particular, if he moves the rook on the f-file (which offers the longest defence), either immediately or after 54 Kg5 Qh3 (seeing that 55 g4? Qe3 wins at once), the black king will advance up the board.
  • Then the queen returns down the board to hassle the white king from in front. Eventually, the rook has to give up control of the f-file, allowing the black king across and behind the pawn.
  • Black can then win the pawn, and then win with queen vs. rook.

The first three points above are exemplified in a superb endgame by the Czech master Emil Richter. From the position after 53...Qh1!, play continued 54 Rf2 Ke4 55 Re2+ Kd3 56 Rf2 Ke3 57 Rf4 Qb7 58 Kg5 Qg7+ 59 Kh4 Qh6+ 60 Kg4 Ke2 61 Rf5 Qg6+ 62 Kf4 Kf2 63 g4 Qd6+ 64 Kg5+ Kg3 65 Kh5 Qd7 66 Kg5? (making it easy) 66...Qg7+ and "White" resigned. I've put White in inverted commas there, because the moves were actually 76 Qh8! through to 89 Qg2+ and 1-0 Em.Richter-G.Stoltz, Karlovy Vary 1948, since the colours were reversed.

My game followed V.Burmakin-V.Lazarev, Werfen 1993, with 55...Qc1+! being a critical improvement. Black has to prevent the g-pawn from advancing freely. In the final position (where I claimed the game on time), White can in fact play 57 g4, but then the annoying pin on the rook means he has to waste moves with his king, allowing Black to get round behind; e.g. 57...Qd2 58 Kf5 Ke3 59 Kg5 Qd8+ 60 Rf6 Ke4 61 Kg6 Qg8+ 62 Kh5 Qh7+ 63 Kg5 Qg7+ 64 Rg6 Qe5+ 65 Kh6 Qe7 66 g5 Kf5 67 Rg7 Qf8 68 Kh7 Qa8 69 Rg6 Qh1+ 70 Kg7 Qh2 etc.

Incidentally, the Lomonosov tablebases show that the win can be considerably shortened by refusing the a-pawn on move 39. Apparently, 39...Qxa4 (its sixth choice) is mate in 59, whereas 39...Qe3! and 39...Qf7! are both mate in 37. I rejected those for two reasons:
1. White could have forced the game position by playing 39 Rf2 first.
2. Why on earth would Black refuse the pawn? I'd never refuse the pawn over the board.

Except that now I might. The a-pawn is not going anywhere. Yes, White can defend it with the rook, forcing the queen to blockade it for the moment. But when the black h-pawn comes up the board, White will then have to take it with the g-pawn, giving him two rook's pawns – until one of them drops off, after which he'll only have a rook's pawn, and on an unfavourable square, making the win that much more straightforward.

Damn, I love endgames.

Monday, 6 May 2019

048. No Problems

White: O. Bouverot - Koshnitsky Memorial, 2002

Two posts ago, I wrote: “After 1 e4 e5 2 Nc3 Nc6 3 Bc4 Black can play simply 3...Nf6 4 d3 Na5 and has no problems.”

Checking that statement against my own praxis, I came across the game below – which began as a Bishop's Opening: 2 Bc4 Nc6 3 Nc3 etc – where I managed to create problems for myself with an impatient break in the centre (12...d5!?), as my opponent's accurate moves demonstrated (14 Nb5!, 17 Kb1!). I later drew anyway after an exchange sacrifice (31...Bxg3!?) for a near-fortress on the light squares.

Looking at the opening, my notes indicate that I also considered 2...Nf6 3 d3 c6 4 Nf3 Be7!? 5 0-0 d6, reaching a Philidor set-up where White has already played d2-d3. This gives Black time for a quick ...Nbd7-f8 or even ...h7-h6 and ...g7-g5. But my favourite 2...f5!? I dismissed with the terse note “2...f5?! 3 d3”. I'm not sure now what I was worried about at that stage in my investigations.

As for 4..Na5, I'd thought Bronstein had remarked on the efficacy of this move for Black, but it turns out he was referring to 2 Nf3 Nc6 3 Bc4 Bc5 4 d3 Nf6 5 Nc3 d6 6 Bg5 Na5, as in V.Korchnoi-D.Bronstein, USSR Championship, Moscow 1952 (0-1, 53). Here's what he had to say about that:

Not so long ago, when Korchnoi had already become a serious contender for the world crown, I happened to ask him: “Why don't you play the Two Knights’ any more these days?” The grandmaster looked at me in amazement and muttered: “Because of
6...N-QR4, of course. You mean to say you didn't know that yourself?”

And 6...Na5 there still seems fine for Black, as with 4...Na5 here. So why does no one (much) play the Vienna/Bishop's Opening hybrid any more these days? Because of
4...N-QR4, of course. You mean to say you didn't know that yourself?

Or if they do... Searching for the position after White's 4th move (filters: date 2010-19, both players rated 2500+ Elo) in fact brings up 41 games (at non-rapid/blitz time limits). But within that sample 4...Na5 scores a hefty 58.6% (P29, W9, D16, L4) for Black – who clearly has no problems.

Monday, 15 April 2019

047. The Drawback to Reversed Openings

So there's a sharp system you like very much as Black. Why not play it reversed as White with an extra tempo? How useful would that be in the critical lines? If you're attacking the kings on opposing wings, say, an extra tempo might be very useful.

But critical lines are not the issue. What you need to think about is the non-critical lines, especially those which nobody plays because they offer White nothing and merely cede equality straight away. That's the drawback. In the reversed opening Black doesn't require an advantage. Equality will do just fine, thanks. And what are you going to do then as White? The game below is a case in point.

I wrote about the Scotch: Steinitz Variation (4...Qh4) in Game 19. The critical lines here involve throwing a knight into b5 with either 5 Nc3 Bb4 6 Be2 Qxe4+ 7 Nb5 or 5 Nb5 at once. Other ideas, such as 5 Nf3 or 5 Qd3 or 5 Nc3 Bb4 6 Qd3, allow Black a good position. A reversed version arises in Mengarini's Opening after 2 Nc3 Nf6 3 a3!? d5 4 exd5 Nxd5 5 Qh5. As you'll quickly realize, the extra a2-a3 rules out any ...Nb4 ideas here, so White can look forward to a “good position” in one of the less critical lines.

However, there was one sequence I'd forgotten about: 4...Qh4 5 Nc3 Bb4 6 Qd3 Nf6 and now 7 Nf5!?. In all my books, GM Lev Gutman's 4...Qh4 in the Scotch Game is the only one even to mention 7 Nf5.

As Gutman shows, White can ignore the attack on c3 after 7...Qxe4+ 8 Qxe4+ Nxe4, because 9 Nxg7+ Kf8 10 Bh6 Nxc3 11 Nh5+ and 12 Bg7 then forks c3 and h8. I'd since looked at this myself and concluded that Black's best response is to target the h1-rook in turn with 11...Ke7 (or 11...Ke8) 12 Bg7 Ne4+ 13 c3 Bc5, when 14 Bxh8 Nf2+ 15 Rg1 Ng4 16 Rh1 Nf2 leads to a draw by repetition. All well and good. Black is fine.

But how does a2-a3 help White in the reversed position? Answer: It doesn't. White has to take the draw in the same way. So when my opponent answered 5 Qh5 with 5...Nc6 6 Bb5 Qd6 7 Nf3 Nf4!, I stopped short and thought: “Bollocks!”. No further comment required.

Monday, 18 March 2019

046. Creating Problems

White: J. Shepley - C&DCCC Sinclair Trophy, 2019

As mentioned in Game 9: “I once wrote an article on 1 e4 e5 2 Bc4 f5!? (the Calabrese Counter-Gambit) for a special issue of Tim Harding's magazine Chess Mail (May 1997).” The same article also had brief analysis of a variation I dubbed ‘The Calabrese Counter-Gambit Deferred’, in that it goes 1 e4 e5 2 Nc3 Nc6 and if 3 Bc4 then 3...f5!?.

This can transpose to the Calabrese itself if, as usually occurs, White responds with 4 d3. But there some independent lines, and I gave a couple of paragraphs on 4 exf5 and 4 Bxg8. The latter of those I appear to have mangled:

“(b) 4 Bxg8 Rxg8 5 d3 (5 Qh5+ g6 6 Qxh7 Rg7 7 Qh3 fxe4 or 7 Qh8 Nd4) 5...d5 6 Qe2? (6 Qh5+ g6 7 fxg6 Rxg6 8 Qxh7 Qf6) 6...Bxf5 7 Nf3 Qd6 8 Bd2 0-0-0 Bixby-Curt, USA 1904 (via 2 Bc4 f5).”

Obviously 5 d3 d5 doesn't make any sense. It should have been 5 exf5 d5, but then 8...Qf6 (in the second bracket) drops the d5-pawn; and I can't find any reference to Bixby-Curt elsewhere, after either 5 d3 or 5 exf5. My original file in fact gives (respectively) 5 exf5 and 8...Bxf5 (Stockfish says 8...Qg5 is better) and doesn't mention “Bixby-Curt” at all. Okay, this was back in the day when we were – or I was – still typing stuff manually into word processor programs instead of entering them in ChessBase files. Mangling was an occupational hazard.

At least the bracketed 5 Qh5+ etc seems to be correct. My own file carried this further: 5...g6 6 Qxh7 Rg7 7 Qh8 Nd4 “-/+” 8 Kd1 (8 d3 f4) 8...Qg5 9 Qh3 (9 g3 fxe4 10 Nxe4 Qf5) 9...d5! 10 Qg3 (10 Nxd5 fxe4; 10 Nf3 Nxf3 11 Qxf3 fxe4) 10...Qh5+ 11 f3 fxe4 12 Nxd5 Rf7 “-+”. Stockfish doesn't have much of significance to add to that. And in the game below, played over 20 years later, my opponent made only four more moves before resigning.

Returning to the opening, I find it a little odd that 3...f5!? hasn't been seen more in chess praxis. MegaBase 2018 has just five games (dating back to I.Rabinovich-A.Flamberg, Triberg 1914), despite Black scoring 4/5. CCDatabase 2018 has another three, with Black scoring 2½/3. In total that's 6½/8 (81.25%) for Black. Even if we add the moves 4 d3 Nf6
5 Nf3, as recommended in Ovetchkin & Soloviov's book The Modern Vienna Game (Chess Stars 2015), and which can arise via various move orders, the figures are still in Black's favour: P51 W28 D5 L18 (59.8%).

I guess it's the professional mindset. After 1 e4 e5 2 Nc3 Nc6 3 Bc4 Black can play simply 3...Nf6 4 d3 Na5 and has no problems. Whereas for us non-professionals, “no problems” may be less appealing. Personally, I'm willing to weigh potential problems (for me) against practical problems (for my opponents). And if they solve theirs, then I have to solve mine. But if they don't...

Wednesday, 27 February 2019

045. AlphaZero nand AlphaZero

Black: AlphaZero - all-play-all tournament,, 2018

I've been reading Game Changer (New in Chess 2019), Matthew Sadler and Natasha Regan's book on AlphaZero: tech company Deep Mind's self-learning AI. AlphaZero was given the rules of chess, played 44 million games against itself in nine hours, and was then the strongest chess player in history, as demonstrated by its trouncing of the world champion computer program, Stockfish 8, in a 1000-game match: W155, L6.

AlphaZero is now my favourite player ever. Garry Kasparov loves it because he sees a mirror of his own style: dynamism as a priority, material secondary to the initiative. But AlphaZero is also like Karpov: subtle prophylaxis, restricting the opponent's counterplay, the opponent's pieces, keeping control. So yes, AlphaZero favours dynamism but only for itself. This is shown in a very simple way in a given graph plotting the number of legal moves each side has available at any point in the game. AlphaZero always seems to have more. It's Kasparov combined with Karpov, enhanced by a computer's incomparable tactical ability. Nigel Short called it God.

Matthew Sadler called it a lunatic. That was while going through a game where AlphaZero, as Black, sacrificed three pawns and its kingside structure for the rather nebulous gains of a well-placed knight, open files for its rooks, and an offside enemy bishop.

Its long-term initiative proved to be so great that even Stockfish was unable to contain it. Sadler's (and my) delight reflects the notion that technological advancement can sometimes seem like magic.

But then Sadler gets down to the business of trying to understand AlphaZero's chess, analysing how it plays, how it thinks, identifying its recurrent ideas, and making it comprehensible to a human readership. Regan does the same thing from the technical side, interviewing the Deep Mind team to gain insight into how AlphaZero actually functions, how it learns, and again how it thinks. And these two approaches “meet in the middle”. Game Changer is an extraordinary piece of work.

You'll notice that my opponent in the game below is AlphaZero. But not the AlphaZero. Just someone who has taken the name “AlphaZero” as their moniker. All the same, "human" AlphaZero is a strong and original player. We've had six games so far (four still ongoing) and I don't look like winning any of them. In the one below "it" adopts an unusual defence to the King's Gambit: 2...exf4 3 Nf3 Nc6 4 d4 f5. This is supposed to be bad, but winning a pawn and wrecking Black's structure was not enough to get me past opposite-coloured bishops, even with rooks on.

Well, there's no use in wondering what the real AlphaZero would have done because AlphaZero doesn't play the King's Gambit. Or 1 e4. AlphaZero plays 1 d4 and sometimes 1 Nf3. In other words, after due and deep consideration, AlphaZero has definitively pronounced: by test, 1 e4 not best.