Tuesday 27 June 2023

081. Nah

White: aperturaf - all-play-all tournament, ChessWorld.net, 2018

In his June 2023 ChessPublishing column on 1 e4 e5, GM Tony Kosten includes the game D.Naroditsky-M.Timmermans, Titled Tuesday (blitz) 2023, which “demonstrates a dangerous line that hinders Black from castling and culminates in the (...) position after 12 f4. This is tempting for White as Black can quickly find himself in trouble in this line if he doesn't know exactly what he is doing. Here Black played a good novelty, but went astray immediately afterwards and was crushed.”

GM Naroditsky also livestreamed the game, which has since appeared on youtube. At 3:13 he concludes, having won in 19 moves: “Yeah, this Bc4 line basically busts the line. I mean, I've played like this with Black, and it's just so unpleasant to play this. It's like, you can never castle. The engine always finds some ways to stay alive, but like from a practical point of view it's like, it's insanely hard to play. Yeah.”


The 7 Bc4 line is certainly challenging (see Game 75), but not the way he played it (7 Bc4 Na5?! 8 Qd3?! Be7 9 Nc3 etc). As I wrote in my book: “if 12 f4 then 12...Qc6!”.
This is rather more than staying alive. White has absolutely nothing here. More than that,
I prefer Black with the two bishops. And ...Qc6 is not at all hard to find. IM Timmermans would certainly find it with more time on the clock; i.e. in a situation other than being surprised in an online blitz game, and by a strong GM.

In fact I've won as Black after 12...Qc6. Well, to be strictly accurate, Stockfish says I was winning at move 15 (...Ng4), level at move 20, losing at move 28, winning at move 30 (...Bxc4+), losing at move 31 (Na5), level again at move 32, winning again at move 35, before I finally won at move 37. Such are the vicissitudes of 2350-rated 3-0 blitz on lichess.

The game below deviated at move 12. This is mentioned in my book too, up to 17...Bxg4, when “Black has whatever chances are going”. I see Stockfish 15 gives Black a clear advantage. Well, maybe. From the discussion mentioned in the previous post:

The point is not always what the assessment is as where it's going. Work on and is it trending up or down? Is a clear advantage trending to winning, staying where it is, or perhaps switching between a clear and slight advantage. Stockfish can defend a lot of positions it thinks are +0.8, +1, +1.3 for the opponent. Even more than that sometimes. In the end it comes down to, yes, the opponent is better, but can they actually win?

Or, here, can I actually win. As it happens, I did win. Sweet.

Incidentally, the following month Timmermans got to play the line again. This time he bashed out ...Qc6! (in an equivalent position) and was also winning at move 15. Though Stockfish is not too impressed with that game either. Such are the vicissitudes of IM level blitz on chess.com, I guess.

Friday 23 June 2023

080. Stockfish vs. Stockfish

Black: I. Rees - C&DCCC Ward-Higgs Trophy, 2023

Towards the end of 2022 I was minded to respond to an article in the New York Times, in which the author wrote: “Out of the 136 games played in the 32nd World Correspondence Chess Championship, 119 were draws.”

Looking at the crosstable you can see that that 119 is superficially correct. However, some qualification seemed warranted, so I tweeted:

Except that Bock withdrew, so all his "losses" can be discounted; similarly Neto-Pessoa where Black "resigned" a level position. So out of 119 real results, only *three* were decisive. Of these, two were one-move blunders (Edwards-Michálek and Schwetlick-Lecroq). Only Osipov-Schwetlick was a win "over the board". Basically Edwards came first because Michálek, who gave him a knight for nothing, only lost one game, whereas Schwetlick lost two. Well done.

Someone then asked: “Do the humans have any input or is it all down to computers?”
To which I replied:

There's a surprising amount of leeway in most positions, so humans can direct how they want play to go, while engines keep it within the bounds of soundness. But as a real contest between humans it's ultimately futile: if a position is at all sound, Stockfish will defend it.

We went into this a bit more in a thread on the ChessPublishing forum, which included a discussion of the strength (or otherwise) of a “centaur” (that is, a human using an engine) compared to that of an engine alone. My opinion was (and is):

That the centaur might be very slightly stronger relies on quite a lot of expertise: that the human knows how to use the engine, does so diligently, and is sensible enough not to overrule it in critical positions.

But note that I said "not significantly stronger" above. I meant that in a mathematical sense; i.e. there's no significant statistical difference; i.e. even in a best case scenario the centaur is hardly any stronger.

To put it in personal terms: I'm a CCSIM (title from 20 years ago), currently rated ca. 2200 over the board, and have been analysing with engines for many years. But if a CC game is, say, me + Stockfish vs. 1200 player + Stockfish – a prime example of centaur vs. engine – I know from the outset that, if the opening is sound and all my opponent's subsequent moves are approved by Stockfish, there's virtually no chance I'm going to win.

Hence, as I also said, and as I've indicated before: “I stopped playing serious CC in 2006. My only games are a couple each year for Nottinghamshire in the C&DCCC.” The game below is from this season's competition.

There is nothing interesting about this game. I hoped my opponent might play 3...f5. He didn't. He played the Berlin. I tried a line that Magnus had won with, and then – once Black had deviated – an odd-looking idea (14 Qc1) from an earlier correspondence game.
It didn't get me anywhere. The game was drawn. The companion game was slightly more entertaining but still a draw. My two games the previous season, against someone rated 1650 OTB (roughly 130 in old money) were also draws.

To reiterate: As a real contest between humans, correspondence chess is ultimately futile: if a position is at all sound, Stockfish will defend it.

Or to put it another way: Stockfish vs. Stockfish is a draw.

Saturday 31 December 2022

079. The Refutation of the King's Gambit

White: Dougbeer - Thematic tournament, ChessWorld.net, 2022

The title above is from an 11-part 2020/21 series (available for free download) by GM Matthias Wahls, who writes: “In my opinion, the King’s Gambit is clearly refuted by the Schallop Variation.”
Please note that “clearly refuted” doesn't necessarily mean a forced loss. From Wahls’ grandmasterly perspective, an opening is refuted for White if it leads to an inevitable disadvantage. And I am willing to concede that Wahls achieves his objective: White is theoretically worse in the Schallopp, so, on his terms, the King's Gambit can be regarded as refuted.

That's not enough to keep me from playing it. In recent games between the engines, either directly or by proxy, White manages mostly to draw after 3...Nf6. Which is to say, White's disadvantage is not so great that an engine can't defend it. And if an engine can defend against its 3600-rated colleague, I am reasonably hopeful of defending against mine at a lowly 2200 or less. Supposing they're sufficiently well prepared actually to achieve an advantage. Generally, they aren't. In fact, I've never lost to the Schallopp.

From the other side, I am more than reasonably hopeful of winning after 2 f4. To that end, the Wagenbach (2...exf4 4 Nf3 h5) has been very successful for me (over 80% as Black). Yet not as successful as the Schallopp. I've employed 3...Nf6 as an occasional weapon some fifteen times over the years – and won every single game.

This is the most recent example. My eighth move was a novelty. Previously Black has played 8...Ng7, transposing to the main line of Wahls' fourth article. But the engines have drawn against that, so I thought I'd try something else. Had my opponent elected to exchange queens on move 24, the result would probably have been the same. As it was, by the time the queens were due to come off anyway, White had a lost endgame.

Tuesday 27 September 2022

078. Cheating / Not Cheating

White: tripoduk - Chess.com, 2022

Prompted by the ongoing drama about high-level chess cheating / not cheating, evidenced / not evidenced by various algorithmic methods, notably at chess.com, I've been considering my own results again.

As I mentioned a few posts ago, chess.com have an analysis module that goes through your games afterwards, assessing how accurately you've played. My daily scores, always against players rated 700 points below me, are usually in the 90% bracket, which is high. However, if you get a good position out of the opening, are in control from then on, and are hardly ever threatened, you're not induced into making mistakes and are free to find nice moves. Hence your score will be very high. Or so you'd expect. The module does surprise me sometimes.

For instance, one game I won cleanly in 17 moves – all very satisfactory – and my score came back at 61.2%. Say what?! Okay, I could have played a standard Bxf7+ tactic and didn't. It's the sort of thing I watch for in OTB blitz games as I catch my mate out with it quite often. So I think it unlikely I missed it here. More likely I purposely refrained, hoping for the Legall's Mate trick on the following turn, which I indeed got to play, in effect twice (10 Qxe2). Still... 61.2%?

Well, anyway, the module has recently made amends by awarding me two 100% scores. Naturally, I see fewer problems with those ;). But let's examine them all the same.

The first came from a trap in a Dutch Defence. I only really made one move, an obvious one (12 b3) hitting a defending piece. To that point it was all previous praxis, and the four remaining moves followed straightforwardly.

The second was a trap in a Scotch Game. I'd had this before in an OTB league match, as far as 11...Qf6 – or rather as far as 12 Qf3, since the position arrived in reversed form after 1 Nc3 Nf6 2 e4 e5 3 a3 and so on. In that game I developed the queen's bishop before exchanging the other, which is less accurate than switching it round, though I still won in 24 moves, and with 24 b3 mate no less.

Back in the online game, Black was winning by move 12.
Whereupon the module praises 13 c4 for White and 13...Nxe5 for Black, calling the latter “brilliant” – even though its justification was a simple king and rook fork (14 fxe5 Qxe5+) which was only possible because White had just opened the e5-a1 diagonal. Not exactly brilliant. After that I brought my rooks to the centre files (...0-0-0 and ...Rhe8) and won a couple of moves later.

So what do I conclude from this? Nothing. And I don't expect anyone else to conclude anything either. Chess.com have in fact "concluded" something twice. The first time (2014) I was banned for "cheating". The second (2016) I received an automated “Warning about suspicious play”. On each occasion I replied with a detailed exposition of recent games – with reference to minutely-researched openings, numerous databases (OTB, CC, engine tournaments, and my own engine-enhanced analysis), an extensive library (close to 500 books), and suchlike – and that seemed to satisfy them (whether it's mean of me to utilize all my off-board resources in this fashion is another question). I've not heard anything since 2016, and the earlier ban was rescinded. They even gave me a free Diamond membership. Thanks :)

As for the current furore, I'm Team Hans, at least as regards the Sinquefield Cup. Magnus just played rubbish. Nonetheless, the reason I don't play classical chess online myself is...

The possibility of cheating.

Sunday 26 June 2022

077. Errata

White: verdi - all-play-all tournament, ChessWorld.net, 2022

No matter how careful you are, errors always slip through. In my book:

Typographical / input errors
So far I've found two incorrect diagrams, an erroneous reference, and a misquote.
– P207 (Two Knights): in the first diagram, the h-pawn should be on h3.
– P251 (Ulvestad): in the first diagram, there should be a pawn on f3. (I must have tweaked the diagrams in the font and then failed to proof them properly.)
– P285 (Jaenisch): line “D2” should be “E2”.
– P307 (Jaenisch): the final word of the first Swiercz quote should be “piece” not “pawn”.
(I guess I had the word “pawn” stuck in my head.)

Transpositional errors
It seems I didn't quite get all the multitudinous transpositions down.
– P95 (Wagenbach): 5 Qe2 d6 6 Nc3 c6 7 d4 g5 is indeed “another main line”, but one Black needs to avoid. Instead, 6...Nc6 is correct, when 7 d4 g5 8 e5 is line E33.
– P129 (Wagenbach): 7...Bg7 does not run “the risk of landing in an inferior 7 e5 line (E33) after 8 e5”, because 8...Nc6! then transposes to my main line E33 again.

Analytical errors
These remain to be seen or shown. However, I have – of necessity – found a small refinement in one variation:
– P240 (Two Knights): At the end of note e) with 20 Bxe6,
I've now discarded 28...Rb8 in favour of 28...g6. There's no urgency to put the rook on b8, especially as the tempo could cost Black in the event of a timely Rc8 by White, swapping the rooks off. Having been forced (by an opponent) to look at this position in more depth, the way to defend is as follows:

Put the queenside pawns on a3 and b4 with the rook in support. The problem then for White, with just one light square between the a-pawn and promotion, is how to remove both pawns. It will require all three pieces: king and rook to capture a protected b-pawn, with the bishop on the a2-g8 diagonal to halt the a-pawn. In response Black can aim to create another passed pawn with ...h5-h4, ...g5-g4 and ...h4-h3 (to overload the bishop), while the king heads for the centre to eliminate the d-pawn. No white pawns on the board means a draw. Another, not insignificant, factor is that, if the white d-pawn is removed and rooks exchanged, even a single black h-pawn can scupper any win, since ...h4-h3 and g2xh3 will leave White with the wrong bishop for the remaining rook's pawn.

This all played out satisfactorily in the game below. Note that the king goes to f8, rather than g7 (on move 31), to avoid ideas of Rc7 and Be6. And later (move 38), that taking on f7 doesn't get White anywhere: 38 Bxf7 g5 39 Bc4 g4 or 38 Rxf7+ Ke5 39 Bc4 Ra8 are both fine for Black.

Friday 15 April 2022

076. Testing the Theory 2

White: Capalaskine - all-play-all tournament, ChessWorld.net, 2018

This time it's not me doing the testing but (surprisingly) two 3600-rated engines – KomodoDragon and Stockfish – in the ongoing TCEC 22 Superfinal. The theory in question is (again) the 4 d3 Jaenisch, this time with 4...Nf6 5 exf5 Ne7 (see page 287 in my book).

Both engines (a little disappointingly) went for the same line as White: 6 0-0 c6 7 Ba4 (Stockfish actually played 6 Ba4 first) 7...d6 8 Nc3 Bxf5 9 d4 (usually considered to be good) and then 9...e4 10 Nh4 d5 11 f3 Qd7 12 fxe4 dxe4.
Hitherto my own praxis and analysis were my only source here, so I was very interested to see how the engines got on.

Both games (inadvertently) followed one of mine (see below) to move 16. The engines’ joint novelty (17 Nd1) didn't prove a significant improvement as they went on to make two draws. Whereas I (seemingly) turned a draw down at move 33. That's (likely) because I wanted to play it out, rather than thinking I was at all better. In the final position Black is about to regain a missing pawn, hence a draw all the same.

Well, it's nice to see 4...Nf6 holding up, at least for the moment. But, to quote myself: “It’s an exercise in brinkmanship for sure and I don’t always feel up to it. No worries – on such occasions I just play 4...fxe4!.”

Thursday 17 February 2022

075. Testing the Theory

White: Dlanor - all-play-all tournament, ChessWorld.net, 2021

Well, my book is finally out and you can go and buy it... wherever, supposing you haven't already done so. I'd suggest not wasting any time anyway because, as I write in the Introduction: “Stockfish 17 may refute the whole lot in five minutes”. And we're currently up to Stockfish 14.1.

Meanwhile, I'm still testing the theory in my own games, and to considerable depth. For example, in the main line of E22 in Chapter Ten (Jaenisch 4 d3), which follows analysis by GM Dariusz Swiercz (to 27 Bf4), I've since had the chance to test Stockfish's riposte 27...g5!
as far as 28 Be5 Qxc2 29 Qe3 Qc6 30 Rd1 Rfe8 31 Qd4 Re6 (given by me in a bracket), and supported my comment “that Black can, if necessary, await a convenient moment to take on d6 and reach a drawable endgame”. In the game below we liquidated further to rook + pawn vs. rook and agreed a draw.

GM Keith Arkell (or someone) would want me to prove it over the board, I'm sure. To which end, at the end, something like 46...Rg8 47 Kh3 (or 47 f4 Kc5 48 Rd1 Rg4) 47...Rh8+ 48 Kg3 Rg8+ 49 Kf2 Rh8 50 Ke3 (or 50 f4 Rh3) 50...Re8+ 51 Re4 Ra8, intending ...Kd6, should do the trick.

Which also supports my comment here in an earlier post: “If you want to defend the Jaenisch as Black, study the endgame!

Saturday 1 January 2022

074. Online Ratings

White: tripoduk - Chess.com, 2021

Online ratings are – how should I put it? – not very reliable. As evidence for that assertion I present my own. On ChessWorld.net I'm rated 2882 right now and have a highest ever rating of 2960. Eat your heart out, Magnus Carlsen. On lichess.org my blitz (3+0) rating is 2403, which also seems a bit high. I'm not an IM-level blitz player.

On chess.com, my blitz (3+0) rating is 2340. That cost me considerable pain and suffering (in the form of appalling and repeated blunders after playing for too long), so it's the only one I care anything about. My daily chess.com rating, on the other hand, is 2244, down from a highest of 2314. The steep decline there is the impetus for this post.

At blitz my opponents are... whoever turns up. You never have to wait more than 10-15 seconds. My opponents at daily are basically just two, whose ratings are currently around 1500. So whenever I win a game – which I nearly always do – my rating goes up perhaps a single point. Anything else causes it to drop considerably.

I've never lost at daily, but there have been a few draws, as in Game 18 and Game 29. Although the chess.com analysis module usually assesses my play in the 90th percentile – because I'm not challenged seriously enough to make many mistakes – that's not always sufficient to win. And the result of the game below actually made me laugh out loud.

Facing the very rare MacLeod's Opening, 1 e4 e5 2 c3!?, I went for a line suggested by GM Bologan: 2...d5 3 exd5 Qxd5 4 d4 Nc6 5 Nf3 e4!? (instead, 5...exd4 is a Göring Gambit Declined, as in Game 5) 6 Nfd2 Nf6, whereupon my opponent surprised me with a troublesome novelty: 7 f3!?.
This logical move removes Black's presence in the centre, leaving White's sturdy d4-pawn dominating. Black should still be okay, but... I don't much like it. I liked it even less after 7...Bf5?! (7...exf3 was necessary) 8 Bc4, when I was in danger of being worse. Escaping that fate via liquidation but then seeing no winning chances, at move 17 I offered a draw, which was accepted and — whomp! My rating dropped 76 points.

Ha ha ha. It would take me another year of winning every game to get back over 2300. Supposing I managed to do that. And supposing I was the tiniest bit bothered.