The reason is that following the moves ...g6 ( a move which displaces the second obstruction on the long diagonal) and ...Nc6 White has a potential threat of Nc3xb5 followed by Nd4xc6 and then Qc3 forking the black knight on c6 and the black rook on h8, a resource which GM Peter Leko indicated in the commentary.

I wouldn't call 17.Ncxb5 easy to miss, especially for a player of Carlsen's caliber. It's a two move finesse that ends in a fork. 16.Nc6 was a rather careless tactical mistake.

According to here (which apparently has every game Carlsen has ever played and some he has even thought about playing.) 7.Be3 was a first for him.

Yes (going by the D.B. here again) also a first for Esipenko but the speed in which it came out this had been in his lab sometime in the past.

His first real think came after Carlsen's odd looking 12...Qh4 for which the only reason I can see was to unsettle Esipenko, get him to the chessboard and out of his analysis which he had surely up his sleeve - you do not quickly sac pawns v the W.C. without back up. )

Then came the missed trick on c6. As White v an uncastled King you look for such shots. As Black it is easily missed. Witness the same pattern...

Melnikov - Kuznetsova, 13th Chigorin memorial, St Petersburg 2005

click for larger view

Black played 23...b3 (OOPS!) and resigned after 24.Qc3 hitting loose thing on c6 and h8.

This transpo from Be2 to ...e6 Scheveningen was a favorite of Kasparov. Could see Be3 played but in different move order. I'm not sure of all the subtleties in move order there but the early g4 by Esipenko stood out as the novelty for me. Think I remember Eljanov won a nice game with an early g4 not that long ago in a Scheveningen but think the usual is f4 to hinder ...e5 and 0-0 and prepare kside play.

Thought it was very tough play for Black. Maybe OK for Kasparov but even he had to play a lot of defense.

Karpov vs Kasparov, 1985

Hi SChesshevsky,

"... the early g4 by Esipenko stood out as the novelty for me. "

It has been played before D Olofsson vs I L Ljubicic, 2013 (a corres game, pawn sacs in these will be looked at very carefully.)

In the linked game.

click for larger view

Black played 11...Bf6 and it turned out to be a bit of a hair raising stramash.

---

Andrey called out the Magnus bluff with 13.Rg1

click for larger view

Offering the h-pawn and perhaps the Queens off going into a Queenless middle game two pawns down.

13...Qxh2 14.0-0-0 Qh6 (perhaps the idea behind Qh4) 15.Qxh6 gxh6.

Though the double pawn on an open file h-file will get hovered up (no rush there) and White active bits surely compensate. White has a whole host of loose pawns and squares to attack.

It looks like both players agreed White is very good in that line.

Full video analysis:

https://youtu.be/dUwO6x5ktZY

As the long-time esteemed contributor <Abdel Irada> noted long ago:

Here, White decides to speed things up a bit with 8. g4. He can because Black... well, Black lets him. We've seen Magnus push the b pawn to threaten White's knight, just to get out in front of White's kingside pawn push. I was surprised by the Bishop exchange since it eliminates an active piece for Black.

And, just when you think this will transpose into a normal if slightly less complicated Sicilian, we get 12. ... ♕h4? I'm sure White spent a lot of time trying to figure out if there was a method to the madness. But without the bishop on b7, this move is... strange, especially given the open g-file. White has the right response, he calls the bluff with 13. ♖g1, castles queenside and attacks.

The line beginning 6.Be2 is actually a positionally based variation, with the normal response 6....e5 7.Nb3 Be7 leading to White playing against the backward pawn, while trying to keep his opponent's queenside and central ambitions in check. The variation which starts with 6.Bg5 is razor-sharp, with theory stretching past move 20 in numerous lines, in which a single slip can be fatal.

<....Often White's bishop will be played to g5 and then retreated to e3 against the Sicilian, followed by the g pawn push....>

Yes, I like Bg5 vs. Scheveningen but it's often just asking for the poisoned pawn in Najdorf. Something probably don't want to get into without understanding and comfort.

Felt I understood it OK but never felt comfortable. Tried bringing the N back to block after seeing Karpov do it but dislike N's on b & c setups. So eventually mostly went with anti-Najdorf early Bb5+ Canal or g3, Bg2 lines.

Having looked more closely at the game, I completely agree that Magnus played very badly. 10... Bxg5 is a beginner's pawn grab, a move which those at club level would have rejected; but the real abomination was 12... Qh4 - beyond appalling: I find myself wondering if Magnus intended to lose to Esipenko!?