"Robert Byrnes's 7 Nge2! is best" (Bobby Fischer)

Two alternatives to 8...Nd7 are 8...f5 attacking the g4 pawn and 8...Bd6 attacking the weskened black squares

9...Nh6 develops the N on the edge of the board. 9..Nb6 avoids this, although it costs time.

An alternative to 11...g5 is 11...f6 clearing the square f7 for the manoeuvre ...Nf7 after which ...Bd6 or ....Nd6 may follow

15 Nxe6 gains the bishop pair as Bobby Fischer might have done.

18...f5?! looks like an error as White can exchange his g4 pawn for the f5 pawn and isolate it. 18...Ng8 may be better even though it loses time.

If 18...f5?! is dubious, 19...Nxg4? looks like a serious mistake because it breaks Black's King side pawn structure. It is as if Aronian is giving Nakamura the fruits of a minority attack for nothing. This may be a losing mistake. The eventual result of 19...Nxg4 is that White wins the f and g pawns for his g4 pawn, and eventually the game.

An alternative to 19...Nxg4 is 19...Nf7 with ...Nd6 to follow and Black may have greater hope of saving the game

Karpov is only underrated by those who forget he was world champion for 10 years and defended his title against some incredible players while racking up tons of tournament wins and game victories. No one had held the world championship for ten consecutive years before him since Lasker! On endgame technique alone he beat another great endgame player, Korchnoi twice in World Ch. matches in several of their games.

So to say that Naka's endgame play was "Karpovian" is a high complement (although in this game I think it's undeserved).

I imagined this response after your similar take on the Keres Attack. I can't say I disagree. Good to think for yourself.

In order to better understand whole variation, let's step back few moves and look at 1963 Botvinnik-Petrosian WC match. They played QGA few times for draw. Botvinnik tried Tarrasch as black, also draw. On game 12, Petrosian tried Charousek (3...Be7) main idea was to avoid Carlsbad aka Exchange variation in QGD. Carlsbad variation allows black B on c8 to get free but in exchange (pun intended) white gets very active position for strong attack. This was also a favorite of Kaparov and one game showing its strength is Kasparov vs Ulf Andersson, 1988 this game also has 6.Qc2 called Reshevshky in USA, key modern move. Opening names often depend on country (e.g., in Russia 1.e4 c6 2.Bc4 is never called Hillbilly Attack LOL).

Charousek is also called Petrosian variation because he liked it so much as shown here. Strong line in Charousek is 5.Bf4 c6 6.e3 Bf5 often called Alatortsev but his name is also given to variations in Winawer French, Alekhine and Semi-Slav. 5.Bf4 is associated with Steinitz who used it in his later years, most notably Steinitz vs Lasker, 1896 one move 4 without exchange on d5, where 6.c5 (or 5.c5) is key move. Later on there were refinements involving Qc2 where transpositions could take place, good example is Kasparov vs Short, 1988 see discussion of transpostion there.

B on f5 in Alatortsev is strong for black, Bd3 to exchange Bs does not show any promising edge, neither did Ne2-Ng3 to harrass B. In WC game 12 (Botvinnik vs Petrosian, 1963) 7.g4! was tried, this move became main weapon it still is today. At first, 7.g4! looks like it merely compromises pawn structure but it is brilliant, modern move. One obvious reason is that black must give up b1-h7 diagonal, trying to hold onto it with 7...Bg6 gives white advantage through 8.h4! (e.g., 8...Bxh4 9.Qb3 b6 10.Nf3 Bb7 11.Bb8 and black position falls apart). Playing h6 merely compromise black K-side, making castling short too dangerous so that black K gets in jeopardy.

After 7.g4! Be6 (best), Botvinnik tried first of three moves 8.Bd3 to simply contol diagonal. While strong, game resulted in draw, so Botvinnik tried second move 8.h3 next time (Botvinnik vs Petrosian, 1963) resulting in win. For fun, here is computer eval after 7...Be6 showing three moves and their starting lines, using computer so early in game necessitates deeper computation.

Houdini_20_x64: 31/69 9:42:06 359,119,617,652
+0.10 8.h3 Nd7 9.Qc2 h5 10.gxh5 Rxh5 11.Be2 Rh8
+0.06 8.Bd3 Nd7 9.h3 g5 10.Bg3 h5 11.gxh5 Ngf6
+0.03 8.h4 Nd7 9.g5 h6 10.g6 f5 11.Qb3 Nb6
+0.02 8.Qb3 Qb6 9.f3 Nf6 10.Bg3 0-0 11.Nge2 Qxb3

Few years later Botvinnik tried 8.h4 sharper most aggressive move (Botvinnik vs Spassky, 1970) this attracted attention of Kasparov who used in game 21 of 1985 WC, being already ahead 2 points (Kasparov vs Karpov, 1985). Karpov came up with brilliant defense that avoided 10...Ngf6 in earlier Spassky game, where 11.f3 h5 12.Bd3 gave white solid advantage through greater space and f5 square (plan shifted to Q-side in this particular game so this advantage is not well illustrated by play). Moreover, B can retreat for Nf4 also strong plan. White K-side pawn weakness is illusory here, white can castle long or simply go Kf1-Kg2 for safe position.

Instead, after 9.h5 Karpov played 9...Nh6! strange looking but it helps stop white from strangling black position by putting pressure on g4 with idea of f5 to attack white pawn structure. Now 10.Be2 Nb6 and if 11.Nh3 then 11...g5! creates pressure ion white K-side, showing some drawback to 7.g4! as played in this Naka game (12.hxg6 hxg6 with equal double-edged position). Instead, Kasparov played 11.Rc1 in 1985 WC game and rather than passive 11...Bd6 Karpov could have played 11...Nc4 for more comfortable game, white gets some problem with K safety. In this game, Naka's natural 13.Bg3 was subtle mistake giving black initiative, computer shows 13.a4! as best move to keep some tiny edge.

And i include Fischer's Bxh2 move against Spassky in that!!