Near Perfect?

Scoring 99.6% accuracy on Chess.com's computer-generated assessment might be considered an achievement, even if the game was 100% preparation. However, the game in question featured several risky and dubious moves--a line that I play in blitz and other casual games, but would never attempt in a serious game.

Note in the image from the website's game report that Black had a slight edge early on, even though it was short-lived. During this phase of the game, both players were making so-called book moves. Not all book moves, however, have equal merit. It does seem that the methods of evaluation built into the website's analysis feature does not count dubious or refutable book lines against a player's accuracy score.

Stripes,J. (1838) -- Internet Opponent (1805) [C44]
Live Chess Chess.com, 16.01.2020

1.e4 e5 2.Nf3 Nc6 3.Bc4 Nf6 4.c3?! Bc5

4...Nxe4 5.Qe2 d5 is also "book" but White scores poorly.

5.d4

This move is fun to play in blitz and often enough leads to games like the present one, but I doubt I would play it in a serious game.

5.d3 is better, and not surprisingly was played this week in Wijk aan Zee among Grandmasters at the Tata Steel Chess Tournament.

5...exd4 6.cxd4 Bb4+

White to move

7.Nc3

7.Bd2 Bxd2+ 8.Nbxd2 d5 9.exd5 Nxd5 10.Qb3 Na5 11.Qa4+ Nc6 seems about equal (11...c6 12.Bxd5 Qxd5 13.0-0)

7...Nxe4 8.0-0 Nxc3

Now White has an advantage

8...Bxc3! 9.d5 Bf6 10.Re1 (10.dxc6 bxc6 Black is clearly better) 10...Ne7 11.Rxe4 d6 12.Bg5 Bxg5 13.Nxg5 and White is worse.

9.bxc3 Bxc3

9...d5 10.cxb4 dxc4 11.Re1+ Ne7 White has a lead in development, and perhaps Black's advanced pawns are vulnerable. 12.Bg5

10.Qb3

10.Ba3 is preferred by Stockfish 10 on my computer, but Chess.com's version likes my move better. 10...d5 (10...d6) 11.Bb5

Black to move

10...Bxa1

Now White is winning. I like showing this position from Greco to my students. Black is ahead a rook and two pawns when you count all the material on the board, but all of White's pieces are in the game. Most of Black's forces are locked up. They are spectators, rather than players.

10...d5 is Black's last chance for equality 11.Bxd5 0-0 12.Bxf7+ Rxf7 (12...Kh8 13.Ba3 Bxa1 14.Bxf8 Nxd4 15.Nxd4 Bxd4) 13.Ng5 Be6 14.Qxc3

11.Bxf7+ Kf8

11...Ke7 12.Bg5+ Kf8 (12...Kd6 13.Qa3+ Nb4 14.Qxb4+ Kc6 15.Qc5#) 13.Bxd8

12.Bg5 Ne7

12...Nxd4 tests White a little more, but I won a long game from this position last month. White had an easy game.

13.Ne5

I've had this position in at least two other blitz games. It appears in several of Greco's model games and is one I've used often for instructive purposes.

13...d6

White to move

This was a new move to me, but I find that it appears six times in my copy of the ChessBase database.

13...Bxd4 14.Bg6 d5 15.Qf3+ Bf5 16.Bxf5 Bxe5 17.Be6+ Bf6 18.Bxf6 Ke8 19.Bxg7 is in Greco.

14.Bg6

Is a response suggested in Greco's models, despite the absence of Black's d6 move there. The checkmate threat is simple enough.

14...d5 15.Qf3+ Bf5 16.Bxf5

Black to move

16...g6

16...Bxd4 17.Bg6+ Nf5 (17...Kg8 18.Qf7#) 18.Qxf5+

17.Bh6+

Another move that exists in Greco, but with a small difference in the placement of one or two other pieces.

17...Ke8

17...Kg8 18.Be6#

18.Bxg6+ Nxg6 19.Qf7# 1-0

Black's errors offered an opportunity to reproduce one Greco's instructive model games with a new twist. Such minor alterations to the basic idea were his method as near as I can discern from the several books that have been made from his manuscripts.

The evaluation of my performance by Chess.com's software highlights that such resources should be treated with skepticism and using sparingly.