Miniature

IT is not often that one of the best players in the world loses in under 20 moves - still less that the game ends in checkmate.
But here is an example drawn from Tartakower and du Mont's 500 Master Games Of Chess (notes in italics are algebraicised from the book).
Johannes Zukertort - Adolf Anderssen, 1865*
Spanish Cozio
One of the many tragedies which, from time to time, are enacted around the ominous square f7.
1.e4 e5 2.Nf3 Nc6 3.Bb5 Nge7
The Cozio Defence, which variation embarrasses Black's game more than any other, and which has no longer any adherents (the book was published in 1952 by G Bell & Sons - I have the 1975 Dover reprint).
4.c3
The most energetic line of play is 4.Nc3 g6 5.d4 exd4 6.Nd5 Bg2 7.Bg5 h6 8.Bf6 etc.
Less good is 4.d4 at once, because  of  4...exd4 5.Nxd4 d5, and Black frees his game.
Most popular today is 4.0-0, which has been played by Karpov, Kramnik, Shirov and Leko. The mainline runs 4...g6 5.c3 Bg7 6.d4 exd4 7.cxd4 d5 8.exd5 Nxd5 9.Re1 Be6 10.Bg5 Qd6, with a position that has been reached more than 200 times in ChessBase's 2019 Mega database. White scores well with the continuation 11.Nbd2, but there are plenty of deviations possible for both sides long before this (for instance, I have a plus score with both 5...d6!? and 8...Qxd5!?, although neither move is held in high regard).
As for "the most energetic line of play," Alexey Dreev in Anti-Spanish: The Cozio Defence (Chess Stars, 2014) implicitly rejects 4...g6, and another quite popular reply 4...Ng6. Instead he concentrates on "a more modest and not so risky move," 4...d6.

Position after 4.Nc3 d6

Dreev states: "The position resembles in its structure the Old Steinitz Defence, but with an early sortie of White's knight to c3. In this variation, Black wishes to obtain a solid and stable position and would not avoid simplifications."
I faced 4...d6 at Tunbridge Wells last year. The round-three game Spanton (163) - Freddie Hand (208) continued 5.d4 ("This is the way for him to fight for a slight opening edge" - Dreev) 5...exd4 (Dreev recommends 5...a6) 6.Nxd4 Bd7 7.0-0, which Dreev says is "one of the positions of the Steinitz Defence in which White maintains a slight but stable positional edge" (1-0, 60 moves).
Not that I remembered when I began researching this article, but I actually played 4...d6 in a training game in 1992 against the Mephisto Polgar computer. That game continued 5.d4 Bd7!? (the most popular move in Mega19, but the engines Stockfish10 and Komodo9 reckon it is a mistake) 6.d5 Nb8 7.Bg5 (the engines prefer 7.Be3 or 7.Bxd7+!?) h6 (this seems a modest improvement over 7...f6, which was played in the tournament game Mephisto Polgar - Mark van der Kraan, Dieren 1991, won by the computer in 57 moves) 8.Bxd7+ Nxd7 9.Bh4 g5 10.Bg3, when White is surely better, but the game was drawn in 38 moves.
4...d6
Here 4...d5 5.Nxe5 dxe4 is indicated.
Tartakower and du Mont were writing long before Bobby Fischer played 6.Qe2 in a blitz game in 1971, which seems to be the improvement Dreev mentions as rehabilitating 4.c3 as "a serious weapon." The Russian grandmaster goes on to state that "all Black's three possibilities (4...d5, 4...g6, 4...a6) are about equally strong." But he says the latter two moves lead to "more stable, manoeuvring positions," while 4...d5 results in "complications full of tactical tricks." He does not mention Anderssen's choice, which nevertheless is still sometimes tried by strong players.
5.d4 Bd7 6.0-0 Ng6
In his obstructed position, Black cannot afford the time for such evolutions. Better is 6...g6.
7.Ng5 h6
A fatal weakening of the position. But after 7...Be7 8.Qh5, White still has the better of it.
I wonder what Tartakower/du Mont had in mind after 7...Be7 8.Qh5 exd4, which was played in Wolfgang Benischek (2070) - Dirk Nitsche (2235) in a German inter-club match in 2002. White immediately went wrong in that game with the wildly optimistic 9.cxd4?! Nxd4 10.Nxf7?? Instead of 9.cxd4?!, the engines give 9.Bxc6, with equal chances whichever way Black recaptures on c6.
8.Nxf7?!
This works swimmingly in the game, but a coming improvement by the engines suggests White should have settled for 8.Nf3 with the more pleasant-looking position.
8...Kxf7 9.Bc4+ Ke7??
This move, which was also the choice of a player rated 1905 in a Portuguese game in 2004, is not commented on by Tartakower/du Mont, but it seems to be the real reason for Black's loss. The engines give 9...Ke8, when 10.Qh5, which is so devastating in the game, can be met by 10...Qf6. After 11.f4 exd4 12.f5, Stockfish10 gives 12...Nce5!? (Komodo9's 12...dxc3 may also be enough for a slight edge) 13.fxg6 Bg4, when White has three sensible options, 14.Bf7+, 14.Rxf6 and 14.Qxg4, but each is assessed by both engines as better for Black. There are a lot of complications here, but the bottom line is that 9...Ke8 is better than the text.
10.Qh5 Qe8?
Falling into the abyss. He should at least have played 10...Be8.
Unfortunately, as the engines point out, 10...Be8 also leads to mate, viz 11.Bg5+! hxg5 12.Qxg5+ Kd7 13.Qf5+ 14.Ke7 Qe6#.
11.Qg5+
500 Master Games Of Chess has the less aesthetically pleasing 11.Bg5+.
11...hxg5 12.Bxg5#
*The game, one of a series of non-tournament, non-match games between Zukertort and Anderssen, was apparently played at the Prussian city of Breslau (now Wrocław in Poland).