Leela and Stockfish play the Fried Liver Attack
Two Chess Players, Italy c. 1590, by Ludovico Carracci

Leela and Stockfish play the Fried Liver Attack

Avatar of ArnieChipmunk
| 7

In the ongoing marathon TCEC Superfinal best of 100 match between Leela Chess Zero (LCZero) and Stockfish, the two heavyweight engines are playing a variety of gambits and popular openings. In Games 59 and 60, they had a go at the Fegatello Attack, or Fried Liver, an ancient variation in the Two Knight's Defence already known from the first half of the 16th century.

In this line, White sacrifices a knight on f7 in order to 'fry' the black king (fegatello means 'fried liver' in Italian) in the centre of the board. It was supposedly first played in an actual game by the Italian Giulio Cesare Polerio (c. 1550-c. 1610) against his compatriot Domenico around 1602. Polerio was one of the most prolific authors of his time and wrote several important chess books, which were translated and read across Europe until the 18th century.

The Fegatello Attack is an important piece of evidence in establishing when castling first became common in Europe. The basic position of the attack arises after the following moves:

The first occurence of this position is in the so-called German Manuscript, dated ca.1530-1560. The manuscipt is kept in the Bibliothèque Nationale in Paris, France, and contains a note written by a librarian that says the manuscript is in German and written in the 16th century. We don't know the name of the author. The line continues, rather primitively, with the moves 7...Qf6? 8.Bxd5+ Ke8 9.Bxc6+ Qxc6. 

A much better defense for Black, however, is the brave 7...Ke6! 

(Note that if White is not allowed to castle in some way - bringing the rook into play - then ...Ke6 is simply the refutation of White's sacrifice. Hence we can concude some form of castling was already being exercised when chess players started looking into the Fregatello.)

The black king move, protecting the knight on d5, was first described in the so-called Boncompagno Manuscript No. 1, probably written around 1580-1590. The Dutch chess historian Peter Monté called it "the most curious and mysterious of all 16th century chess documents." The manuscript, headed Questo libro è di Giulio Cesare Polerio Lancianese al suo comando, e del Amici à presso, is a collection of games and analysis by various chess players from the 16th century, gathered over a period of years, perhaps decades, by Polerio. 

PolerioPolerio

The line with 7...Ke6 is attributed specifically to the Spaniard Don Carlos d'Avalos, of whom we know little more than that he lived in Naples, Italy around 1590. Polerio's analysis continues 8.Nc3, followed by 9.0-0. An important point, however, is that according to the manuscript, White castles in the "Napolitanean way", meaning the White rook jumps to e1 straight away! Obviously, this makes a huge difference since with a rook on e1, the line is virtually winning for White.  

When, not much later, 'regular' castling (also called alla Calabrista) with the rook to f1) became more and more common, the Fegatello Attack was found playable again for Black by Polerio and other Italian analysts. In a second, also undated but probably slightly later manuscript - the Leon Manuscript,  further analysis (likely also by Polerio) is given: 

The fact that it was seriously analysed in this and subsequent books/manuscripts, provides important clues as to when modern-day castling became the norm.

It took centuries before theoreticians agreed that the move 8...Nb4! was actually a better try for Black. For instance, the authorititaive Handbuch des Schachspiels (1874) by P.R. von Bilguer still preferred 8...Ne7. One of the reasons was that for a long time, 9.a3 seemed like a very strong move. This rook sacrifice was already given by Polerio and analyzed by De La Bourdonnais in 1833. It was also played only a few years ago by Alexei Shirov.

8...Nb4! was the last 'book' move given to the engines, and LCZero and Stockfish both chose 9.0-0! which is indeed the best move. Two interesting theoretical battles ensued. The second game featured the current main line, where Stockfish seems to have improved the theory with 21...Rc8!N

Both games ended in a draw. They're given here below, including variations provided by the engines.

It looks like after almost 500 years, the Fegatello Attack is still alive