9...a5 and the Human Escape Route in the Era of Chess Engines

On a weekend evening in Binh Duong, I reopened my chess software on an old laptop and set up the position after White's ninth move. 9.b4. This is the Bayonet Attack — the sharp blade of the classical King's Indian Defense, the Mar del Plata Variation. White has just pushed a queenside pawn and sent a message: I will not let you rest. Beside a cup of tea gone cold, I played a promotional video for a DVD on the move 9...a5. The presenter, a player specializing in the King's Indian Defense, called this move the "antidote" for those tired of memorizing lines against computers. I sat there, listening to his argument, and wondered: does a pawn moving sideways truly open an escape route, or is it just a beautiful dream of freedom? The King's Indian Defense is one of the oldest and most complex defensive systems in chess. Starting with 1.d4 Nf6 2.c4 g6 3.Nc3 Bg7, Black deliberately concedes the center before counterattacking with ...d6 and ...e5. The Mar del Plata Variation — with 7...Nc6 — is the peak of this system, where both sides launch a two-wing race: White masses pieces on the queenside, Black storms the kingside. In the 1990s and 2000s, this variation appeared densely in top-level tournaments. Garry Kasparov, who once wielded the King's Indian Defense as a devastating weapon in title matches, helped enshrine Mar del Plata in every player's textbook. But as analysis engines grew stronger, players began memorizing lines dozens of moves deep. That is when the Bayonet Attack became the favorite weapon of those wanting to escape an opponent's careful preparation. In more than twenty years of following chess, I have noticed a pattern: whenever the machine tightens its net of calculation, humans seek to slip into dark corners the machine cannot see. The Bayonet Attack is exactly such a dark corner. The move 9...Nh5 has long been Black's mainstream answer. It opens the position wide, forcing both sides into sharp lines the computer has calculated to the very end. The problem is that when your opponent has prepared twenty moves at home with Stockfish, the Black player easily falls into a passive position right from the opening. Lines like 9...Nh5 become a one-way road leading to positions both sides already know by heart. The DVD I watched proposes a different path: 9...a5. Instead of opening the kingside, Black advances a queenside pawn, attacks White's b4-pawn, and prepares ...c5 or ...Be6. This is not a new move — it has appeared sporadically in top-level practice as a space-gaining option. But the author has systematized it into a complete repertoire, with a central argument: this is a position of strategy, not of tactics. The most appealing part of this product's argument is its pragmatism. The author does not promise Black an objective advantage. He admits the position is equal in evaluation. But he argues that in practical play, if Black understands the plans, the position becomes dangerous for White — because White must find their own way through a structure they rarely encounter. In other words, the value of 9...a5 lies not in the engine's evaluation bar, but in the psychology of the opponent sitting across the board. It is a gamble by someone who understands people. But this is where I must stop and ask a question. The claim that "computers do not know how to handle these positions" was once true in the era of Fritz and Rybka. But with today's Stockfish and Leela, strategic positions of the King's Indian type are evaluated quite accurately. The machine is no longer "blind" to long-term plans as people once believed. This product's anchor is slowly rotting with every engine update. Worse still, the DVD provides not a single number to prove its case. No engine match rate, no ACPL figure, no win-loss statistics from a database. It is all argument in words. That turns the claim "9...a5 is better than 9...Nh5" into an assertion, not evidence. A promise without an invoice. And there is a paradox of format. A static DVD cannot be updated after publication. Meanwhile, opening knowledge today flows through subscription platforms like Chessable, where every line can be corrected within weeks. This is the structural disadvantage of a physical product — a book frozen amid the endless flow of opening theory. Sixty days listening to recordings, I imagined the cheers of empty seats. Back then I learned that a chess position, like a match, is not measured by results alone. The pitch holds not only the ball, but also the calluses of human life. The chessboard is the same — it holds not only moves, but also players fumbling to find a path that does not require memorization. If 9...a5 helps a club player understand the King's Indian Defense more deeply instead of memorizing mechanically, then it has fulfilled its mission. Whether it is truly "better" is a question perhaps only time and databases can answer.