This is the fastest famous smothered mate in opening theory: six moves, one quiet queen move, and the most natural developing move on the board loses instantly. It has claimed real masters, most famously Edward Arlamowski against Paul Keres in 1950, and Alekhine used the same idea in a simultaneous display back in 1935. If you play the Caro-Kann as Black, this trap is mandatory knowledge.
In the 4...Nd7 Caro-Kann, White plays the sly 5.Qe2, hiding a pin on the e-file behind the e4-knight. If Black develops with the natural 5...Ngf6??, the knight jumps in with 6.Nd6#: the e7-pawn is pinned and cannot capture, and the king is smothered by its own pieces.
After 1.e4 c6 2.d4 d5 3.Nc3 dxe4 4.Nxe4 Nd7, Black plays a main-line system: the knight on d7 prepares ...Ngf6 so that Black can recapture on f6 with a piece instead of wrecking the pawns. Now 5.Qe2!? looks clumsy, even bad, because it blocks the f1-bishop. That is the disguise. The real point is that the queen stands behind the e4-knight on the e-file: the moment that knight moves, the file opens against Black's king. 5...Ngf6?? attacks the knight and develops a piece, the most natural move in the position, and loses on the spot: 6.Nd6#!. The pawn cannot capture, because exd6 would expose the e8-king to the queen on e2, an illegal move. The king cannot run: d8 holds the queen, and d7, e7, f7 and f8 are all occupied by Black's own pieces. One knight delivers mate in the middle of the board on move six.
5… Ngf6 — 5...Ngf6?? (ply 10) is the losing move, and it is the cruelest kind: the developing move Black was planning all along. The right version of the same idea is 5...Ndf6, taking with the d7-knight, because it clears d7 so the queen on d8 covers the d6 entry square: then 6.Nd6+?? simply loses the knight to 6...Qxd6, and after the normal 6.Nxf6+ Nxf6 Black is completely fine. The modest 5...e6 also defends, since it frees the f8-bishop to answer Nd6+ with ...Bxd6.
When an enemy knight sits on e4 with the queen lined up behind it on e2, treat the e-file as loaded: any knight jump comes with a discovered pin against your e7-pawn. Count again who actually controls the entry square d6. In this exact position, the difference between 5...Ngf6?? and 5...Ndf6! is everything: the second one keeps the d-file open so the queen guards d6. As a habit, before each developing move in the opening, spend two seconds on every check your opponent has: this whole trap is a single knight check that turns out to be mate.
Because the pawn is pinned. Once the knight leaves e4, White's queen on e2 looks straight down the open e-file at the king on e8. Capturing with the e7-pawn would expose the king to check, which is illegal, so the mating knight on d6 is untouchable.
Yes. The most famous victim was Edward Arlamowski, mated in six moves by Paul Keres at Szczawno-Zdroj 1950. Alekhine had sprung the same idea in a 1935 simultaneous exhibition, and it still appears constantly in online blitz at every rating level.
A trap only works if your opponent makes the mistake — strong players sidestep these, which is why each page also shows how to avoid it. Every line here is checked legal with the same engine the board runs, and every checkmate is verified.