The Worst First Moves in Chess

I was inspired by the notes in one of my clubs to think and write about the worst possible first moves. I’ll consider the worst first move candidates for White, and for Black in response to 1.e4 and 1.d4, and ultimately compose a personal top three.

The worst moves are actually easy to find in terms of raw engine evaluation, and on the other website you can challenge WorstFish, which is conveniently programmed to play the very worst moves possible based on Stockfish 14 (the current version is 17.1). I’ll also try to put into words the reasoning behind my list.

My knee-jerk reaction was that 1.f3 had to be the very worst move for White. It doesn’t help control the centre, it takes away the f3-square for the Knight, and it weakens King safety. There are no redeeming qualities, apart from maybe reaching setups in which f3 is useful, such as if Black plays a King’s Indian Defence and you have time to put your pawns on e4, d4, and c4, leading to a Sämisch. If Black just puts their pawns in the centre, f3 looks especially bad, and the absence of the Knight on f3 can be felt the most.

Another logical candidate is 1.g4. White doesn’t put a pawn in the centre while offering Black a clear target to attack. It also compromises kingside safety. Among its fans, there are actually two ways to play it: some like to follow up with h3, others like to push on with g5. Anyway, I was quite surprised to see just how badly the engine evaluates this move. The main line goes like this, with White actually going for neither h3 nor g5, when Black has superior development, control over the centre, and will inevitably gain more time against the White Queen:

Moves like 1.a4 and 1.h4, which beginners are so fond of playing, are not actually as bad as they seem. 1.a4 is probably quite significantly better than 1.h4, though, because White is much more likely to want to play 0-0 instead of 0-0-0. White therefore severely limits their options by playing 1.h4. Also, h4 is generally only strong against a castled Black King, while a4 can take valuable queenside space (more on that later).

Moves like 1.Na3 and 1.Nh3 don’t actually weaken any pawn structure yet and do develop pieces, even if it’s to suboptimal squares. The engine evaluation of these moves, interestingly enough, is quite a lot worse than all of 1.a3/a4 and 1.h3/h4, so this is where my human brain fails.

Moves like 1.a3 and 1.h3 very often come in handy and are probably the best of the bad moves. Take this position, in which a3 serves a clear purpose by preventing Bb4 and making the d4 push more appealing:

Moves like 1.c3, 1.d3, and 1.e3 are passive but not actively harmful to White’s position. Transpositions are also still easily possible. If you follow up 1.c3 with d4, you can easily transpose into a Colle. 1.d3 can conveniently be continued with Nf3 and g3, leading to a King’s Indian Attack. 1.e3 works well with a quick c4.

Both 1.b4 and 1.f4 are not great, but they are far from candidates for being the worst move. Black can equalize easily against these setups and maybe even claim a slight edge, but it’s nothing catastrophic.

1.b4 doesn’t weaken the kingside, prepares to develop the Bishop to b2, and takes some queenside space. In the main line, Black goes e5 and Bxb4, while White goes Bb2 and Bxe5. White has essentially gotten a centre pawn for a flank pawn. It’s not as great as it looks, because Black can win time against the Bishop on e5 and make use of the e-file later, but it’s not entirely nothing.

1.f4 controls e5 and, with that, a part of the centre, and practically speaking, players who like the Dutch Defence as Black can feel at home in these positions.

As for advice on how to meet these weird moves, consider how they do or don’t fit in with your usual setup. 1.a4, for example, is usually a good move in Anti-Sicilian systems, taking space and preventing a6–b5 ideas from Black, so 1...c5 would probably not be the best answer. On the other hand, in 1.e4 e5 systems, with the exception of very deep Ruy Lopez lines, the move a4 is almost always bad, weakening the b4-square. With that in mind, the best answer would probably be 1...e5, making d4 an unappealing response for White and thus limiting their options while taking a firm stake in the centre.

My personal top three worst moves for White would thus be, to my own slight surprise:

1.g4
1.f3
1.h4

As for Black’s worst moves against 1.e4 from White, I was surprised that the aforementioned WorstFish selected 1...b5 as the worst move instead of 1...f5, the clear candidate for the top spot. I’m inclined to think that, especially for a human, it would be much worse to lose the f-pawn than the b-pawn. 1...b5 is basically a reversed 1.b4, with a clear tempo down, meaning you can’t win the e-pawn back with the Bishop.

I initially thought it was going to be quite hard to find number three for this list, with suitable candidates to my human brain being 1...f6, 1...g5, and 1...h5, but as it turned out, according to my Stockfish 17.1, 1...g5 was actually even worse than 1...b5 and 1...f5, despite not immediately losing a pawn. The main line goes like this in the diagram below: after 3...g4, the best move is not actually 4.Qxg4 (4...d5), but 4.Nc3, and after 4...d6, material is still at least even. However, I think that being a pawn down for basically nothing is much worse for a human to deal with than the final position in my 1...g5 diagram, where the extremely high engine evaluation in White’s favour stems from other aspects, such as control over the centre, a lead in development, and a clear target on the kingside.

My personal top three worst moves for Black against White’s 1.e4 would thus be:

1...f5
1...b5
1...g5

In response to 1.d4 from White, WorstFish goes for 1...g5, which makes sense, since it loses a pawn and weakens the kingside. My other candidates were 1...e5 (sorry, Englund fans, but losing a centre pawn can’t be great, no matter what people on YouTube might say), 1...f6, and 1...h5.

1...e5, the Englund Gambit, involves losing a centre pawn in the hope of creating an open, tactical game. It’s generally considered a weak gambit, heavily relying on surprise value.

1...f6 has basically all the downsides of 1.f3 for White discussed earlier, but with the additional downside of being a tempo down.

Even though the engine considers 1...h5 to be less bad than 1...e5, I’ll still go for 1...h5 as the move to round out my personal top three. I feel that in games between humans, the practical value and traps of the Englund Gambit far outweigh the objectively less bad move 1...h5, which doesn’t cost a pawn.

My personal top three worst moves for Black against White’s 1.d4 would thus be:

1...g5
1...f6
1...h5

Guess I’ll be sticking to my London and Caro-Kann for now 😊