The Most Disgusting Chess Move I’ve Ever Seen in Chess

Every so often you see a move that doesn’t just surprise you, it seems to literally defy the rules of chess.
This game produced one such move.
The move you're about to see steps a pawn onto a square that looks “obviously” defended, leaves the queen hanging, and yet it’s not a bluff. It’s a declaration: I don’t need my queen to kill you.
This post is supposed to be educational, but I’ll be honest: it’s also an appreciation piece. Because if you’ve ever wondered what “initiative over material” looks like when taken to a borderline illegal extreme, this is your specimen.

---

The game

So we’re here, in the middle of an absolutely insane skirmish, played in 1973 in France. Our protagonists are Émile Joseph Diemer and Fritz Tromsdorff.
Diemer was a fiercely imaginative attacker, best known for championing the Blackmar–Diemer Gambit — an opening he promoted with almost missionary zeal. His play was bold, speculative, and often electrifying, reflecting a personality that could be just as intense away from the board. (At one point, a bizarre incident even led doctors to advise him to stop playing chess — a suggestion he very clearly ignored.)
Tromsdorff, by contrast, was a respected French master with a steadier, more classical style: disciplined, principled, and therefore the ideal foil for Diemer’s uncompromising aggression.
The opening is labelled as a Pirc, though it started out as a Jobava London — and only “transitions” once Black refuses to play ...d5. In this game White signaled their intent very early: g4, g5, h4, h5... a full pawn storm and out of book almost immediately. And while the opening has some interesting moments, it’s still nowhere near what is about to happen, so that's why we're starting off at move 29.
With the last move, Black took the bishop on d2, which, incredibly, White has no intention of recapturing.
29. O-O-O!?

Yep. Just ignoring the knight and castling instead.
The idea: White wants Nxg7 with tempo and doesn’t want to deal with Bh6+ interference if the king recaptures on d2 (since that check would chase the bishop off its crucial squares). To be honest, objectively, both choices — recapture or castle — are roughly similar, and the position remains completely insane but somehow about equal.
Engine logic in these positions is basically thinking: “Yes, this is on fire, but it’s a *balanced* fire.”
(29. Kxd2 Bh6+ Presumably what White didn’t like.)
29... Qf6
(29... Rxe6
Taking the annoying knight “in the throat” on e6 wasn’t really an option.
30. fxe6+ Kc7 31. Rxd2 Bh6 32. Nf5
And because of the g-passer, White is winning.)
30. Nxg7
This was the point of 29. O-O-O!?.
White grabs the bishop with tempo, hitting the rook on e8. And Black can never really recapture on g7: the f6+ ideas would expose Black's poor king placement. So the rook has to move — and Black does it with tempo by hitting the queen.
30... Re3

Now every queen retreat has drawbacks:
- Qh2 can run into ...Re2.
- Qg4 can meet the unpleasant ...Be2.
- Qh1 is playable, but visually exiles the queen, and sometimes allows ...Ne4–f2 forks.
- Qg2 (apparently best) reduces control of the h4–knight and, crucially, gives up second-rank influence — and you’ll soon see why that matters.
Also: in many “queen-promotion” races, Black can often answer by playing ...Ne7 at the right moment, giving up a knight for the new queen. So White can’t just abandon everything and hope the promotion wins the game cleanly.
All of that helps explain why White doesn’t retreat the queen at all, and instead tries to counterpunch by attacking Black’s queen.
31. Nh5?

It looks clever, but objectively it’s a mistake.
31... Qg5
32. Qg4?

A very interesting idea: offering a queen trade in an unconventional way.
If Black takes the queen, White has Nf6+ — a fork that wins back the queen — and that exchange would drastically favor White.
But White missed what might be the most spectacular move I’ve ever seen. That’s why the engine preferred: }
(32. Qg2
White likely played Qg4 partly to keep the h5–knight protected, but if you look closer, that “problem” isn’t really a problem: White can roll the g-pawn or just eliminate the d2–knight and cool the attack down.
The *real* difference is that Qg2 gives the queen at least some influence over the second rank. You’ll see soon why that’s crucial.
So it was in this position, after Qg4, that Black uncorked:)
32... b3!!

This is the move this whole post is dedicated to — and it’s absolutely brilliant.
Black pushes a pawn onto a square that appears defended, and in doing so **completely abandons the queen** for multiple moves — not as a swindle, but as an operating principle.
The attack is organized with a “half-dead” knight on d2, the other knight, the rook, and potentially the bishop. The queen is optional. The king is the target.
33. axb3
White’s practical response: take the pawn that just “suicided” itself. Seemingly logical. But let's look at a few alternatives:
(33. Qxg5
Accepting the queen sacrifice leads to forced mate.
33... bxa2
There is no defense against ...a1=Q, because:
34. Kxd2
fails to
34... Re2+
and the king is forced back onto the first rank, where the rook no longer guards a1 — leading to a clean lateral mate:
35. Kc1 a1=Q#)
(33. a3
Trying to “step away” also loses, and the brilliancy has barely started.
33... Rc3!!
Nothing short of spectacular. Black still ignores the hanging queen, and now sacrifices a rook on top of it.
White *must* take: ...Rxc2 is a mating threat, and capturing the d2–knight doesn’t solve it because ...Rxc2 still lands with devastating effect.
34. bxc3 dxc3
Now the only way to prevent a B2–type mate is to take on b3 — and that leads to one of the most beautiful mating constructions I’ve ever seen:
35. cxb3 Bd3!!
There is absolutely no stopping Nxb3 mate. White can throw in one last Nf6+ check, but Black just walks back and the mate remains unavoidable.)
(33. Nf6+
Throwing in the check changes nothing. Black can ignore the hanging knight and play:
33... Kc8
(33... Qxf6
is also possible and winning — but the point is: Black can often just slide the king back to the 8th rank and ignore these “desperation checks.” So Nf6+ doesn’t fix anything.))
(33. cxb3
Capturing with the other pawn looks like defense — it’s not.
33... Nxb3+ 34. axb3
(34. Kb1
doesn’t help and leads to forced mate:
34... Bd3+ 35. Rxd3 Re1+
Now Kc2 is met by the unusual mate ...Na1, while Rd1 runs into another gorgeous sacrifice:
36. Rd1 Qc1+! 37. Rxc1 Rxc1#)
34... Rc3+ 35. Kb1 Bd3+ 36. Ka2
(36. Rxd3
also fails:
36... Qc1+ 37. Ka2 Nb4+ 38. Ka3 Nc2+ 39. Ka4 Qa1+ 40. Kb5 Rc5+ 41. Kxc5 Qa5#
Again: absurd beauty. A mating net with White’s king dying on the fifth rank.)
36... Nb4+ 37. Ka3 Rc6
and now a combination like ...Nc2+ with ...Ra6, or ...Qe7+ with ...Ra6, ends it.)
33... Nxb3+?

This was played in the game, but it may actually complicate the win. The engine’s cold-blooded best is:
(33... Nb4!!
Another ice-cold move — and now you can see why 32. Qg2 was superior: if White’s queen had been on g2, it could have captured the d2–knight, and this whole mechanism wouldn’t work the same way.
Now:
34. Kxd2
leads to mate in 1:
(34. Rxd2 Re1+ with mate coming soon)
(34. Qxg5
Accepting the queen sacrifice again leads to a quick, stunning mate:
34... Na2+ 35. Kxd2
(the only move to take the knight)
35... Re2#
White, up 14 points of material, loses to a rook, knight, and bishop cooperating like assassins.)
34... Re2#)
34. Kb1

White has to retreat, because:
(34. cxb3
transposes: if White takes b3 with the c-pawn first and then accepts the knight sacrifice with the a-pawn, it’s just pawn captures in the reverse order — and we already know it ends in mate after
34... Rc3+)
34... Rd3??

I appreciate Black continuing the cascade of sacrifices, but the engine says this misses the win. Apparently the best practical route was to repeat:
(34... Nd2+ 35. Kc1
(35. Rxd2 Re1+ also fails)
35... Nb4
and we’re essentially back to the earlier mating ideas. The fact that b3 is gone doesn’t actually save White.)
35. Qg1??

Looks logical: reconnect queen and rook, try to stabilize.
But apparently this loses again — and without an engine, it’s ridiculously hard to navigate all the nuances in this rollercoaster. }
(35. Rf1
was apparently correct: stepping away so the rook isn’t hit by tactics.
35... Nd2+ 36. Kc1
Even allowing scary double-check ideas, Black has nothing better than repetition — because the queen on g4 is still guarded by the Nf6 fork trick (where we started).)
(35. cxd3
Worth noting: accepting the rook sacrifice here is pure masochism. The engine gives:
35... Bxd3+ 36. Ka2 Nb4+ 37. Ka3 Qd8 38. f6+ Kc6
White’s king likely won’t survive long, though it remains sharp and mistake-prone.
35... Nd2+

Correct.
36. Ka1 Qd8
With the queen coming to a5, this should be crushing. White was supposed to be done for... although apparently ...Nb4 was even quicker.
37. Nf6+
Trying to muddy the waters with a “grand scheme” check.
37... Kc8
Black is not interested in the knight. Black wants blood.
38. Rxd2

The knight must be eliminated, otherwise ...Qa5 would be mate. Now we get a mostly forced sequence that *should* be an easy win for Black — but after surviving in a hurricane this long, it’s very human to slip.
38... Qa5+ 39. Kb1 Nb4

This should win the queen with ...Qa2+ and ...Qa1 (assuming White defends against mate).
40. Rxd3 Qa2+ 41. Kc1

And now, for the love of God, I cannot comprehend why Black didn’t take the queen with ...Qa1+ here. I genuinely don’t understand what happened — maybe just a glitch — but instead Black takes the rook and lets the win evaporate.
41... Nxd3+??

(41... Qa1+ 42. Kd2 Qxg1
was obviously the way.)
42. Kd2 Qa5+ 43. Ke2?

This leads to a draw by force, and I completely understand why White chooses it. After surviving the storm and seeing Black refuse to take the queen, it’s human to take the perpetual and walk away.
Still, it’s worth noting:
(43. c3
would objectively be winning. The king can run from checks and the g7 passer becomes dangerously real — though there’d still be room for many mistakes.)
43... Ne5+
We’re officially back to equal.
44. Kf2 Qd2+ 45. Kg3 Qg5+ 46. Kf2 Qd2+ 47. Kg3 Qg5+

and the game ends in perpetual.
So yes: 32... b3 really is the most disgusting move I’ve ever seen — brilliant enough that it feels like it defies the laws of chess. A pawn walks onto a “defended” square, the queen is left en prise for multiple moves, and yet the attack continues with relentless force, all in service of one idea: the White king is exposed, and the second rank is everything.

---

Why 32...b3!! is “disgusting” (and what to learn from it)

A few big educational takeaways you can actually use:

1. Second-rank control is sometimes more valuable than material.
   A lot of White’s “natural” defensive instincts fail because Black’s heavy pieces and knights are poised to invade along the 2nd/1st rank with tempo.
2. A hanging queen isn’t always “a free queen.”
   If taking the queen opens a mating net (promotion mate, lateral mates, rook lifts, forced checks), then the queen is basically bait.
3. Coordination beats inventory.
   In multiple lines, Black’s queen is irrelevant; the kill is delivered by rook + knight(s) + bishop, in perfect cooperation.
4. In chaos, one tempo is a lifetime.
   Moves like ...Rc3!! and ...Bd3!! aren’t “pretty.” They’re clinical: every move forces, every move narrows the king.

---

Closing

I’ve seen a lot of tactical shots. I’ve seen a lot of queen sacrifices. But 32...b3!! has a special kind of menace: it’s not just a sacrifice — it’s a rejection of normal chess logic.
It says: “You can have my queen. You’re not getting your king.”
If you only remember one thing from this post, make it this: don’t evaluate tactics by material first. Evaluate them by king safety + forcing moves + coordination — because sometimes the “obviously losing” move is the one that makes the board collapse.