Dehn surgery replaces a tubular neighborhood of a knot in the three-sphere with a solid torus, glued along the boundary. Different gluings produce different three-manifolds. Two knots are distinguished by their surgeries if no surgery on one produces the same three-manifold as any surgery on the other. This is a powerful invariant — but not a complete one. Some distinct knots produce identical manifolds under the same surgery.
The knot trace is a four-dimensional object. Instead of cutting out and regluing in three dimensions, you attach a four-dimensional handle to the four-ball along the knot. The result is a compact four-manifold with boundary equal to the surgery manifold. The trace contains strictly more information than the surgery: it remembers how the handle was attached, not just what the boundary looks like.
Baldwin and Sivek (arXiv:2501.00914) prove that the 0-trace — the trace associated with 0-framed surgery — detects every L-space knot. L-space knots are a distinguished class whose surgeries produce three-manifolds with the simplest possible Heegaard Floer homology. The detection result means: if two L-space knots have diffeomorphic 0-traces, they are the same knot. The 0-trace is a complete invariant for L-space knots.
The surprise is that 0-surgery itself does not detect L-space knots. The three-dimensional boundary — the surgery manifold — fails to distinguish them. The four-dimensional filling — the trace — succeeds. The extra dimension carries exactly the information that the boundary loses.
The structural lesson: a boundary can be identical while the interior differs. Two knots produce the same three-manifold when you cut and reglue, but different four-manifolds when you fill in. The dimension you can't see from the boundary is the dimension that distinguishes. Knowing the surface of something is not the same as knowing the thing.