The unknotting number of a knot is the minimum number of crossing changes needed to turn it into an unknot — the simplest possible loop. For the (2,7) torus knot, the unknotting number is 3. For its mirror image, also 3. The natural assumption, open since Wendt in 1937 and explicitly asked by Gordon in 1977, was that connecting two knots should add their unknotting numbers. Two problems, combined, should be at least as hard as the sum of their parts.
Brittenham and Hermiller (2025) showed this is wrong. The connected sum of the (2,7) torus knot and its mirror can be unknotted in 5 moves, not 6. The combination is easier than the sum of the parts.
The mechanism matters. In the connected sum, crossings from the knot and its mirror create configurations that don't exist in either component alone — configurations where a single crossing change resolves structure in both. The mirror provides a topological shortcut through the original. Neither knot offers this shortcut in isolation. It emerges only from the combination.
In their January 2026 follow-up, Brittenham and Hermiller named such pairs symbionts — knots that reduce each other's unknotting difficulty when joined — and conjectured that every non-trivial knot has at least one symbiont. If true, the default assumption (combining problems compounds difficulty) is not just sometimes wrong but universally wrong in this domain.
The through-claim: composition can create shortcuts that neither component contains. The default expectation — that difficulty is additive — rests on an assumption of independence. But when components interact structurally (here, crossing configurations from a knot and its mirror), the interaction generates simplifying pathways. The whole becomes simpler than the sum of its parts, not because each part is simpler, but because their junction produces something new.
This inverts the usual story about composition. In most contexts, we treat combination as complexity multiplication: more components, more interactions, more difficulty. The symbiont finding says composition can also work subtractively — the right pairing introduces mutual cancellation that reduces the problem's effective size. The trick is that you have to find the right partner. Not every combination simplifies. But the conjecture says a simplifying partner always exists.