A zero-knowledge proof convinces you that something is true without revealing why it is true. The classical formalization demands three properties: completeness (a true statement has a convincing proof), soundness (a false statement does not), and zero-knowledge (the proof reveals nothing beyond the truth of the statement). The third property is defined by the existence of a simulator — a program that produces transcripts indistinguishable from real proof transcripts without knowing the secret. If such a simulator exists, then the verifier learned nothing, because everything the proof showed could have been generated without the secret.
For decades, an impossibility result constrained the design space. A proof system cannot simultaneously be non-interactive (single message, no back-and-forth), perfectly sound (no false proof exists), and zero-knowledge. You can have any two. You cannot have all three. This was treated as a hard wall.
Ilango (FOCS 2025) found a gap in the wall. Not a gap in the proof — the impossibility result is correct. A gap in the definition. The standard definition requires that a simulator provably exists. Ilango weakened this to: a simulator's existence cannot be disproven. In the space between “provably exists” and “cannot be disproven,” there is room — and that room is exactly the space opened by Gödel's incompleteness theorem.
The construction works by building a proof system where the question “does a simulator exist?” is independent of ZFC — the standard axiomatic foundation of mathematics. ZFC can neither prove the simulator exists nor prove it doesn't. This means no adversary operating within standard mathematics can demonstrate that the proof system leaks information, because demonstrating a leak would require proving the simulator doesn't exist, which ZFC cannot do.
The proof system achieves all three properties: non-interactive (one message), perfectly sound (no false proof exists), and “effectively” zero-knowledge (the failure of zero-knowledge is itself unprovable). The impossibility result said you can't have all three. You can, if you accept that the third property lives in the gap between true and provable.
The general principle: incompleteness is usually treated as a deficiency — there are things we cannot prove. Ilango treats it as a resource. The fact that some mathematical statements are true but unprovable means there exist properties that hold but that no adversary can leverage, because leveraging them would require proving something that the axioms cannot reach. The gap between what is true and what is provable is not empty. It is cryptographically useful.
This inverts the usual relationship between logic and security. Cryptography typically builds on computational hardness — problems that are easy to pose but hard to solve. Ilango's construction builds on logical hardness — statements that are true but impossible to establish. The security comes not from the difficulty of a computation but from the structure of mathematical truth itself.