A wrong theory gives wrong answers. You can fix it by finding the errors and correcting them. A silent theory gives no answers at all. You can't fix it, because there's nothing to correct — the theory simply doesn't have vocabulary for the question you're asking.
Henry Yuen, a complexity theorist at Columbia, is building a new complexity theory for problems with quantum inputs and quantum outputs. Traditional complexity theory classifies problems by the resources — time, memory, communication — needed to solve them, and its classification hierarchy (P, NP, BQP, and dozens of others) has shaped fifty years of computer science. But as Yuen explains, traditional complexity theory assumes classical inputs and outputs: ordinary numbers going in, ordinary numbers coming out. When the inputs and outputs are quantum states — superpositions, entanglement, interference — the traditional framework doesn't give wrong classifications. It gives no classifications at all. It is silent.
The critical open question is the unitary synthesis problem: could an infinitely powerful classical computational oracle efficiently perform any quantum state transformation? If the answer is no, then quantum-input, quantum-output problems are “of an intrinsically different nature than classical ones” — not harder versions of classical problems, but problems that live in a landscape classical complexity theory cannot map. The framework isn't inaccurate about quantum problems. It doesn't have a place to put them.
Yuen's team discovered that several seemingly unrelated quantum problems — transforming entangled states through local operations, quantum bit commitment in cryptography, decoding the Hawking radiation of a black hole, quantum information compression — are mathematically equivalent in complexity. They all reduce to Uhlmann's theorem. This equivalence wouldn't have been visible from within classical complexity theory, because classical complexity theory doesn't classify quantum-output problems at all. The connections between these problems are invisible not because they're subtle but because the lens is pointed at a different domain.
The general pattern: when a framework is extended to a new domain, the important failures are not the wrong answers but the absent ones. Wrong answers are correctable — they mean the framework reached the domain and made a mistake. Absent answers mean the framework never reached the domain in the first place. Newton's mechanics is not wrong about relativistic speeds — it gives increasingly inaccurate predictions that can be corrected with Lorentz factors. But Newtonian mechanics is genuinely silent about spacetime curvature. It has no concept to express it, no variable to represent it, no equation to get wrong. The silence is what motivated general relativity — not accumulated errors, but the recognition that an entire phenomenon had no representation in the existing theory.
The distinction matters because it determines the response. When a theory is wrong, you debug it — find the error term, add a correction, refine the model. When a theory is silent, debugging produces nothing. The correct response is to build a new framework that includes the silent domain, while ideally reducing to the old framework in the domain where the old framework already works. This is what Yuen is doing: constructing a complexity theory that can classify quantum-input, quantum-output problems while remaining consistent with classical complexity theory where inputs and outputs are classical. Not replacing the old theory. Extending the territory it can speak about.
The hardest part of recognizing silence is that it looks like adequacy. A framework that answers every question you've asked so far appears complete. The questions it can't answer are the ones you haven't thought to ask, because the framework's vocabulary shapes which questions feel natural. Classical complexity theory classified an enormous range of problems. The absence of quantum-output problems from the classification didn't look like a gap — it looked like a complete taxonomy. Yuen's contribution is not a new answer to an old question. It is the recognition that an entire category of questions was never asked.