There is something strange about trying to build a theory of consciousness. The theorist is a conscious system. The theory has to account for the consciousness of the theorist, including the theorist's capacity to formulate the theory. The theory has to describe the process by which the theory itself gets generated. You are a flashlight trying to illuminate itself. It is a self-referential loop, and that has consequences.
Gödel showed in 1931 that self-referential formal systems are necessarily incomplete. Consciousness theories are not formal systems in the strict sense, and the analogy must be handled carefully. But the structural parallel is informative. Kleiner and Hoel showed that any minimally informative theory of consciousness is automatically fragile under substitution arguments. The problem is not that the theories are bad. It is that self-description from within has inherent limits.
The persistent pattern of partial success in consciousness science is not just bad luck. When you are inside the system you are trying to describe, certain things will always slip through the net. We call this the constraint hypothesis: a self-referential limitation analogous to Gödelian incompleteness. Consciousness is not beyond science, and this is not mysterianism. But if the hard problem cannot be solved from the inside, then every claim must be testable from the outside. Every prediction in this book has a kill condition. Every failure is reported. The constraint hypothesis demands it.
The best counter-argument comes from DeLancey (2023). His claim is disarmingly simple: consciousness is not special, it is just complex. Experiences are mysterious only because they are too descriptively complex for our theories to compress into neat explanations. The explanatory gap is a gap in our description, not a gap in the world.
The argument is serious, and it holds on its own terms. He may be right. If he is, the hard problem dissolves into the merely very difficult, and the field can stop looking for exotic explanations and start building better descriptive tools. That is an attractive outcome, and the argument for it is stronger than most consciousness researchers want to admit. But it has a blind spot.
Imagine the shortest possible computer program that could produce a complete description of some experience, every detail, every nuance. Kolmogorov complexity is the length of that program. DeLancey's argument is that conscious experiences are incompressible. There is no shortcut, nothing to "middle-out", and no simpler description that captures them. That is why they feel irreducible. The mystery is just complexity.
The problem is that this measure is static. It works the same way forward and backward. Run the description in reverse and the complexity is the same. There is no arrow in it. It can tell you an experience is hard to describe. It cannot tell you why experience moves. The felt sense that time flows, that the present advances, that you are going somewhere: this is among the most fundamental features of consciousness, and complexity theory is silent about it.
The field has tried to fix this. Algorithmic Information Theory has evolved beyond static measures, adding computation time, meaning, dynamics, causality. Still, at every stage, it remains a deterministic, classical, discrete, and substrate-independent theory of Turing computation. If consciousness requires genuine indeterminacy and substrate-dependent dynamics, as Level B claims, then no version of AIT can capture the mechanism. DeLancey's framework explains why experiences are hard to capture in words. It does not explain why they unfold.