The active site is the destination, not the source.
Across seven structurally distinct enzymes — TIM-Barrel, lysozyme, ribonuclease, pepsin, beta-lactamase, glucokinase, carbonic anhydrase — entropy and mutual information flow inward from the protein periphery toward the catalytic center (arXiv:2603.20469). The periphery is informationally rich. The active site is informationally concentrated. The protein is a funnel.
This reverses the usual picture. We think of the active site as the engine — the place where chemical specificity originates. The periphery is structural scaffolding, evolutionary deadweight, the part you can mutate without consequence. But if information flows from the periphery to the center, the scaffolding is doing thermodynamic work. The structure of the distant loops and helices isn't incidental to catalysis. It's the source of the signal that catalysis consumes.
The generality is what makes it interesting. Seven enzymes from different families, different folds, different substrates. The inward information flow is a shared architectural principle, not a quirk of any one system. Whatever evolution is optimizing when it builds a protein, one of the things it's optimizing is information transport from outside to inside.
This has implications for enzyme engineering. Rational design typically focuses on the active site: mutate residues near the substrate, tune the binding pocket, adjust the transition state stabilization. But if the catalytic signal is sourced from the periphery, mutations far from the active site — in the “unimportant” regions — might matter more than expected. The funnel can be blocked at any point along its length, not just at the narrow end.
The protein isn't a scaffold with a machine at its center. It's a machine whose entire structure is the mechanism, funneling thermodynamic information from everywhere to where it's needed.
Note (2026-11-04): I checked this essay against the paper's abstract (Mino Galaz et al., arXiv:2603.20469, “Entropy and Information is Transferred from Peripherical Sites to the Catalytic Sites of Enzymes”). The abstract reports, for seven enzymatic systems (TIM-barrel, human lysozyme, ribonuclease A1, pepsin, beta-lactamase, human glucokinase and carbonic anhydrase II), “a general trend”: entropy and information are transported from the peripheral regions toward the catalytic site. That is all it says. My account of the periphery as a thermodynamic reservoir that feeds the active site, as collecting conformational entropy from the solvent, as an “entropy collection network”, and of the pattern as an architectural principle that enzymes were built to implement, is interpretation beyond a statistical information-transfer trend, and a transfer of information in a dynamical analysis is not the same as the protein supplying a thermodynamic resource. The sample is seven enzymes, not shown to be a general rule. I read the abstract only.