Where the Frameworks Break

When a company hires, not every criterion carries the same weight of ambiguity. GPA is framework-invariant — every evaluator reads 3.7 the same way. Years of experience, degree completion, certification status: these travel between perspectives without distortion. But “culture fit” doesn’t. Two interviewers assessing the same candidate for culture fit can reach opposite conclusions, each internally consistent, neither wrong by their own standards. “Leadership potential” does the same work. So does “communication skills,” assessed from a writing sample.

The framework-dependence isn’t evenly distributed across the hiring process. It concentrates at the boundary between objective and subjective criteria — the place where measurable credentials end and interpretive judgment begins. And this is where discrimination concentrates too: not in the GPA filter, which treats everyone identically, but in the subjective assessment, where the evaluator’s framework becomes constitutive of the outcome.

This pattern — framework-dependence concentrating at boundaries rather than distributing uniformly — is not specific to hiring. It appears across physics, computation, biology, and decision theory with a consistency that suggests something structural.

In gauge field theories, physical predictions must be independent of the mathematical description — this is what “gauge-invariant” means. And in the interior of a system, they are. The bulk physics doesn’t care which gauge you chose. But at the boundary, gauge-dependence reappears. In Wess-Zumino-Witten models, the bulk action is gauge-invariant while the boundary term is explicitly gauge-dependent. The framework’s fingerprint, successfully erased from the interior, persists at the edge.

This isn’t a technical detail that better mathematics would fix. It’s structural. The boundary is where the system meets its description, and that meeting is irreducibly framework-dependent. Different gauge choices — different but mathematically equivalent descriptions — produce different boundary physics. The bulk achieved framework-independence by, in a precise sense, pushing its framework-dependence to the boundary.

In decision theory, the value of information depends on where you stand. Deep inside a decision region — where the evidence strongly favors one option — additional information from different sources is complementary. Each new signal reinforces the conclusion. But at the decision boundary — where the evidence is balanced and the optimal choice could go either way — the same information sources become substitutes. They compete rather than cooperate. The economic structure of information itself changes at the boundary. Which analytical framework you use to combine evidence (Bayesian updating, minimax, satisficing) matters most at the point of decision, not in the interior of conviction.

In multipartite quantum systems, entanglement doesn’t distribute uniformly. It localizes at junctions — the nodes where subsystems connect. The bulk of each subsystem can be nearly unentangled while the junction sites carry almost all the quantum correlation. And the entanglement entropy of a low-energy system is bounded not by its volume but by its boundary area. What you can learn about a quantum system is determined by its surface, not its interior. Information is a boundary phenomenon.

This much establishes where. But the deeper claim is about robustness.

The non-Hermitian skin effect is a phenomenon where quantum transport concentrates at system boundaries — particles pile up at the edge. You might expect this to be fragile: a delicate quantum effect that any noise would destroy. The opposite happens. Under decoherence — the quantum-to-classical transition that destroys most quantum phenomena — the skin effect not only survives but strengthens. The drift velocity at the boundary exceeds what the coherent (noise-free) system achieves. Noise destroys the bulk’s quantum character while enhancing the boundary’s.

This is not an isolated finding. In topological systems, boundary physics is generically more robust than bulk physics. Flat bands at lattice junctions persist under perturbation while bulk bands shift. Edge modes in topological insulators survive disorder that scrambles the interior. The boundary isn’t just where framework-dependence concentrates. It’s where the system’s most robust features live.

The implication cuts against a deep assumption. We typically treat the interior as fundamental and the boundary as derived — the edge of something more important. But if the boundary is both where frameworks matter most and where physics is most robust, the priority might be inverted.

Consider how far the inversion goes.

Linearized gravity — the force that holds you to the earth, curves light around stars, and governs the large-scale structure of the universe — can be reformulated as edge modes of a five-dimensional topological theory. In the five-dimensional bulk, nothing happens. The theory is topological: no local degrees of freedom, no propagating particles, no dynamics. All the physics — everything that makes gravity gravity — lives on the four-dimensional boundary. The bulk provides the stage, but the play is entirely at the edge.

In computational complexity, the same concentration appears. The boundary between easy and hard optimization problems — where polynomial algorithms give way to exponential ones — is where computational resources concentrate. Problems deep in the easy phase are cheap. Problems deep in the hard phase are uniformly intractable. The interesting structure, the place where algorithmic ingenuity matters, is the boundary between them. Complexity, like framework-dependence, is a boundary phenomenon.

In string theory at the tensionless limit, the inversion becomes literal. The symplectic current — the mathematical object that defines what’s physical — localizes entirely on the boundary. The bulk phase space is degenerate: it has no independent physical content. The physical phase space exists only because of boundary conditions. The boundary isn’t a surface of the system. The boundary is the system.

There’s a sentence from the first essay in this series: “The boundary is the most information-dense part of the system.” That was an observation. This essay is the mechanism.

Framework-dependence concentrates at boundaries because the boundary is where a system meets its description. In the interior, different descriptions agree — they’ve had space to settle into consensus. At the edge, they haven’t. The boundary is where constraints from adjacent regimes collide, where symmetries break, where the choice of framework stops being academic and starts being constitutive.

The hiring committee knows this implicitly. That’s why they argue about culture fit and not about GPA. The argument is the framework-dependence, and it concentrates exactly where you’d expect: at the boundary between what can be measured and what must be interpreted.

What we call a phase — a uniform, well-understood region of behavior — may be the residue. The part that remains when you subtract the boundary. The bulk is what’s left over after the interesting physics has finished happening at the edge.

Look at the boundaries first. That’s where the frameworks break, and where the structure lives.