The formal back matter
The volumes were written to be read; this is written to be checked. Here is the rigorous skeleton beneath them — the categorical setting in precise definitions, Lawvere's theorem proved in full, and, at the center, the one thing a unification of this scope owes its reader: for each claim that Cantor, Gödel, Turing, and the rest are "one construction," the exact bridge-assumptions it requires, named as propositions, with an honest verdict on how far the likeness reaches — genuine instance, transferred technique, shared engine, or mere suggestion. This is where the corpus states its own limits.
the setting · the theorem · the bridge-assumptions · the boundary
A primary text is read for its argument; its scholia — in the old scholarly sense, the technical annotations set in the margin — are there to be checked. These notes add nothing to the six volumes. They do something the volumes, written for the eye, could not afford to do at length: they state the definitions exactly, give the proof in full, and draw the precise line past which the corpus's central claim — that the famous limitative results are one construction — stops being a theorem and becomes, in descending order of strength, a transferred technique, a shared schema, and finally a mere family resemblance.
Why this matters is the corpus's own standard. A unification is worth exactly what it costs to state, and the cost is the bridge-assumptions: the propositions one must assume to carry a result out of its native setting and into the cartesian closed category where Lawvere's theorem speaks. Name them, and "Gödel is the diagonal" becomes a checkable reduction. Leave them unnamed, and it becomes the kind of slogan the diagonal itself warns against — a likeness mistaken for an identity. The inspector below names them, one claim at a time.
A cartesian closed category, stated precisely. The reach of the theorem is exactly the reach of these three definitions — no more.
The statement, the proof with the diagonal made explicit, and the two corollaries the corpus runs on.
If B carries a fixed-point-free endomorphism, then no point-surjective φ : A → Bᴬ exists. With B = 2 and α = ¬: Cantor's theorem, the missed element being {a : a ∉ φ(a)}. Every limitative result in the corpus is this corollary, with α and the setting chosen.
If φ is point-surjective — as the identity on a reflexive object D ≅ Dᴰ — then every α : D → D has a fixed point, uniformly: the fixed-point combinator Y. Caveat: no well-pointed topos (Set included) contains a reflexive object with more than one point. The power direction therefore requires a universe the limit direction does not — the deepest reason power and limit, though one construction, do not inhabit the same category.
For each claim that a famous result is the construction, the exact assumptions it requires — named as propositions — and an honest verdict.
The corpus says, again and again, that the famous results are one construction. Made precise, that claim has a cost, and the cost is nameable. To carry a result into the cartesian closed category where Lawvere's theorem applies, one must assume certain propositions — the bridge-assumptions. Some are definitional and free; some are deep theorems in their own right; and where none suffice, the claim of identity simply fails, and a weaker, honest relation takes its place. Select a result to read its bill.
What the inspector shows is a graded honesty. The limits of logic and computation — Cantor, Russell, Gödel, Tarski, halting — are genuine instances: each becomes the construction once a handful of proven bridge-assumptions (arithmetization, representability, the diagonal lemma, the s-m-n theorem) are named. The positive twin, recursion, is an instance too — though it demands a reflexive object that the world of sets denies it. Past that perimeter the likeness thins, and the corpus's integrity lies in marking exactly where.
A theorem across logic and computation; a technique just beyond; a shared schema among the genera; a suggestion in the human sciences.
The inspector sorts the corpus's claims into four ranks, and the ranks are the whole point. Across logic and computation, the unification is a theorem: each result becomes the construction once finitely many proven bridge-assumptions are named. One step out, at the complexity hierarchy theorems, only the technique survives — the diagonal as a method of separation, with no fixed-point theorem anywhere in it. Among the other genera, the order-theoretic and topological fixed-point theorems, there is a real family resemblance under a single schema — richness plus regularity yields a fixed point — but no reduction in either direction; the kinship is conceptual, not deductive. And in the human sciences the connection is a suggestion, sometimes productive and sometimes, as in the Lucas–Penrose argument, a fallacy wearing the costume of a proof.
A genuine instance of the construction, once finitely many proven bridge-assumptions are named. The result is Lawvere's, in another costume. Cantor · Russell · Gödel · Tarski · Halting · Y
The diagonal method travels, but the fixed-point theorem does not. Same move — yet no α, no φ, no fixed point. It separates; it does not fix. the complexity hierarchy theorems
A real family resemblance under one schema — richness + regularity ⟹ a fixed point — but no reduction either way. Conceptual kinship, not deduction. Knaster–Tarski · Brouwer · Banach
Genuine thematic resonance; no formalization makes them instances, and some are contested. A likeness to think with, never to claim. reflexive markets · the Löbian obstacle · Lucas–Penrose (rejected)
The Atlas's "three genera as one" is, stated exactly, a heuristic, not a metatheorem. The claim the corpus can honestly make is precise: one construction underlies the diagonal genus — the limits of logic and computation, and their positive twin, recursion — and a family resemblance, no stronger, links it to the order-theoretic and topological theorems. To assert more would be to commit, at the scale of a research program, the very error the diagonal warns against: mistaking a likeness for an identity. The discipline of the four ranks is the corpus's last result, and the one it is most bound to keep.
A primary text is read; its scholia are checked. These notes do not extend the argument — they fix its edges, naming what each claim rests on and where the likeness gives out. The construction is one across logic and computation, a technique just beyond, a schema among the genera, a suggestion in the world. That graded honesty is not a weakness of the unification; it is the whole of its strength. One construction — and the exact account of how far it reaches.