The Original

Fix the right notion of structure. An algebra for the operation X ↦ 1 + X is a set X with a chosen point z and a map s : X → X — a place to start and a way to step. A homomorphism between two such algebras is a function commuting with start and step. In this category ℕ is the initial object: for every algebra (X, z, s) there is exactly one homomorphism h : ℕ → X, namely the one fixed by h(0) = z and h(n+1) = s(h(n)). That formula is primitive recursion, and the theorem that it always defines a unique total function is Dedekind's recursion theorem. The single map out is the whole of ℕ's claim to be the source.

This is the original's signature, and it is one nobody can forge. To be the source is to map uniquely into every algebra; the property mentions all of them at once. An impostor receives its arrow from ℕ like everything else — but, as the next movement shows, it cannot send one back, and that asymmetry is the whole of the difference. The genuine article is not the model with the most structure or the longest theory. It is the one at the beginning.

Here is the subtlety that makes the original hard to name. The impostors are not failures of arithmetic; they obey every Peano axiom. In particular each impostor M satisfies "every element is 0 or a unique successor," which says exactly that the map [0, s] : 1 + M → M is a bijection. So every model of arithmetic, impostor included, is a fixed point of X ↦ 1 + X: they all satisfy 1 + M ≅ M. The successor operation has not one fixed point but a whole continuum of them. ℕ is distinguished among them by being the initial one — the least — and that is a property no single model can wear on its sleeve, because it is a statement about how the model sits relative to all the others.

the all-subsets quantifier, in disguise  Now the reason first-order logic cannot name the original is plain. Initiality is a property held over all algebras at once — quantify across every (X, z, s) and demand a unique map out. That is a statement about the whole surrounding category, and no first-order theory can see the category it lives in; it speaks only of its own elements, never of the totality of structures. The same content, phrased inside one model, is the demand that the standard part exhaust the model — and to say "the standard part" requires quantifying over the subset of standard elements, the very "all subsets" first-order induction can only approximate schematically. So The Impostors and The Original are one coin. The rivals are the non-initial fixed points; the original is the initial one; and the logic of Gödel and Tarski cannot tell them apart because it cannot pronounce the word initial.

If the obstacle is that first-order logic cannot quantify over all subsets, the remedy is a logic that can. Second-order induction states the principle in full: ∀X (0 ∈ X ∧ ∀n(n ∈ X → n+1 ∈ X) → ∀n n ∈ X), the quantifier ∀X ranging over every subset of the domain, not merely the definable ones. Feed it the set of standard elements — the image of the unique map from ℕ — and the axiom forces that set to be the whole model. The nonstandard tail cannot survive; the standard part exhausts M; and M ≅ ℕ. Said categorically: demanding that the standard part exhaust the model is demanding that the model be initial, and the all-subsets quantifier is precisely what lets you demand it. Second-order logic can name the source.

The rescue is genuine, and worth pausing on. Everything The Impostors built — the forced order type, the coded interior, the recursion-theoretic floor — was the anatomy of models that exist only because first-order induction is a schema over definable properties. Restore the quantifier to its full range and the entire menagerie is excluded in a single line. ℕ is not merely the nicest model; it is the only one. The Gödel sentence, true but unprovable, has a truth value because there is one structure for it to be true in, fixed exactly by the universal property the schema could only gesture at. For the length of this movement the ground is firm. The trouble is what the firmness costs — and it begins on the very next page.

The quantifier over all subsets is not free. A logic earns a usable proof system only if its theorems can be enumerated by a machine — and full second-order logic cannot. Its set of validities is not merely undecidable; it is not even arithmetical, lying beyond the reach of any first-order definition over ℕ. There is no sound, complete, effective calculus for it, not because none has been found but because none can exist. The logic that singles out the original is a logic in which "valid" is, in the strongest sense, beyond computation. You can use it to describe ℕ uniquely; you cannot use it to reason, because there is no walking on this ground.

The second cost is quieter and goes deeper. Second-order induction excludes the impostors only by quantifying over all subsets of the domain — and that presupposes that "all subsets of ℕ," the full powerset, is one determinate thing to quantify over. The categoricity is genuine: in any universe with a settled ℘(ℕ), second-order PA pins ℕ exactly, and nothing here is sleight of hand. But The Relativity showed that whether the powerset is determinate is just what set theory leaves open — CH's independence is the proof that ZFC does not fix even the structure of ℘(ℕ). So the determinacy second-order logic borrows to name the original is exactly the determinacy set theory cannot certify. The definiteness of ℕ was drawn on the definiteness of the powerset, and that draft must be honoured one level up — by a theory no more uniquely fixed than the arithmetic it was sent to ground.

Gather the whole. The Impostors asked whether ℕ could be told from its rivals, and the answer turns out to be yes — by a universal property, by initiality, by the all-subsets quantifier of second-order categoricity. The original is real: there is a mark the copies lack, and we have named it. The genuine natural numbers are the initial fixed point of the successor, the source from which every model receives its arrow, the unique structure second-order induction pins.

But the mark is exactly the power first-order logic surrenders to gain completeness and compactness, and that power is double-edged. Its validity decides the continuum hypothesis; its definiteness is borrowed from a powerset set theory cannot fix; and the moment you can reason in it again — Henkin's fork — the impostors return. So ℕ's uniqueness and ℕ's unreachability are not two facts but one: the very property that distinguishes the original is the one the logic of Gödel and Tarski cannot pronounce. The fixed point at the root of arithmetic is genuine, and out of reach, for one and the same reason. The original exists. The descent's only lesson is that you cannot point to it from where you are standing — and that to gain a vantage from which you could, you must climb to a logic that reopens the seam one universe up.