The Work
W2 — The Structure of Logic
What constraint is — why bets are unavoidable, why checking is easier than finding, and the architecture of the Logic Corpus.
"Logic," in Work 2, means something more fundamental than formal logic. Formal logic — propositional calculus, inference rules — is one product of what Work 2 studies, not its subject. The subject is constraint itself: why anything follows from anything, why some knowledge is forced and some left open, why narrowing happens without guiding. Work 1 works within constraint; Work 2 makes constraint the thing to be explained.
It is the second half of the structural pair. It takes the medium Work 1 operates in — the constraint that shapes every distinction — and asks what it is, such that it produces the specific features Work 1 identifies: feedback, self-capture limits, the internal logic of the constrainability gradient. This is the framework's most technical territory, and parts of it are among its least settled. What follows states the established direction and the core claims, and marks the more speculative ones honestly as it goes.
The ground it works from
Work 2 inherits the foundation Work 1 establishes — the minimal ground, the fundamental activity of differentiation and abstraction, the constrainability gradient — and turns from using it to examining it. Work 1 argues that things follow from other things; Work 2 asks what "following from," "entailing," and "constraining" actually are. The self-referential character is deliberate, not a flaw: the structure of logic examining itself, and finding that the examination exhibits the structure it describes.
The core claims
Four claims form the argumentative spine. These carry the most weight.
Combinatorial explosion is the structural origin of constraint. Each differentiation doubles the space of possible configurations; the space grows exponentially while any actual path through it stays linear. Constraint is the gap between the space and the path. In a combinatorially vast space exhaustive search is impossible, so any navigator must commit — impose fixed bottoms, treat some things as settled, and search only the narrowed space those commitments allow. The necessity of betting is not a human weakness but a structural consequence of drawing distinctions in time.
Checking is structurally easier than finding. Verifying a configuration against constraints traces downward through a fixed hierarchy — the variation does the work. Finding one requires building upward through bets — imposing structure the variation does not provide. Recognizing a face versus drawing it, comprehending a sentence versus composing it, judging a proof versus discovering it: not loose analogies but the same asymmetry wherever constraint is navigated. It is also the shape of the famous P-versus-NP question, which Work 2 engages directly.
The constrainability gradient has an internal logic. The gradient is not a brute fact but the product of identifiable, interacting factors — the density of the variation, the width of the epistemic gap, the transparency of the underlying structure, the richness of the surrounding web, what the consciousness brings, and whether other consciousnesses are involved. Work 2 claims to explain why physics sits where it does and psychology where it does.
Heuristics are systems of fixed bottoms that make vast spaces navigable. A heuristic's quality is how well its fixed bottoms match the actual constraint structure of what it meets — which is why heuristics work spectacularly on structured instances and fail on worst-case ones.
Claims with formal contact
A second set of claims engages established mathematics directly. These are agreed in direction and under active development.
The lens-space picture: an algorithm is implicitly a choice of decomposition lens, and hard problems resist compression across every lens efficiently reachable — if P≠NP, NP-complete functions look irreducibly spread in all of them. The self-capture pattern extended to efficiency: Gödel, Tarski, and the halting problem show that formal systems cannot fully capture their own operation; P≠NP would show that formal computation cannot efficiently capture its own generative side — the framework's strongest structural prediction, and its most carefully qualified. And a unified account of AI approaches: successive methods — classical machine learning, transformers, graph networks, diffusion models, reinforcement learning, quantum computing — read as regions of one lens space, each a quantitative expansion against the same qualitative wall.
The boldest claim
The most speculative territory, and the one held most openly: navigation may be structurally distinct from computation. The brain may not be computing in the formal sense but navigating — no clean solver-problem separation, many timescales at once, contradiction tolerated as tension rather than treated as fatal. The strong version, that navigation is irreducible to computation, is speculative; the weak version, that navigation is computation with exceptionally well-matched fixed bottoms, is defensible. Either way the contribution is the vocabulary for the distinction itself — and the honest marking of where the framework's formal contact ends.
The method: structural analysis and formal engagement
Work 2 proceeds by cross-domain structural analysis — examining how the same dynamics play out across computation, biological cognition, mathematics, and meaning, surfacing features invisible from within any single domain — joined to formal engagement, where the framework meets established mathematics and reports the contact points with full honesty, marking each as direct, partial, structural, or none. A second strand is the mapping project: systematically deriving from fixed bottoms and watching the derivation encounter, in its own practice, the very features the theory predicts — the tree's vastness, the ease of checking over finding, the necessity of heuristics. The discipline is that prediction comes first. "This will be hard" is not a prediction; "hard in these specific ways, at these depths, for these structural reasons" is.
The shape of the work
Work 2 is planned as the Logic Corpus — a Prelude and five books — with a formal and a general paper. The structure is tentative; this is scoping, not settled architecture.
Prelude
Creates the felt experience of constraint before analysis begins — encountering something that resists, that narrows without guiding, that forces bets and punishes bad ones; the felt difference between checking and finding, between recognizing and generating.
Book 1 — The Nature of Constraint
The foundational derivation: from the fundamental activity through combinatorial explosion to constraint as the gap between space and path. Why bets are structurally necessary, what fixed bottoms are, the verification/search asymmetry, the gradient's internal logic. Everything else rests on this.
Book 2 — Computation and Navigation
The boldest claim given sustained treatment: the structural distinction between formal computation and navigation, the biological-cognition analysis, the unified account of AI as lens-space navigation, and the open scope question about whether the computational picture captures what cognition does.
Book 3 — The Formal Face of Constraint
Where the framework meets existing formal knowledge — the lens-space picture, the self-capture pattern extended to efficiency, the plurality of logics as a structural necessity — with honest status-marking throughout: convergence where it exists, gaps where it doesn't.
Book 4 — The Mapping Project
Systematic derivation from fixed bottoms: the experiment that tests the theory, watching the framework's own derivation exhibit the dynamics it predicts.
Book 5 — The Philosophy of Logic
The synthesis: what the account of constraint means for logic, inference, and necessity as such — why formal systems have the properties they have, and why what we have traditionally called "logic" is one expression of something deeper.