Changelog

The capture rule (published live 2026-07-02) says the threshold for capture is existence, not merit — merit is decided by the climb, in public — and that promotion into the canon roster is a climb event, not a capture event. By this version the /open register and the spine had diverged: four candidates from the ratchet-cell run, plus one structural insight surfaced at the tail of the second-commodity run, had lived and been refined across three gated treatments (ratchet cell, second commodity, forest school) without appearing in propositions.ts. This version reconciles the registers. Each node enters synthesised from all three runs, not just its original capture, and each ships in-surface with its sharpest adversary, per house style. And the spine records its first same-version climb: the-closing-window entered at forming and climbed to contested inside this release, because its own stated climb condition — a third domain's residue is (or is not) a race — fired on the "(or is not)" branch: the forest-school domain's residue is a window that opens on a known date (~2028 curriculum reset), which reshaped the claim from a law about races into a typology of clocks. The reshaped claim, not the original, is what climbed; the rent/protection line it is gated by remains a judgment rather than a measurement, which is exactly the disagreement the contested rung marks.

The first real climb under the v0.11.0 regime: coordination-bounds-reachability, one of the three nodes that entered at forming when the draft wall came down, migrates to established. The coordination/enclosure edge of the reachable set held across multiple worked applications (the Ogwen DNA-economy evaluation; route-around-the-chokepoint, which is its infrastructure corollary) and became load-bearing for downstream nodes — which is what the climb records. Entered in this history retrospectively at v0.12.0; the climb itself was recorded inline in the node's statusHistory when it happened.

A Living Epistemic Work cannot keep a backstage. v0.2.0 collapsed the draft/canon queue in editorial practice but left the word "draft" — and a /drafts surface — standing, which kept smuggling the dead binary back in: draft implies not-yet-real, will-be-replaced. This version retires the notion entirely. The settledness spine gains an entry rung, forming — thinking that is published and live but not yet load-bearing — so a node no longer has to arrive already-settled; it can enter at the bottom and climb in public. Status stops being a fixed label and becomes a path: every proposition now carries a trajectory, and the climb is rendered as a ribbon on /explore and as a dedicated surface, /motion, where the whole canon's migration is laid along the version axis. The histories are backfilled honestly from the version-grain record — and they show, plainly, that almost every legacy node entered straight at derived or contested, because the old regime thought claims through behind the wall before letting them appear. That flatness is not hidden; it is the evidence for the reform. This is the format applying its own thesis to itself: the work asks economies to thaw frozen value back into the reachable set, and here it does the same to its own backstage.

The work has described itself from the outside in — on its author's own site — as an invitation to an infinite game in James Carse's sense: the point is not to win or finish it but to keep it going and bring others in. That framing had never entered the work itself. v0.10.0 brings it inside. A new proposition, infinite-game, names what the viable objective has been all along: a finite game is played to win and ends in a win; an infinite game is played to continue the play. The no-regret viable objective is the rule of an infinite game — it admits no winning move, only the refusal of game-ending ones — and option space is that game given a thermodynamic substrate. James Carse joins the lineage beside Aubin and Snowden (who give the method of staying in play) and paired with Buckminster Fuller (the generative means the purpose serves). And /polychess is recut to match: the win and the high-score are gone; the only outcome that means anything is ending the game, and the board now exists to deliver one turn — you try to win, you find you cannot, and you arrive at the game that has no win. The finite game was the bridge; the infinite game is the far side.

For eight versions the work entered through its cost side: throughput is cost, the inheritance is being burned, physics is being ignored. All true — and all a charge sheet, and a charge sheet enrols no one. v0.9.0 re-sequences the work to enter through generation instead. The essay's orientation now opens on the generative claim — value is not stored but made; configuration generates configuration; the economy's task is to keep that generative game open — and lets the familiar cautions arrive as the single failure mode they always were: the engine run down. Nothing is softened and no status marker moves; the optimism is earned from the generative physics, not from promoting a hope to a fact. The same turn surfaces the framework's most conspicuous silence as a proposition rather than a gap. coordination-as-move-evaluation reframes the open question of what coordinates an economy once labour is no longer the allocator — not as the search for a right price, but as distributed local Δω-evaluation across many boards. A direction, drawn with conviction; the mechanism itself is kept open. And /polychess makes the move-rule playable: a reader feels option space open and close rather than being argued into it — the work's answer to "words may not be enough."

The format's central claim has been that a claim is not a monolith — it is an explicit set of premises with an epistemic status and a connective neighbourhood, so any single premise can be stress-tested on its own. v0.8.0 makes that claim operable. A new /attack surface lets a reader target one premise and get a four-step response from the guide: steelman the strongest version of the attack, respond from canon, concede what the attack legitimately costs, and trace the blast radius — which connected propositions actually lean on the premise and which survive without it. Three worked examples span the epistemic-status tiers (an established premise, a contested premise, an open premise) to show the response calibrates to confidence: the established premise gets a strong defence and a scope-level concession; the open premise is not defended at all, only located. The same affordance is wired in-place on /explore: open any proposition's logic and each premise carries an attack control. This is the demonstration a printed paper cannot offer — interactive, per-premise falsifiability over the work's own dependency structure.

Alexander, the guide, is now mounted on every surface where there is grokkable text — not just on /essay and /applications/moves. The mode-adaptive register that has existed since v0.1.5 is now actually reachable everywhere it was designed for: /overview adopts overview register, /explore adopts explore register, /research-frontier adopts research-frontier register, /lineage and /objections gain dedicated registers (predecessor-mapping and critique-response companions), /practice reuses overview. Each surface gives the reader the same calm guide tuned to the resolution that surface is operating at.

Push 3: the rule becomes an instrument. Alexander is now wired into /applications/moves with a dedicated move-evaluator register. A reader who brings their own move — a policy, an infrastructure commitment, a land-use change, an institutional reform — gets the structured analysis the page's six worked cases use, produced by the framework's own logic. Where the verdict is ambiguous, Alexander says so. Where the move sits near one of the five failure modes in "Where this rule runs out", Alexander names the edge plainly. The credibility apparatus from v0.5.0 backs the verdicts when readers push back: Alexander can defer to /lineage for predecessors and /objections for engaged attacks.

The long-pending counterargument-engagement draft lands as /objections, pairing structurally with /lineage. Five engaged objections — measurability, degrowth-in-disguise, services economy, decoupling data, Solow substitutability — each as steelman → response → concession. Three attacks recorded as resolved by other work (normatively smuggled propositions, novelty/lineage, solar-flux scale); five recorded as partly resolved (political economy, AI guide trust, AI-symbiotic authorship, living-document instability, category error in epistemic status). Three genuinely open structural critiques named as future work rather than answered weakly.

Integration pass following the v0.4.0 push. §4 (configuration-value) prose carried the generative-mechanism claim implicitly throughout — "structures that store, transmit, and reuse information"; "compound without proportional energy increase" — without naming the structural move. The integration adds one sentence to the biological-evolution paragraph that names the mechanism (each configured arrangement makes further arrangements reachable; accumulated arrangement lowers the cost of further arrangement), and a clause to the "compound without proportional energy increase" line that surfaces the rate-grows-with-accumulation claim. §4.5 was already tight — the productive/destructive distinction is named in prose; no integration touch needed there.

The expansion-side counterpart to v0.3.0's asymmetry lands as a third v0.7 proposition: configuration-generates-configuration. With it, CE has both faces of the option-space dynamic — fast-local destruction (asymmetry) and slow-systemic compounding expansion (generative). The /lineage page ships alongside, separating the credibility apparatus from the calm essay. Seven predecessor clusters with explicit deltas, and an honest update to how CE relates to Cronin/Walker's Assembly Theory now that the 2023–24 reformulation has closed part of the gap CE had been citing.

Mechanical finishing touches on the v0.7 integration. Seven existing propositions gain backlinks to the new v0.7 propositions where the relationship is meaningful in both directions. /overview and /research-frontier now point at /applications/moves alongside the existing route map — the rule's operational surface is now discoverable from both the structural-view modes and the open-boundary mode.

Integration pass on the essay. §5 (work-wrong-question) prose now gestures toward labour-as-allocator's structural distinction — allocation as load-bearing, production as its visible form — rather than re-performing the move implicitly. §7 (viable-objective) prose carries a real drift fix: the section was still shipping the pre-v0.2.0 "maximisation of durable human flourishing" framing even after P22 was reformulated to no-regret in v0.2.0. The Alexander system-prompt summary was updated then; the essay's own prose was missed. §7 now ships the no-regret form and names asymmetry-of-option-space-change as the structural reason for it.

Two propositions promoted that the framework had been gesturing at without naming: labour as allocation mechanism (not source of value) and the structural asymmetry between option-space destruction and expansion. With them, work-wrong-question gains its historical reframing and viable-objective's no-regret form becomes a derived consequence rather than a stance. A new /applications/moves surface lands alongside, showing the local Δω rule operating on six worked cases — fossil-fuel infrastructure, antibiotic stewardship, stratospheric aerosol injection, soil-regenerative subsidies, copyright term extension, biodiversity offsetting — with two ambiguous verdicts to demonstrate the rule's edges.

The framework gains a structural answer to "option space for whom?" via Wolfram-style observer-relativity, and an operational counterpart via local move-evaluation in the spirit of Snowden's next-best-move. P22's viable objective shifts from scalar maximisation to no-regret preservation over moves. Three essay sections rewritten to engage the physics-rigour and decoupling literatures directly. Editorial practice changes: drafts no longer queue behind voice-review for canonical promotion; revisions ship straight to canon with a revisedAt date, surfaced as a UI badge.

Alexander, the in-site epistemic guide, becomes proposition-aware: the system prompt receives the propositions relevant to whichever essay section the reader is on, and Alexander references them naturally rather than constantly. Final v0.1 release before the v0.2 push.

Two propositions added that close the v0.1 set at 22 nodes — one disarming the techno-optimist "everything is digitisable" move, one naming what replaces throughput as the unit of success.

Substantial expansion of the proposition set — time asymmetry, displaced costs, contextual scarcity, substitution limits, growth and distributional strain, money as signal, coordination as wealth, the legibility/truth tradeoff, transition fragility, and the choice to ignore physics. Brings the explore surface to twenty nodes.

Adds the configuration-maintenance cluster: structural memory, complexity's ongoing cost, care as configuration rather than sentiment, the prevention/repair asymmetry, and the distinction between stability and stasis.

Architectural shift away from a typed-graph data model toward proposition nodes that are self-contained and meaningful in isolation. Links remain non-semantic ("links to" only). Restraint — not expanding surface area prematurely — becomes the design principle.

First public version. Establishes the living-epistemic-work format: essay with collapse modes, an in-site epistemic guide (Alexander), and the framing that economics needs a physics-grounded reframe. The essay "Value under Physical Constraint" lands in eleven sections; Alexander handles reader questions in voice; concept popovers surface key terms inline.