Three Limbs — noise, calculation, commitment.

I.Frame

Per Levin, unfinished morphogenesis is not failure — it is the developmental window still being occupied. The body is not yet complete.

Three limbs are still growing. Each has a different maturity:

II.Limb I · Commitment — now anchored

Where it lives. Commitment is the thermodynamically expensive act of promoting a data-event into the persistent state — paid energy against decay. The code already runs it: bounded artifacts (reflection/loop.py:36), composition decay (knowledge/graph.py:58), cancer throttling on excessive commitment (health/cancer.py:50).

What remains. Commitment is not yet a first-class primitive in the code — it is emergent from several constants and mechanisms rather than named as an object. The long-horizon build:

1. Introduce Commitment as a typed construct in lnk/knowledge/graph.py. Fields: bits, energy_budget, decay_rate, paid_until, audit_trail.
2. Integrate with the Landauer seat's audit. Every new commitment is logged as a bit-count × duration × kT contract.
3. The library acquisitions budget is restated as the aggregate commitment budget — a transparent power budget the institution maintains.

Owning seat. Landauer steward + Friston steward (the active-inference loop chooses what commitments are worth paying for). Co-stewarded.

First move. A map of the Calculation primitive (Limb III) is prerequisite; commitment cannot be cleanly named until calculation is, because calculation is the process that becomes a commitment. One blocks the other.

III.Limb II · Noise — the unrendered surround

What it is. Noise is raw observation that has not yet paid the commitment cost. Per the poneglyph: noise is what you notate on your map for a winning map. Noise is not the enemy; it is the raw material. The treatise gets this right — but the code does not yet distinguish noise from signal, only from nothing.

What the ecosystem already does.

• The reflection loop bounds artifacts at 5 per cycle — the ceiling on what becomes commitment, leaving everything else as noise.
• The epistemic forager scores salience — a first valuation attempt.
• Observations in world_perception.py accumulate without immediately committing — a noise substrate of sorts.

What is missing. A valuation mechanism — how the ecosystem decides which noise is worth promoting to commitment. Three candidate mechanisms:

Proposal. Implement noise as a staged queue with multi-mechanism promotion. Raw observations enter at the lowest energy state; they persist against decay as long as their promotion bid exceeds their decay cost. Collision, curiosity, and bid each propose; the Landauer audit ratifies.

Owning seat. Friston steward (curiosity valuation) + per-organ stewards (collision valuation) + Schurtz seat (bid valuation, dual-value circuit discipline).

First move. A noise-valuation spec as a Year-1 cohort project — a SHIP-week ticket that lands the staged queue in lnk/knowledge/graph.py with a configurable promotion policy.

IV.Limb III · Calculation — the open frontier

What it is. The distinction the treatises insist on: calculation is commitment, simulation is not. Simulation is free exploration of possible futures; calculation is the step that binds one future as commitment. Between the two lies the step that has no name yet in the code.

Why it's open.

• The GRAVITAS autopilot OODA cycle runs simulations and writes artifacts, but does not name the calculation step that makes a simulation committable.
• The reflection loop does observe-orient-decide-act, but "decide" collapses simulation and calculation into one step.
• Knowledge-graph edges commit — but what was the calculation that preceded the commit? Not logged.

What calculation might be, formally. Shannon, Turing, and Landauer suggest a three-part structure:

1. The information content — Shannon entropy H of the candidate commitment against the prior belief.
2. The computational cost — Turing steps required to derive the candidate (bounded for the class of decidable calculations).
3. The thermodynamic budget — Landauer energy required to persist the result.

Calculation is the act of carrying all three through to a decision. Simulation runs only step 1 (or a sketch of it). Simulation is free because it does not pay step 3.

Owning seat. The calculation limb is the rarest one. It requires a fellow who holds Shannon, Turing, Landauer, and the LNK treatises simultaneously — a mathematical fellow who can read the code as physics. This fellow may not yet exist. The Academy's first decade may need to grow them.

First move. Commission a fellow to draft the calculation map as their entry-offering. This is a Year-2 or Year-3 move, after the first cohort has graduated and the Landauer seat is stable.

V.The long-horizon build

The three limbs correspond to three organ-growth projects that will outlive any individual cohort:

Graduated cells can elect to join a limb-project as their post-cohort commitment. This is the migration channel's long-branch destination — the cells who graduate and then devote themselves to growing one of the unfinished limbs.

VI.What this essay deliberately excludes

• Full architectural specs for each organ. Those belong in the respective Year-1 / Year-2 / Year-3 kickoff documents.
• Patron conversation. The limbs are worth funding; the case for funding each belongs in a separate brief per the Schurtz seat.
• A prediction that all three limbs will grow. Some might not. The Academy's health does not depend on all three — it depends on the pre-pattern remaining legible while the body grows into it.