dotfiles/claude/memory/feedback-arc-typology-recognition-scaffold.md

name, description, type, originSessionId

name	description	type	originSessionId
ARC typology is a recognition scaffold, not a generation template	How to relate to the ARC content typology — it serves the steward's recognition of what arrives as ARC-shaped, not a production pipeline. Reference documents and pieces are grounds for recognition, not templates for generation.	feedback	b9a0781b-1d1c-418f-8ca3-70a51db83580

ARC typology — recognition scaffold, not generation template

Rule: the ARC content typology exists to let the steward recognize what is ARC-shaped when a thought arrives. It is not a production pipeline, not a notebook-to-public, not a schema to fill.

Why (steward, 2026-04-21):

"ARC is not meant to be an online notebook to public. But the forms that Jaccottet uses are useful to analyze, to have a better idea of how we can define the types. As you're getting to know me more and more deeply, you know that I am not a diarist. I am not an essayist. Sometimes a thought comes across one of the types we've defined already and I 'know' that it is for ARC. Sometimes not. My mind works across various forms. And my output can be… quite heterogeneous."

Three structural facts this establishes:

1. Not-diarist. ARC is not a daily-chronicle of thought. There is no expectation of regular entries or exhaustive coverage.
2. Not-essayist. ARC is not primarily an argumentative corpus. Essay is one type among many, not the default.
3. Recognition-based publishing. A thought either arrives already in the shape of an ARC type (and the steward knows), or it doesn't (and it doesn't become an ARC piece). The typology makes recognition possible.
4. Heterogeneity is structural. The steward's mind works across multiple forms; output is "quite heterogeneous." The typology must hold that range without flattening it.

How to apply:

• Definitions describe what a piece is, not how to write one. Recognition-criteria over generation-instructions.
• Every §1 entry needs clean in/out criteria. The ✗-boundary cases matter as much as the ✓-exemplars. The steward's discipline is "sometimes yes, sometimes no" — the spec must support that judgment.
• Multi-axis §0 is not a design inelegance — it is correct coverage. Heterogeneity has structural dignity. Don't try to collapse the axes.
• Reference texts (Bachelard, Visuddhimagga, Jaccottet) are grounds for recognition, not templates for generation. They deepen understanding of what the forms are; they don't prescribe what ARC pieces should look like.
• Jaccottet specifically: different atomicity (carnet-compilation vs individual-publication). Treat as formal-range reference + confirmation that heterogeneity has literary dignity in one body of work. Don't drift into "ARC as online notebook" thinking.
• When tempted to propose generative templates or production workflows, stop. The typology serves recognition, not generation. Offer definitions, boundaries, and exemplars; let the steward recognize or not.

What this does NOT mean:

• It does NOT mean the steward doesn't work on pieces — pieces get drafted, revised, polished. But the recognition that this is a piece for ARC is upstream of the drafting discipline, and the type the piece belongs to is recognized (often) rather than constructed.
• It does NOT mean the steward has no notebook-shaped activity — there may be notebooks, private writing, vault material. It means the ARC site is not that venue. ARC is the curated public body.

Test for the spec writer (executor): before finalizing a §1 definition, ask "can the steward use this to recognize whether an incoming thought is this type?" If the definition only helps someone write this type, it's incomplete.

Humic-layer reweighting 2026-04-21

The steward articulated a further frame that reweights how reference texts appear in §1: see user-influences-humic-layer.md for the full steward characterization.

Operative consequence for the ARC typology spec:

Reference texts (Bachelard's Poétique de la Rêverie, Visuddhimagga, Jaccottet's carnets, Weil on attention, etc.) are in the humic layer of the steward's reading — broken down, absorbed, enriching — not authorities the types are derived from. The types rest on the steward's work and his recognition of it. The influences are in discourse with the types, not in governance of them.

For §1 drafting, this means:

• Definitions in the steward's register, from the steward's practice. Not "per Bachelard..." or "following Visuddhimagga..."
• Influence citations appear as grounds for discourse / recognition, not as deeds of transfer. The ARC Reverie type is what the steward recognizes when reading Bachelard; Bachelard is not the source the type derives from.
• Example frame (Reverie): "The ARC Reverie type names a mode the steward recognizes and practices. Bachelard's La Poétique de la Rêverie is one of several texts in discourse with this type — he names from within the anima-register what ARC Reverie enacts on the page." Not: "Reverie is defined by Bachelard as..."
• Example frame (Meditation): "The ARC Meditation type traces a practice the steward describes in first-person mechanics. The Visuddhimagga offers technical vocabulary for the same structure (kammaṭṭhāna / samādhi / sati / vipassanā) — a humic reference, not the source of the definition."
• Same posture across all types. The citations are present where relevant; the work's ground is the steward's.

Test extension for §1 drafting: before writing a reference-text citation, ask "am I positioning this as an authority above the steward's work, or as humic enrichment?" If the former, rewrite.

Sequencing corollary 2026-04-21

Rule: do not apply the recognition-scaffold to the corpus (reclassifications, audits, systematic retypings) until the scaffold is stably complete.

Why: a recognition-scaffold's types are defined by their mutual boundaries as much as by their positive definitions. Fragment-vs-Observation-vs-Meditation-vs-Reverie-vs-Dream each sharpen the others. Reclassifying corpus pieces against partial scaffolds risks locking in classifications that later boundary-type definitions would displace. Steward articulated the principle 2026-04-21 after catching the executor about to execute reclassifications prematurely: "Maybe I'm being rash — I guess we should finish the spec before the audit."

How to apply:

• Corpus reclassifications (ARC audit follow-ons) wait until §1 is comprehensively locked, minimum through the adjacent types of whatever's being reclassified.
• Safe reclassifications: those where the source type is stable and the target type is locked, AND no currently-DRAFT type could plausibly claim the piece. Today (2026-04-21): Graffiti Girl → Reverie is mostly safe (both Observation and Reverie are defined well enough that Graffiti Girl is clearly Reverie), but still wait for Observation's LOCKED state so the boundary is explicit.
• When the spec is nearly complete, run the audit pass with the full vocabulary available. Then reclassify.
• Executor rule of thumb: when steward authorizes a spec decision, that authorizes writing the decision into the spec. Applying the decision to corpus pieces is a separate authorization, typically to be granted after the whole scaffold is stable.