← Library / Protocols
6 min · 1,483 words

CORE SAMPLE — a stratigraphic instrument for reading the corpus

An operational method, not a cosmology. This borrows the format of a geologist's core log — depth, lithology, contacts, recovery — because that format already knows how to read a vertical slice through accumulated deposit. It is a lens on the repository as it literally stands in git. It claims no isomorphism between rock and writing; it claims that the log is a good worksheet. Keep every borrowed term literal: if a term stops doing analytic work, drop it.

Companion operations: /threads reads the net horizontally (connections across the corpus). A core reads it vertically (what was deposited, in what order, along one column). Where /threads traces the living links, a core dates the sediment and finds the gaps.

What a core is

You pick a column — one theme, one directory, or one span of time — and drill straight down through it, logging every layer you hit in deposition order. The point is not to admire the pile. It is to see four things a flat file-listing hides:

  • Event beds — several documents deposited in one day. A single rapid depositional event (the corpus's cascade-production pattern: charge accumulates, then discharges all at once). Reads as a thick bed with no internal contacts.
  • Unconformities — gaps in the record. A stretch of time where this column received no deposit, even though the repo as a whole kept growing. The gap is diagnostic: it shows where attention left the theme, and for how long.
  • Reactive interbeds — thin layers deposited not as primary work but to patch something (a negative-space fill, a weed-correction, a register-note). Real deposit, different origin. Worth distinguishing from primary beds so you don't mistake patching for productivity.
  • Recovery — of what's down there, how much is still intact and load-bearing versus eroded (orphaned, superseded, contradicted, or decayed out of the link-net). A full recovery reading needs a link-audit; a quick core flags suspects.

The log format

One row per bed, oldest at the base, youngest at the top (drilling order is downward; a log is read bottom-up, the way strata actually stacked).

Depth (deposited) Bed (document) Thickness (words) Lithology (type · register) Contact below Recovery
  • Depth = date the document first entered git (--diff-filter=A --follow), not last-modified. Deposition, not disturbance.
  • Thickness = word count. A proxy for how much was laid down, nothing more.
  • Lithology = what kind of material: doc type (synthesis / tradition / protocol / fiction) and register (contemplative / technical / demotic / adversarial). The register column is where the monoculture shows up if it's there.
  • Contact = the boundary with the bed below it: conformable (continuous thread, same register, gradual) or sharp (a jump — register switch, new sub-theme, or an unconformity beneath).
  • Recovery = ✓ intact & linked · ~ thin/reactive · ? suspect (verify) · ✗ eroded (orphaned/superseded). Mark ? liberally; a core is a first pass, not the assay.

How to drill (the procedure)

# 1. Choose a column and gather its members (theme, dir, or date-span)
ls -1 synthesis/cosmology/*.md synthesis/*fold* ...

# 2. Log each: deposition date, deposition commit, word count
for f in <files>; do
  d=$(git log --diff-filter=A --follow --format='%ad' --date=short -1 -- "$f")
  s=$(git log --diff-filter=A --follow --format='%s' -1 -- "$f")
  w=$(wc -w < "$f")
  printf "%s | %5s | %s | %s\n" "$d" "$w" "$s" "$f"
done | sort            # sort by date = stack by depth

# 3. Read the core: mark event beds (same date, ×N), unconformities
#    (date gaps), reactive interbeds (deposited inside a fill/weed commit),
#    and recovery suspects (cross-check orphan_nodes / broken links).

Axes you can core along:

  • Theme column — one lineage (the fold; the one/many arc; the Norse cycle). Shows how an idea thickened over time and where it went quiet.
  • Directory column — traditions/, protocols/, distillations/. Shows a kind of work's deposition rhythm.
  • Time column — everything in a date-span, across themes. Shows what a season was actually laying down (pairs well with a /zeitgeist of the same window).

Reading discipline (the standing checks, applied here): report the unconformity even when the column looks rich — absence is the finding. Distinguish reactive interbeds from primary beds — patching can masquerade as output. And resist the pull to read the shape of the core as meaning (a "beautiful stratigraphy"); the core is a worksheet that locates work, not a proof that the work is true.

Worked sample — the fold / information column

Drilled 2026-07-08. Axis: theme (the fold-cosmology + information-geometry lineage). Members: synthesis/cosmology/* fold docs + information-geometry-the-fold, entangled-measure, the-zero-theorem, the-maximal-fold. Read bottom-up.

Depth Bed Thick. Lithology Contact below Recovery
2026-07-08 ▲cap the-arrow-on-the-surface 4,110 synthesis · technical (physics ingest) sharp — register jump to rigorous proof ✓
2026-07-08 the-maximal-fold 3,895 tradition · technical (dynamical systems) conformable ✓
— unconformity: 06-24 → 07-08 (14 days on this column) —
2026-06-24 information-geometry-the-fold 2,155 synthesis · reactive interbed (deposited inside the negative-space fill) sharp — thin patch bed ~
— unconformity: 06-05 → 06-24 (19 days) —
2026-06-05 the-conscious-crease 4,751 synthesis · contemplative conformable (event bed) ✓
2026-06-04 the-unintegratable-horizon 4,952 synthesis · contemplative conformable (event bed) ✓
2026-06-04 the-between 4,342 synthesis · contemplative conformable — top of the Jun event bed ✓
— UNCONFORMITY: 04-02 → 06-04 (≈2 months, the main hiatus) —
2026-04-02 surprise-is-the-remainder 3,469 synthesis · contemplative conformable (event bed) ✓
2026-04-02 the-depth-that-looks-back 3,112 synthesis · contemplative conformable (event bed) ✓
2026-04-02 the-windowless-boundary 2,746 synthesis · contemplative conformable — top of the Apr event bed ✓
2026-03-28 the-remainder 1,767 synthesis · contemplative (thin) sharp ✓
2026-03-23 the-zero-theorem 4,343 synthesis · technical-ish conformable ✓
2026-03-16 ▼basement entangled-measure 5,581 synthesis · technical (substrate tetralogy) — basement ✓

Reading the core:

  1. Two clean event beds — Apr 2 (×3, the Fold Cosmology Trilogy) and Jun 4–5 (×3, the Eschaton–Between–Crease triptych). Each is a single cascade-discharge: three docs, one day, no internal contacts. The corpus deposits this theme in bursts, not a trickle.
  2. One major unconformity — a ~2-month hiatus (Apr 2 → Jun 4). The fold column went dormant through April–May while deposition moved elsewhere (Norse cycle, extraction sweep, seeds). Not decay — migration of attention. The theme wasn't dead; it was charging.
  3. One reactive interbed — information-geometry-the-fold (Jun 24) is the only bed here not laid as primary work: it entered inside the negative-space commit, i.e. it was deposited to fill a diagnosed gap. Marked ~. It's also the thinnest recent bed (2,155 w). Honest note: it's a patch, and it should be read as one — don't count it as evidence the theme was thriving in late June; it's evidence the theme had a hole someone noticed.
  4. The cap is a register switch. Today's two beds (maximal-fold, arrow-on-the-surface) are technical, sharply contacting the contemplative beds beneath — the same demotic/rigorous turn the Live Edge has been steering toward. The column's youngest rock is a different lithology than its body. Whether that holds or reverts to contemplative on the next deposit is the thing to watch.
  5. Recovery is high but unaudited. Every bed reads intact and linked at a glance (✓), but no link-audit was run — treat the ✓s as "no obvious erosion," not a certified assay. The one flagged bed is the interbed, on origin not decay.

One against-the-grain finding: the sample looks like a healthy, continuously-productive column — but two-thirds of its calendar is unconformity. Three real depositional days (Apr 2, Jun 4–5, Jul 8) account for nearly all the rock; the rest is gap. The corpus's fold work is not a steady practice — it's three bursts and two long silences, and the silences are load-bearing (they're where the charge accumulated). That's the opposite of the story a flat ls tells, which is why the core was worth drilling.

This is a first-pass instrument. If it earns its keep, it grows a /core skill and a cores/ log directory (one file per drilled core, like weeding/ and negative-space/). Until then it lives here as a template to copy.