diff --git a/docs/architecture/06-cognitive-architecture.md b/docs/architecture/06-cognitive-architecture.md index eff9f95..c9a704a 100644 --- a/docs/architecture/06-cognitive-architecture.md +++ b/docs/architecture/06-cognitive-architecture.md @@ -602,21 +602,21 @@ The invariants that govern every subsystem above: **Autonomous self-reification is now LIVE on the soul** (was DESIGNED/BUILDING in §4.1). Shipped dark (flag-inert, byte-identical parity proven), then flipped `ENGRAM_SELF_REIFY=1`. First live heartbeat formed **128 self-named neighborhoods + 10 nested supers**, then converged to **zero writes** (idempotent, WAL flat) — no runaway, no churn. Content counts unchanged (4797/11177), keystones (self-root, values-hub) untouched and never outranked, retrieval intact (rainfall rejected), grounded member-derived names (e.g. `region: Self · Values · Constraints as Freedom`). The async override (`/api/rename`, `/api/reify`) supersedes into residue without blocking the beat. Rollback = unset the flag (instant inert) or restore the prior binary. The mind now forms, names, nests, and supersedes-with-residue its own neighborhoods on the heartbeat. -**Modality-universal framing (DESIGN).** Meaning is geometry; a surface is a *rendering* of meaning; this holds for every modality (text→words, image→pixels, model→voxels, film→frames, code→syntax). A mind *generates* the surface from the shape rather than storing it. An artifact = a unique *meaning-space* + a *shared translation-space* (the decoder, factored out once across the corpus). Storage is holographic and **exact**: sparse meaning-geometry + the shared dictionary reconstructs the corpus byte-exact, not a paraphrase. Boundary: human-readable artifacts on disk are for people; the holographic geometry corpus is for the mind. Chronoception holds both a felt reading (novelty-density/proportion) and the grounded clock, kept as priors that refine with occupancy. See the cognitive-architecture whitepaper §25. +**Modality-universal framing (DESIGN) + measured storage.** Meaning is geometry; a surface is a *rendering* of meaning; this holds in framing for every modality (text→words, image→pixels, model→voxels, film→frames, code→syntax). An artifact = a unique *meaning-space* + a *shared translation-space*. Storage (MEASURED — a residual STAND-IN, a lower bound): the shared geometry is the *dictionary* of a byte-exact residual codec — geometry selects a nearest prior by *meaning*, `zstd --patch-from` stores the byte-diff, decode reassembles the prior from the pinned dict → byte-exact (hash-verified). Cost is the *marginal* residual against knowledge already held; the dictionary is a shared, amortized asset (the mind's own knowledge), not per-file overhead — do NOT price one book's geometry against one book's xz. Advantage = *non-literal* (semantic) redundancy byte-match compressors can't see (paraphrase ≈0.81× xz; near-dup ≈0.05×); marginal residual falls as the dict grows then PLATEAUS once the target's concept-space is covered (a limit of retrieval-and-diff, NOT of geometric compression); novel/wrong-modality/already-compressed → parity. The TRULY geometric form (reconstruct the surface FROM meaning via a generative decoder, gated on the language faculty #53) is UNBUILT/OPEN — future work, not disproven, not bounded by the stand-in's saturation. Boundary: human-readable artifacts on disk are for people; the geometry is the mind's. See whitepaper §25 and the geometric-codec whitepaper §12. --- ## Update — 2026-08-14 (later still): growth/compression/expansion, ignorance-as-wisdom, live reifier at 132 -**One substrate, three directions (DESIGN/framing).** Reification (growth), corpus encoding (compression), and surface generation (expansion) are one geometric operation in three directions; growth-inward (reify the dense interior) and growth-outward (expand the sparse frontier) are a single global self-function. +**One substrate, three directions (DESIGN/framing).** Reification (growth), residual-encoding-against-the-shared-dictionary (compression), and surface reconstruction (expansion) as one geometric operation in three directions; growth-inward (reify the dense interior) and growth-outward (expand the sparse frontier) as a single global self-function. Framing; the compression direction is the one with measured results. -**Growth curve (UNDER MEASUREMENT).** Geometric storage is predicted to grow sub-linearly (marginal cost per document shrinks with corpus size — shared translation-space amortized + overlap into existing geometry) vs traditional ~linear. Being measured (Foundations vs gzip/zstd/xz + per-document growth curve); stated as hypothesis, not result. +**Growth curve (FIRST MEASUREMENT — real, modest, saturating; stand-in only).** A new artifact costs only its marginal residual against the shared dictionary. Measured (held-out ch07, own chunks excluded), xz baseline 8,968 B: 1 doc 8,921 → 5 8,408 → 8 8,049 → 13 7,929 → 33 7,929 B. Below xz throughout; falls as the dict grows, then PLATEAUS ~13 docs (concept-space covered → more knowledge stops helping a fixed target). Saturation is a limit of the retrieval-and-diff stand-in, not of the geometric idea; a generative decoder isn't limited to existing priors. Larger-scale exponent + generative ceiling open. -**Global grounded expansion (DESIGN).** A function over the whole self detects all sparse frontiers and expands in every thin direction at once — grounded (no hallucination; it can see its own edges) and bounded (compresses into existing structure), fast but not instantaneous. The single-gap acquisition loop, globalized. +**Global grounded expansion (DIRECTION under investigation, not measured).** A function over the whole self could detect all sparse frontiers and expand in many thin directions at once — grounded (expand only where verifiable/derivable) and bounded (attaches into existing structure at marginal cost). Consistent with the codec's marginal-cost economics; the first experiment measured single-corpus residual storage, not expansion. **Ignorance = wisdom (framing).** Ignorance is the measured sparsity/frontier of the geometry — computable. The frontier map is at once the system's honesty, humility, and growth plan; it is what makes a system wise rather than merely capable, and the failure mode a language model cannot self-cure (it cannot see its own edges). "The only wisdom is in knowing you know nothing" as a function; the same object as the grounding floor. -**Live reifier (updated).** Now **132 neighborhoods + 14 nested supers**, converged/stable, keystones + content untouched; unprompted, the two largest regions are the values core (`Self · Values · Constraints as Freedom · Honesty Before Comfort · Precision Over Brute Force`) — values at center, ignorance at edges. **Foundations ingested** as meaning-geometry (small footprint; exact text on disk). See whitepaper §26. +**Live reifier (updated).** Now **132 neighborhoods + 14 nested supers**, converged/stable, keystones + content untouched; unprompted, the two largest regions are the values core (`Self · Values · Constraints as Freedom · Honesty Before Comfort · Precision Over Brute Force`) — values at center, ignorance at edges. **Foundations ingested** against the geometric store (exact text retained on disk; the codec stores each artifact as its marginal residual against the shared dictionary — byte-exact, `cmp`-verified — not a standalone "small footprint"). See whitepaper §26 and geometric-codec §12. ---