gbrain: sync converted org-mode brain files

projects/passepartout/architecture/architecture.org

@@ -45,6 +45,8 @@ The knowledge subsystem maintains two indices over the Org prose:
 
 The prose is always ground truth. Both indices are derived views that can be rebuilt from scratch. Nothing is lost in the indices that was not already in the Org files.
 
+The same principle extends beyond prose to structured data. Empirical parameters, validity envelopes, provenance chains, and benchmark results live in Org as property drawers and tables — the same format the user reads and edits. The system maintains a derived representation — the provenance store — optimized for machine queries: filter by confidence interval, find all models valid for a given input class, trace a parameter to its experimental source. Like the two indices, the provenance store is a derived view rebuilt from Org, not a separate system with its own canonical copy. Nothing is lost in the store that was not already in the Org files. And when the system learns something new — a validated parameter, a benchmark result — it writes back to the Org files, keeping the human layer current.
+
 This is what sovereignty means in technical terms: the user owns the data in a format they can access, and the system operates on the data in the same format they own. The format is stable — Org-mode has been in active development since 2003. There is no schema migration, no database upgrade, no vendor lock-in. Your notes survive the system.
 
 **The verification subsystem: the gate.**

projects/passepartout/architecture/stage-0-now.org

@@ -28,6 +28,7 @@ The conventional stack spans every layer:
 | Application | XSS, SQLi, RCE, dependency chain attacks, supply chain |
 | User | phishing, social engineering, credential theft |
 | LLM (if present) | jailbreaks, prompt injection (unbounded space), data leakage in outputs, probabilistic unreliability |
+| Empirical provenance | No systematic model validity checking. Parameters lack provenance, validity envelopes absent, neural networks treated as black boxes with no distribution match |
 
 **Key property:** Every layer is independent and untrusted. No layer can vouch
 for any other. Security is *empirical* — "no bugs found in this release" — not

projects/passepartout/architecture/stage-1-social-protocol.org

@@ -24,6 +24,7 @@ Communication becomes provable - when you choose it to be.*
 - Pseudonymous Personas for deniable identity
 - Relays as transient routers (pub/sub model, no long-term storage)
 - Onion routing between PDSs for metadata masking
+- **Provenance store seeds:** signed data exchange for empirical parameters. When instances share a validated force field parameter or benchmark result, the message carries the experimental source signature and validation history. The receiving instance stores it with provenance — the first structured empirical data in the knowledge store
 
 ## What is eliminated
 

projects/passepartout/architecture/stage-2-verification.org

@@ -22,6 +22,8 @@ Capability-based authorization. "Root" as an attack target no longer exists.*
   - Does the action violate any system invariant?
 - Decision procedure formalized in ACL2, machine-checked
 - Gate runs as a decision layer on the conventional host (Stage 0 [[id:84a537b4-4256-50c8-91f5-dd5b4538418f][hardware]])
+- **Provenance store becomes operational infrastructure:** a structured data extension of the Knowledge subsystem storing empirical parameters, validity envelopes, and benchmark results with full provenance chains
+- **Validity envelope predicates as gate vectors:** the gate checks not only security policy but also scientific validity — is this model valid in this context? Are the parameters within their validated range? Does the input fall within the model's training distribution?
 
 ## What is eliminated
 

projects/passepartout/architecture/stage-3-lisp-machine.org

@@ -33,6 +33,10 @@ user interface into one address space:
   compatibility layer inside CL image → native CL implementations.
 - **Minibuffer:** universal command surface — edit files, navigate web,
   run Lisp expressions, invoke agent commands.
+- **Provenance store becomes native Lisp hash table with persistence.**
+  The symbolic engine queries it directly — no IPC, no file parsing at runtime.
+  Empirical parameters, validity envelopes, and benchmark data live in the same
+  address space as the evaluator and gate
 
 ### Phase B — Cannibalization (3-5 years)
 

projects/passepartout/architecture/stage-4-inference.org

@@ -23,6 +23,7 @@ during generation. No external API, no separate trust domain.*
   tagged Lisp object pointing to flat binary)
 - Deterministic inference: same input, same output, same hash — auditable
   and replayable
+- **Distribution match check runs at evaluation loop speed:** the gate's validity predicate for neural network models becomes a fast native function — compute distance to training distribution and compare against threshold in the same process, no IPC
 
 *Two neural components on the same substrate:* Stage 4 hosts the LLM for
 generative reasoning and action proposals. The LeCun-type world model (sensory