amr/hermes-brain
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Remove old design-decisions.org, update cross-refs to design/ folder

Browse Source
This commit is contained in:
Hermes
2026-06-04 20:24:17 +00:00
parent 2ee60b5b4a
commit 9ce3883005
5 changed files with 8 additions and 87 deletions
Download Patch File Download Diff File Expand all files Collapse all files

10
projects/passepartout/architecture/distinguishing-features.org
View File

@@ -11,7 +11,7 @@ This is a working list of what distinguishes Passepartout's architecture from al
**1. Neurosymbolic (vs pure neural)**
Not an LLM bolted onto a database. The system fundamentally bridges neural and symbolic representations, not as an integration layer but as a unified semantics. See [[id:1c3ec48b-446c-50d2-b53e-126a81f5143f][architecture.org]] and [[id:0a33bd83-ff3c-4eac-bc97-83eb6702051a][design-decisions.org]] for the full architecture and design reasoning.
Not an LLM bolted onto a database. The system fundamentally bridges neural and symbolic representations, not as an integration layer but as a unified semantics. See [[id:1c3ec48b-446c-50d2-b53e-126a81f5143f][architecture.org]] and [[id:e32290a0-a02a-4af7-ae22-243d04a7ac82][Design Decisions]] for the full architecture and design reasoning.
**2. Symbolic above neuro (vs other neurosymbolic architectures)**
@@ -19,15 +19,15 @@ The Gate is the mechanism: an ACL2-verified deductive layer that is authoritativ
**3. Org-mode data store + neural index + symbolic index**
One file format for human and machine. The Org file is not a representation of the data — it IS the data. Both indices (neural embeddings for semantic search, symbolic assertions for formal reasoning) are derived views that can be rebuilt from the Org source. No translation layer, no schema migration, no vendor lock-in. See [[id:design-org-unified-ast][design-decisions.org]] for the full analysis.
One file format for human and machine. The Org file is not a representation of the data — it IS the data. Both indices (neural embeddings for semantic search, symbolic assertions for formal reasoning) are derived views that can be rebuilt from the Org source. No translation layer, no schema migration, no vendor lock-in. See [[id:design-org-unified-ast][Org-Mode as Unified AST]] for the full analysis.
**4. Wikipedia bootstraps general knowledge, LLM bootstraps specialized domains**
General world knowledge seeds from structured sources (Wikidata, Wikipedia infoboxes) at minimal cost. Specialized domains (regulatory compliance, physics, medicine) are extracted by LLM from prose, then verified by the Gate. The bootstrapping is cheap and incremental, not a 24.5M-assertion Manhattan project. See [[id:0a33bd83-ff3c-4eac-bc97-83eb6702051a][design-decisions.org]] for details on the Wikipedia knowledge seed strategy.
General world knowledge seeds from structured sources (Wikidata, Wikipedia infoboxes) at minimal cost. Specialized domains (regulatory compliance, physics, medicine) are extracted by LLM from prose, then verified by the Gate. The bootstrapping is cheap and incremental, not a 24.5M-assertion Manhattan project. See [[id:e32290a0-a02a-4af7-ae22-243d04a7ac82][Design Decisions]] for details on the Wikipedia knowledge seed strategy.
**5. Lisp and homoiconicity — one address space**
Editor, shell, browser, and agent run in the same Lisp image. The Gate is in the evaluation loop itself, not interposed as an OS layer. There is no MMU boundary between components because there are no separate processes. No IPC, no kernel boundary to attack. The system verifies code at the level where code and data share representation. See [[id:1c3ec48b-446c-50d2-b53e-126a81f5143f][architecture.org]] for the full address space argument, and [[id:0a33bd83-ff3c-4eac-bc97-83eb6702051a][design-decisions.org]] for the homoiconicity foundation.
Editor, shell, browser, and agent run in the same Lisp image. The Gate is in the evaluation loop itself, not interposed as an OS layer. There is no MMU boundary between components because there are no separate processes. No IPC, no kernel boundary to attack. The system verifies code at the level where code and data share representation. See [[id:1c3ec48b-446c-50d2-b53e-126a81f5143f][architecture.org]] for the full address space argument, and [[id:e32290a0-a02a-4af7-ae22-243d04a7ac82][Design Decisions]] for the homoiconicity foundation.
**6. Social protocol**
@@ -51,7 +51,7 @@ Because code and data share the same representation in a Lisp address space, the
**11. Cost inversion (80% symbolic at near-zero marginal cost)**
Symbolic reasoning is typically expensive (knowledge engineering, ontology maintenance). Passepartout inverts this: the LLM generates symbolic assertions cheaply, the Gate verifies them deductively, and the Org source stores them as a human-readable byproduct. The expensive part becomes the verification, which compounds with every use. See [[id:0a33bd83-ff3c-4eac-bc97-83eb6702051a][design-decisions.org]] for the token economics analysis.
Symbolic reasoning is typically expensive (knowledge engineering, ontology maintenance). Passepartout inverts this: the LLM generates symbolic assertions cheaply, the Gate verifies them deductively, and the Org source stores them as a human-readable byproduct. The expensive part becomes the verification, which compounds with every use. See [[id:e32290a0-a02a-4af7-ae22-243d04a7ac82][Design Decisions]] for the token economics analysis.
**12. Potential revenue streams from the Social Protocol**