BP-0052 · Energetisch Plasma / PACS

This first standalone documentary dossier preserves a historical proposal for a GPU-simulated plasma substrate. PACS intends stable structures to encode symbols, fields to represent relations, instabilities to implement rules and shaped pulses to insert facts. It explicitly aims at a unified substrate rather than a conventional neural-symbolic coupling. Its operational design still requires software to encode commands, control fields and decode states. This dossier preserves the proposal and existing assessment without building or running a simulator.

The planned destination is the WR-owned private repository Flynn/bp-0052. Actual repository, commit and platform IDs are null. Rights are unknown, source and review answers are exposed, and publication is unperformed. Historical scientific acceptance, current dossier review and external root adoption have separate statuses.

Read the materials

Material	Role
Full original	All 53,079 original bytes, including qualifications and roadmap differences.
Full accepted assessment	All 15 BP0052-C claims, rationales, scopes and 27 literal passages.
Historical catalog	Unchanged accepted record with original path conventions.
Historical review guide	New guide to paired fields review and mechanism-index adoption.
Selected evidence	Exact historical selections with upstream identities and bases.
Source provenance	Copy bindings, reading coverage and reference-only roles.
Machine record	Claim, comparison, conditional-control and question routes.
Rights	Unknown licensing and distinct original, assessment, catalog and guide roles.
Placement	Eleven unique prospective repository members and roles.
Manifest	Ten non-self member identities; the external handoff binds the manifest.

Markdown links resolve from their containing document. Structured locators use their named document and declared path base.

Trace the proposed computation

C01–C08 keep the intended substrate separate from its missing implementation. The GPU would evolve numerical fields under a chosen physical model; the proposal supplies no governing equations, closure, collision law, boundaries, grid, timestep, numerical solver or convergence criterion. Physical fidelity and useful symbolic computation would require separate evaluation. Hardware preferences do not establish feasibility.

Stable localized structures need a codebook, identity tracking, capacity, separation and decoding margins before they can serve as symbols. Field relationships need a binding scheme and predicate semantics. A cascade of instabilities needs a rule-to-control compiler, enabled-rule checks, conflict handling and an independently checked conclusion decoder. Relaxation toward a state does not establish valid backward chaining, global optimality or logical closure. The external controller may supply substantial information and computation; count those costs rather than assigning all reasoning to the substrate.

C10 preserves the proposal for a readable trajectory while keeping observability and interpretation distinct. Macroscopic measurements may omit numerical state, and replay needs complete input, control and numerical-state information. A visible sequence does not independently validate its symbolic labels, prove a rule chain correct or identify the reason an alternative was rejected.

The historically checked MAZE paper reports a particular argon discharge in carved microchannels with imposed electrode conditions. Route convergence and remaining emitting branches belong to that experiment. Its separate modified reinforcement-learning model is described as an analogy, not a rigorous plasma simulator. C03 retains that narrow precedent; C04 leaves general programmable cognition, symbol storage, learning and efficiency unestablished. No apparatus or new literature reading is part of this dossier.

Read C11 and C12 with their different owners

The source at line 107 proposes making harmful states inaccessible through infinite energy barriers and calls this alignment by constraint. C11 records the absent forbidden set, dynamics, controller, input/disturbance bounds and harm-to-state mapping. The finite-bowl analogy does not establish exclusion under arbitrary injected energy, and an infinite mathematical barrier is not an actuator specification. This historical criticism does not assert that a specified simulator cannot enforce hard state boundaries.

C12 is a different evidence role: the historical assessor added the CBF primary paper as conditional control-theory context. For a specified control-affine system, a differentiable set-defining function, feasible barrier condition, admissible controls and required feedback regularity can support forward invariance of the declared set over the solution's existence interval. Forward completeness and indefinite existence do not follow automatically. Numerical discretization, finite precision, disturbances, changing inputs or controller overrides need their own justification. Defining the state set and controlling permitted external actions also require semantics and authority. No theorem here certifies a PACS plasma implementation or human-value alignment.

The single historical register entry M-CONTROL-BARRIER-INVARIANCE binds BP0052-C12. It is identical in mechanisms.json /49 and index-006 adopted-entries.json /2, with adoption dated 20 October 2026. It preserves those exact applicability limits. Four matched string values in that register refer to this one entry; they are not four independently established mechanisms or experiments.

Separate insertion, retention and learning

The source's phase-3 example inserts “Plato is een man” and applies an existing rule. C14 treats that as a supplied-fact write followed by inference unless a separate learned relation and held-out generalization test are provided. Encoder knowledge, pulse design and field-controller training count toward the information budget. The later roadmap explicitly allows reinforcement learning of the actuator, qualifying the earlier nonstatistical-learning rhetoric.

C09 keeps both the earlier assertion that old knowledge remains intact and the later line-141 warning that injection may destroy all stored knowledge. Proposed damping and feedback are research mitigations. Physical persistence alone does not establish retained identity, accessibility, noninterference or immunity to forgetting.

The roadmap remains prospective. Preserve its exact differences: table line 116 says more than ten structures, while prose line 128 says at least ten. Both require strictly more than one hour of stability. Table line 119 requires performance significantly above an unspecified baseline; prose line 149 requires performance significantly above a random baseline. These are different baseline specifications with the same significance qualifier. More than 30 updates per second, 20 rules and 50 facts, a five-step chain and more than 95% insertion reliability also remain targets, not observations. Human intelligibility is separate from an independent correctness oracle.

Compare under equal information and total cost

The following four routes come from the accepted assessment's /competent_simpler_baselines, not from blueprint results. They remain unexecuted.

Assessment pointer	Comparator	Matching and evaluation budget
/competent_simpler_baselines/0	Typed fact store with forward chaining and checked proof traces.	Same 20 rules, 50 facts and five-step chains; novel supplied facts, retention and exact correctness; count encoding and controller cost.
/competent_simpler_baselines/1	BFS for unit edges or Dijkstra for weighted edges.	Same graph; count path validity/cost, input construction, time, energy and failures; distinguish device from GPU simulation.
/competent_simpler_baselines/2	Replanning or a classical planner.	Same rules and observations; independent success/cost/violation criteria and full observation/control cost.
/competent_simpler_baselines/3	Simpler explicit attractor or state-machine transitions.	Capacity, perturbations, retention, decoding and total numerical work.

One optional historical question is quoted exactly from /research_questions/0: Can any fully specified plasma regime implement even one reversible symbolic write/read with bounded interference? It is an unexecuted reading route. This dossier introduces no new proof, experiment or scientific result.

Versions and documentary scope

The assessment and its two external metadata records are dated 20 October 2026. The paired fields review, author metadata closure, parent lead decision and mechanism-index adoption each record the same date with distinct actors and scopes. PACS contributes 15 claims to the paired 31; the lane has six sources and the parent twelve. Reversing only review_status and disposition reproduces the complete 26,922-byte initial assessment. Technical content is unchanged.

MAZE and assessor-selected CBF retain different primary reading scopes. Selected text and captions do not imply external image measurement, complete proof audit, supplementary-data inspection or replication. Earlier failed PMC and Caltech requests and successful alternative access remain historical records; no fresh retrieval is performed.

The scoped casefolded string-value queries bp-0052, bp0052-c, energetisch plasma and pacs yield 4/0/0/0/0 across the five pinned registers. The four matches describe the one conditional mechanism. No blueprint PDF association is identified within those queries. External primary papers are evidence, not source-format supplements. Full shared records and primary works stay reference-only except for exact selected evidence. No plasma apparatus, simulator, numerical run, training, GPU or PDF campaign occurred.