Nooplex Readiness Map¶

Status: Active tracking authority

Purpose: Record what ProbOS has demonstrated relative to the Nooplex architecture, what claim the evidence permits, and which existing decision owns the next gate.

This document tracks readiness, not aspiration and not implementation inventory. A component existing in source, passing a unit test, or being enabled in one local configuration does not advance a readiness tier by itself. A tier advances only when its evidence gate passes through the production paths named below.

Tracking Authorities¶

Question	Authority
Which AD/BF numbers are allocated or shipped?	Live enumeration of Git subjects, GitHub issue titles in all states, and in-flight `prompts/ad-*.md`; the generated AD/BF ledger is a reconciliation view
What is ordered for future implementation?	Development roadmap
What work makes one ProbOS mesh dependable?	AD-1270 Platform Maturity Program / #1324
What is the Tier A P0 seam denominator?	Stable-ID P0 manifest; removed IDs remain reasoned tombstones
What is the initial maintainability/test-cost observation?	2026-08-30 reproducible baseline; discovery evidence, not a passing readiness gate
What evidence supports a release?	AD-1186 Ship Trials / #1123
What does the Nooplex architecture require?	The Nooplex
What readiness claim is currently permitted?	This document

Technical readiness belongs in the OSS repository because it describes how the product works. Commercial positioning may link here but must not maintain a competing technical status table.

Status Vocabulary¶

Status	Meaning
`not-started`	No active implementation owner or evidence program
`planned`	Decision and owner exist; implementation has not passed its gate
`experimental`	A working path exists, but security, repeatability, scale, or evidence is insufficient for supported use
`supported`	Versioned configuration, production-path exercise, recovery, security floor, and release evidence pass
`research`	A falsifiable hypothesis is being studied; it is not a product capability claim
`disconfirmed`	Required evidence failed a predeclared criterion

unknown, skipped, not exercised, and judge unavailable are never synonyms for passed.

Readiness Tiers¶

Tier	Name	Current status	Owning program	Permitted claim
A	Dependable Cognitive Mesh	`planned`	AD-1270 / #1324	“ProbOS is an alpha implementation of a governed Cognitive Mesh.”
B	Secure Multi-Mesh	`planned`	AD-1196–1198 / #1140	“ProbOS has an experimental federation transport.”
C	Nooplex Core Fabric	`not-started`	AD-1303 / #1356 bounded pilot planning; AD-1304 / #1357 conformance	“ProbOS provides local foundations for a future Nooplex Core.”
D	Emergence Research	`research`	#1139 staged research planning; complete evidence program pending	“ProbOS is an experimental platform for testing the Nooplex hypothesis.”

2026-09-07 planning update: the Captain requested roadmap/AD/issue follow-through. The planning record assigns bounded owners, not implementation admission or passing evidence. Readiness statuses and permitted claims above remain unchanged; no tier has advanced through document edits.

Tier A — Dependable Cognitive Mesh¶

Tier A establishes one ProbOS vessel as a supported Cognitive Mesh: heterogeneous agents, shared memory, governed action, durable work, learning loops, human participation, recovery, and truthful operational evidence.

Required evidence

AD-1185 defines and boots the supported profile.
AD-1270a distinguishes present, configured, advertised, activated, exercised, and healthy.
Every active Tier A P0 seam in the canonical manifest crosses its real producer and consumer.
Runtime, startup, configuration, and cognitive orchestration retain compatibility while ownership moves behind bounded facades.
Every durable store declares lifecycle, criticality, retention, backup, and restore disposition through AD-1256.
AD-1265/1266 pass snapshot -> verify -> restore -> reboot -> production-read.
P0 paths carry correlated run/tool evidence, latency, errors, and bounded resource measurements through AD-1152 and AD-1195 or equivalent owners.
Supported capabilities have no unresolved Critical or High security/governance defect; otherwise they report unavailable or degraded.
AD-1186 Ship Trials are non-vacuous: skipped, absent-data, and judge-unavailable outcomes are inconclusive or blocking, never passing evidence.
AD-1270g keeps product claims synchronized with executable authorities.

Exit claim

ProbOS is a supported implementation of one governed Cognitive Mesh.

This claim does not imply authenticated federation, a Global Knowledge Fabric, or evidence of emergent general intelligence.

Tier B — Secure Multi-Mesh¶

Tier B connects independently running Cognitive Meshes without shared process memory, filesystem, or database authority.

Existing owners

AD-1196: bind did:probos identities to Ed25519 keys with rotation and revocation.
AD-1197: sign canonical federation envelopes and prevent replay.
AD-1198: authenticate transport admission and prove behavior on the existing two-node cluster.
AD-703: distribute fleet policy; its version/conflict semantics must be settled before Tier B is supported.
AD-1152 and AD-1195: preserve correlation and durable event meaning across the boundary.

Required evidence

Two separately started nodes communicate over real sockets with no shared storage.
Source, target, topic, body, and attachment references are authenticated and tamper-evident.
Duplicate, replayed, stale-key, malicious-peer, partition, restart, and clock-skew cases are exercised.
Policy version mismatch fails closed for safety-critical operations.
Cross-node traces preserve causality and bounded provenance.
Latency/error/resource SLOs are measured for local and federated paths.
Fault injection proves bounded degradation and recovery.

Exit claim

ProbOS supports an authenticated federation of sovereign Cognitive Meshes.

Use “federated ProbOS cluster” or “multi-mesh substrate” before this gate. Do not call transport connectivity alone a Nooplex.

Tier C — Nooplex Core Fabric¶

Tier C implements the paper's Core Fabric rather than only forwarding intents between nodes.

Owner status: AD-1303 / #1356 owns a planned, bounded human-claim -> cross-mesh reuse -> correction pilot; AD-1304 / #1357 owns its independent-peer conformance and sovereign-exit evidence. Human claims retain AD-1199/#1136; local derived-knowledge reassessment is AD-1302/#1355. Neither the pilot nor the local ontology, graph, memory and trust prerequisites substitute for the full required architecture below. Broader Core Fabric work remains unscheduled until bounded follow-on decisions are admitted.

Required architecture

Global ontology with explicit pluralism, versioning, deprecation, migration, and conflict review.
Cross-mesh schema registry and compatibility negotiation.
Federated semantic query with routing, score normalization, partial-result semantics, and provenance.
Embedding-space alignment or an evidence-backed alternative that preserves semantic meaning across heterogeneous meshes.
Multi-mesh knowledge graph with trust-weighted integration, quarantine, retraction, and lineage.
Model-of-models for mesh capability, health, latency, cost, and routing fitness.
Cross-mesh governance with policy reconciliation, human override, and independent watchdog evidence.
Byzantine/Sybil/poisoning controls at knowledge-ingestion boundaries.
Privacy, confidentiality, data-residency, and provenance-minimization contracts.
Human and agent knowledge entering the same governed lifecycle, including validation, deprecation, archival, retraction, and deletion.

Exit claim

ProbOS implements a prototype Nooplex Core Fabric across multiple sovereign meshes.

This is the earliest tier at which the unqualified architectural term “Nooplex” is appropriate. It still does not demonstrate emergence.

Tier D — Emergence Research¶

Tier D tests the paper's falsifiable research hypothesis. It is not a release milestone and must permit disconfirmation.

Required research program

Use the refreshed #1139 as the staged cooperation-research planning owner. AD-1188/#1125 remains retired; AD-1231's hybrid-coordination decision is not reopened.
Compare the full system, independent-mesh ensemble, and predeclared ablations under identical models, tasks, budgets, and evidence policies.
Measure cross-domain synthesis against an ensemble baseline.
Replace or clearly separate vessel-local collaboration proxies from the paper's cross-mesh $TC_N$ measure.
Detect novel coordination against a frozen initial-protocol catalog.
Track held-out capability growth and accumulated knowledge longitudinally for at least six months.
Report performance per dollar, latency, energy/resource use, and coordination overhead.
Use expert human panels with predeclared rubrics and inter-rater reliability.
Publish confidence intervals, effect sizes, multiple-comparison corrections, and disconfirmation outcomes.

The paper's planetary-scale profile remains a long-horizon condition: at least 50 meshes, 500 active agents, three regions, 100,000 cross-mesh knowledge operations per day, and sustained federated memory at the stated scale. Smaller experiments may test early effects but cannot claim planetary validation.

Permitted claim before evidence

ProbOS is a research platform for evaluating whether governed federated cognition produces measurable emergent capabilities.

Permitted claim after evidence

The tested ProbOS federation showed evidence for specific predeclared emergence criteria under the published conditions.

Never generalize one successful demonstration into “ProbOS achieved AGI” or “the Nooplex hypothesis is proven.”

Cross-Cutting Evidence Programs¶

These do not form a fifth tier. They supply evidence required by more than one tier.

Program	Existing owner	Readiness role
Supported configuration	AD-1185 / #1121	Defines what is actually supported
Ship Trials	AD-1186 / #1123	Release evidence for Tier A and later tiers
Cooperation research	#1139; AD-1188 remains retired	Local and staged cross-mesh hypotheses, baselines and claim discipline
Event persistence	AD-1195 / #1132	Defines what historical claims event storage can answer
Trace correlation	AD-1152 / #1079	Connects ingress, reasoning, tools, artifacts, and outcomes
Agentic risk/cost bounds	AD-1208 / #1154	Begins resource governance for individual turns
Secure federation	AD-1196–1198 / #1140	Tier B identity, envelopes, and admission
Fleet policy	AD-703 / #479	Tier B/C policy propagation; conflict semantics remain required
Store lifecycle/recovery	AD-1256, AD-1265, AD-1266	Tier A durability and Tier C storage foundation
Platform maturity	AD-1270 / #1324	Tier A evidence and maintainability
Goal validity and external effects	AD-1300 / #1353, AD-1301 / #1354	Contracts before broader unattended consequential missions
Human claims and derived reassessment	AD-1199 / #1136, AD-1302 / #1355	Knowledge contribution, correction and enrolled downstream consumers
Core Fabric pilot and peer conformance	AD-1303 / #1356, AD-1304 / #1357	Bounded Tier C experiment, not full-tier completion

Remaining Program Gaps¶

Do not file one issue per bullet. Reuse the bounded planning owners below; when additional work is scheduled, keep independent findings actionable without duplicating existing epics. Planning ownership is not evidence of implementation.

Core Fabric beyond the pilot: AD-1303/1304 own the first bounded crossing. Full global ontology/schema evolution, heterogeneous alignment, global knowledge graph, model-of-models and broader cross-mesh governance still need bounded follow-on contracts.
Operational Science: platform SLOs, performance/cost accounting, load shedding, chaos engineering, and concrete shadow deployment/rollback.
Knowledge Lifecycle and Human Contribution: AD-1199 owns human claims and AD-1302 owns enrolled derivation reassessment. A complete lifecycle across every substrate, citation effects, archival, erasure and rights-aware provenance remains broader than those contracts.
Federation Adversarial Assurance: Byzantine knowledge validation, Sybil resistance beyond key possession, poisoning quarantine, privacy-preserving cross-mesh cognition, and data-sovereignty enforcement.
Emergence Evaluation: #1139 owns staged planning, with protected/representative release trials under #1123. Complete multi-mesh, cross-mesh $TC_N$, novel-coordination, longitudinal and human-panel evidence is still pending and is not implied by a pilot result.

Update Rules¶

Update this file only when a readiness gate, owner, or permitted claim changes. Routine implementation detail belongs in the owning AD or issue.
A tier changes status only when its required evidence artifact is committed and linked.
Every status change records the commit, supported profile, evidence artifact, and residual risks.
GitHub issues track executable work. Do not create issues merely to mirror this table.
Use one epic per active tier/program. Existing owners remain authoritative: #1324 for Tier A, #1140 for Tier B, #1356 for the bounded Tier C pilot, and #1139 for staged cooperation research.
The 2026-09-07 request schedules the bounded Tier C/D planning recorded above, not live experiments or the entire planetary program. Additional epics require separately scheduled scope.
AD-1270g should generate the readiness summary exposed in README; this full document remains human-reviewed architecture narrative.
Commercial documentation links to this status rather than duplicating it.

Current Assessment¶

As of the 2026-09-07 planning reconciliation (not a new live-system measurement):

ProbOS contains unusually advanced local governance, memory, trust, learning, self-modification, durable crew, and HXI mechanisms.
AD-1270 is planned but not yet complete, so Tier A remains planned, not supported.
Federation transport exists, but authenticated multi-node admission and real-socket evidence remain open, so Tier B remains planned with an experimental substrate.
Tier C has strong local prerequisites and a planned integrated pilot with independent conformance; implementation/evidence has not advanced its readiness status.
Tier D has useful local instrumentation, a narrow Sigma ablation and a refreshed research owner, but no complete multi-mesh, longitudinal, statistically governed evidence program has passed its gates.

ProbOS therefore appears capable of evolving toward the Nooplex without a foundational rewrite. The next architectural step is to prove the Cognitive Mesh unit cell, then federate sovereign units securely, then build the Core Fabric as a new layer rather than enlarging ProbOSRuntime into a planetary controller.