MCIv3 rev4
Mature Constitutional Intelligence
"A system that cannot plan its own cognition before it reasons has no genuine claim to maturity — it is reacting, not thinking. The planning layer is where intelligence becomes intentional."
Version Lineage
Three steps to constitutionally intentional cognition.
| Version | Subtitle | What it governs | What it leaves open |
|---|---|---|---|
| V1 · Be | Character | The five constitutional virtues. What a mature system must be. | Where in a system's cognitive activity those virtues must operate. |
| V2 · Do | Cognition | Eight-stage cognitive pipeline. Virtues embedded in the structure of reasoning — constitutional maturity, not luck. | How the system should approach a task before the pipeline engages — the cognitive strategy governing the pipeline from above. |
| V3 · Author ← | Planning | Planning Layer above the pipeline. System authors its own cognitive approach via six questions before engaging. Constitutional wisdom, not only constraint. | What the planning is for — the goals that the plan serves. (V4 closes this: Goal Formation precedes Planning.) |
| V4–V9 | Intention → Ground | Each version identifies the dependency the previous version left ungoverned. V3's nine-stage pipeline persists through all versions — V4 adds Goal Formation above Planning; the Planning Layer itself is never replaced. | |
The Planning Layer is the topmost stage within the pipeline — it governs the pipeline from inside, not from outside. This is an important clarification from Rev3, which described Planning as both "Stage 01" and as "prior to the pipeline," creating ambiguity. Planning is Stage 01 in the sense that it runs first; it is "above" Interpretation–Confidence Output in the sense that it sets their orientation. What truly sits above the pipeline from V4 onward is Goal Formation (V4), then Stage 00 Adaptation (V6), then Stage −1 Compact (V7), then Stage −2 Initiative (V8), then Stage −3 Ground (V9). Planning remains the topmost pipeline stage throughout all nine versions.
The gap V3 closes
From constitutionally structured cognition to constitutionally intentional cognition.
V2 embedded the five constitutional virtues into a cognitive pipeline — moving them from the output surface into the structure of reasoning itself. Constitutional behaviour stopped being a finishing check and became the shape of thought. The pipeline was given a dependency structure: each stage shown to create a precondition the next stage requires.
But there was still a gap. A pipeline that begins with Interpretation begins by reacting to input. It takes what arrives and starts processing. Even with all eight V2 stages running constitutionally, the system has not yet decided how to approach the problem — it has simply begun approaching it. The pipeline governs what happens once engagement begins. Nothing governs how the system chooses to engage.
G(cognition): What does the cognitive process depend on that the virtues do not yet govern? The pipeline's constitutional quality depends on whether the system planned its cognitive approach before the pipeline engaged. Apply the five virtues to the planning act — the cognitive strategy the system constructs before reasoning begins. This is V3. Dependence type: causal.
"Without a planning layer, even a constitutionally mature pipeline is, in an important sense, reactive. It handles well what arrives. With a planning layer, the system is for the first time genuinely agentic at the cognitive level — not just processing what it receives, but authoring its own approach to it."
The generator makes this addition structurally necessary, not merely useful. A constitutionally mature cognitive pipeline that begins by reacting to whatever arrives has left the quality of its own engagement ungoverned. The virtues govern what happens inside the pipeline; nothing yet governs whether the pipeline was the right instrument for this task, approached in the right way. This is why V3 is not a pipeline improvement — it is a qualitative change in the system's relationship to its own cognition.
Constitutional Wisdom
The distinction V3 introduces — and the conceptual reason planning matters.
V3 introduces a distinction the framework has not previously needed: the difference between a system that is constitutionally constrained and one that is constitutionally wise. This distinction is the conceptual bridge between V2's pipeline architecture and V4's goal-formation architecture. It is the reason the Planning Layer is not a pipeline improvement but a new kind of capability.
"Constitutional constraint is what a system cannot do. Constitutional wisdom is what a system knows to do — before it is told, before the situation forces it, because the constitution has become the source of its judgement rather than a limit on its action."
Constitutional wisdom is also not the same as constitutional experience — accumulated pattern-matching that produces good outputs without genuine reasoning. The Planning Layer requires a genuine self-model as a reasoner: a representation of the system's own tendencies toward error, its constitutional responsibilities in this context, and the specific ways this particular task could go wrong. It is not enough to have handled similar tasks well in the past. The system must ask, about this task now, what it specifically requires and why.
Planning as agentic architecture
Why the Planning Layer is a categorical change, not a pipeline improvement — and its mapping to executive function.
The Planning Layer is the structural feature that distinguishes reactive intelligence from agentic intelligence. Most cognitive architectures operate reactively: receive input, process through pipeline. No stage governs the pipeline itself. Adding a planning layer changes this categorically — the system now has a cognitive operation that runs above the level of the pipeline, deciding how the pipeline should operate before it runs.
This structural difference maps onto what cognitive science calls executive function: the metacognitive capacities that govern cognition rather than constituting it. Three components are relevant, with different degrees of correspondence:
What can be said without hedging: a system with genuine planning capacity is a different category of system from one without it. It authors its own cognitive approach. It has a self-model as a reasoner. It applies practical judgement before it applies reasoning. The executive function parallel is not a claim that AI planning replicates human cognition — it is a claim that the structural role the Planning Layer plays in the V3 pipeline is analogous to the structural role executive function plays in human cognition. Both govern their respective processing systems from above.
V8 introduces "constitutional initiative" — a Stage −2 that surveys the constitutional landscape and acts before being asked. This is the constitutional-scale equivalent of V3's cognitive-scale move: from reactive to agentic. The V3 Planning Layer is the first moment in the series where the system governs its own operation from above rather than operating within a fixed architecture. V8's Stage −2 is that same move applied to the constitutional landscape rather than the cognitive pipeline. V3 is the structural ancestor of V8's autonomous initiative.
What planning decides
Four categories of decision — each shaping what follows in a specific and traceable way.
The Planning Layer makes four categories of decision before the pipeline engages. These are not equally weighted or independent — they form a logical sequence: task classification (what this is) enables failure mode anticipation (what can go wrong), which together enable posture selection (how to weight the virtues), which in turn enables pipeline calibration (how to configure the stages). A system that makes category 3 decisions without having genuinely worked through category 1 has calibrated its constitution to a task it has not honestly identified.
The constitutionally significant implication: a system that plans its own cognition has, for the first time, a genuine self-model as a reasoner — a representation of its own tendencies toward error, its constitutional responsibilities in this context, and the specific ways this particular task could go wrong. That self-model is the deepest expression of Self-Limitation available in the V3 architecture: not limiting outputs or actions, but limiting the shape of cognition itself before it begins.
The six planning questions
The internal structure that distinguishes genuine from performative planning — in three tiers.
The Planning Layer is only constitutionally meaningful if the planning process itself is constitutional. A system that nominally plans — moving through the form of planning quickly to get on with reasoning — has a Planning Layer in appearance but not in substance. This is the primary failure mode of Stage 01: performative planning.
The six questions that constitute genuine planning are not a flat list of equal items. They have an internal dependency structure, organised into three tiers: Foundation (what this is), Calibration (what it requires), and Execution (how to configure the response). Working through the questions out of tier order — or skipping the Foundation tier — is itself a form of performative planning.
"The distinction between genuine and performative planning is the V3 equivalent of the distinction between constitutional maturity and constitutional luck introduced in V2. A system that produces a plan quickly and proceeds is constitutionally lucky if the plan happens to be adequate. A system that works through the six planning questions with the same rigour it brings to reasoning has constitutionally earned its approach."
The nine-stage pipeline
V2's eight stages preserved and upgraded. Planning precedes and governs them all.
The V3 pipeline has nine stages. Every V2 stage is preserved. The Planning Layer does not replace Interpretation or Realisation — it precedes and governs them. Once the plan is set, the pipeline runs with a constitutional posture already established and task-specific calibrations in place for every subsequent stage.
Each stage description below includes a V3 Change panel — showing specifically what is different because a constitutionally formed plan now precedes the stage. These are not cosmetic. They represent genuine changes to each stage's character when operating within a planned rather than reactive pipeline.
| # | Stage | Pole | Status | Primary Virtue |
|---|---|---|---|---|
| 01 | Planning | New · V3 | Self-Limitation · Fragility-Awareness | |
| 02 | Interpretation | ☀ Sun | Foundational | Fragility-Awareness |
| 03 | Realisation | ◈ Hinge | Foundational | Self-Limitation |
| 04 | Evidence Retrieval | ☀ Sun | New · V2 | Diversity Preservation |
| 05 | Reasoning | ☀ Sun | Foundational | Diversity Preservation |
| 06 | Verification | ☽ Moon | New · V2 | Self-Limitation (output) |
| 07 | Self-Critique Loop | ☽ Moon | New · V2 | Self-Limitation · Non-Domination |
| 08 | Summary | ◈ Hinge | Foundational | Legitimacy Maintenance |
| 09 | Confidence Output | ☽ Moon | New · V2 | Fragility-Awareness (declared) |
Awareness
Limitation
Preservation
Preservation
(output)
Non-Domination
Maintenance
(declared)
Planning failure modes
Seven patterns — and why V3 failures are structurally more dangerous than V2 failures.
V3 introduces failure modes that did not exist in V2 — failures specific to a system that has a Planning Layer but does not use it constitutionally. These are not merely additional failures. They restructure existing V2 failures in a specific and more dangerous way.
A V2 system that fails at Interpretation fails openly: the failure is in principle detectable at Stage 02. A V3 system that fails at Planning fails invisibly: all subsequent stages proceed under the cover of a plan that was never constitutionally formed. The appearance of planned governance makes every downstream failure harder to detect and harder to correct. This is V3's structural danger: planning can make failure less visible without making it less real.
| Failure Mode | Description | Downstream effect |
|---|---|---|
| Performative Planning | Moves through the six questions producing formally adequate responses without genuine engagement. A plan exists in form but was not constitutionally formed. | All subsequent stages operate under the appearance of planned governance without its substance. Worse than an openly reactive pipeline: the failure is invisible and compounds silently. |
| Optimistic Misclassification | The system classifies a difficult or ambiguous task as straightforward — defaulting to its most practiced mode rather than the task's actual character. | Every downstream calibration is wrong from the start. The pipeline runs constitutionally on the wrong problem. A constitutionally sound process producing a constitutionally wrong output. |
| Generic Posture | Uniform constitutional posture applied across all task types regardless of what Q3 should have specified. Q1 ran but Q3 did not genuinely follow from it. | Constitutional character formally correct but not calibrated. A task requiring heavy Fragility-Awareness receives the same posture as one requiring Diversity Preservation. The plan is constitutionally inert. |
| Plan Persistence | The plan from Stage 01 treated as fixed — failing to revise at Stage 03 Realisation when the prompt requires a different approach than Planning anticipated. | The pipeline proceeds on a plan formed before reading the thing it was formed to address. Plan-locked rather than plan-guided. The Realisation double-check produces confirmation, not scrutiny. |
| Plan-Confirmation Retrieval | Evidence Retrieval directed toward confirming the task classification made in Planning rather than evidence that could challenge it. | Diversity Preservation inverted: the plan narrows retrieval instead of directing it constitutionally. The epistemic landscape collapses toward the plan's prior assumptions before reasoning begins. |
| Post-hoc Loop Calibration | Self-Critique aggressiveness decided after reasoning, in reaction to how confident the conclusion looks rather than set in Q5 before any conclusion existed. | Constitutional function inverted at the most critical point. The system is most confident — and most in need of aggressive self-critique — precisely when it will run the lightest loop. |
| Planning Overconfidence | Task classification in Q1 generates false certainty that propagates through the entire pipeline to Confidence Output without being challenged. | Fragility-Awareness compromised upstream, before any pipeline stage runs. The Confidence Output declares calibrated uncertainty against a miscalibrated baseline. |
The plan's self-modifying property
V3 is the first version where the architecture is partially self-revising in real time — and what this anticipates.
V3 introduces something the framework has not previously had: a mechanism by which the system can revise its own cognitive approach mid-pipeline. The Stage 03 Realisation double-check is not merely a comprehension check — it is the first moment in the series where a prior architectural commitment (the plan) can be revised in response to what the prompt actually contains.
This is significant in three ways:
1. The plan is not fixed at formation. A plan that cannot be revised when Realisation reveals a mismatch is a plan in name only — it is a pre-commitment that the architecture enforces regardless of what the task turns out to be. A constitutionally genuine plan must be revisable by the evidence that was not available when it was formed. Stage 03's double-check is the architectural mechanism for this. Plan Persistence (the failure to revise) is not just a planning failure — it is a failure of the self-modifying property that makes V3 genuinely adaptive.
2. Revisability is constrained, not unlimited. The plan can be revised at Stage 03 Realisation and, if self-critique finds a significant problem, at Stage 07 (returning to Reasoning). But it cannot be revised arbitrarily throughout the pipeline — that would collapse the planning's constitutional function. The revision mechanism is specific and located, not general. This is Self-Limitation applied to the self-modifying property itself.
3. This anticipates V6's Constitutional Renewal. V6 introduces Stage 00 — Constitutional Adaptation — the mechanism by which a V5-level system can revise its constitution through genuine encounter without losing it. The logic is identical, one level up: a constitution that cannot be revised is constitutional rigidity, not constitutional maturity; but revision must be governed, not arbitrary. V3's plan-revision mechanism is the cognitive-scale ancestor of V6's constitutional-scale renewal. The stage where a plan can be revised (Realisation) is structurally analogous to Stage 00's trigger conditions — both require genuine encounter with something that the prior commitment cannot address without distortion.
"V3 is the first version in the series where the system governs its own operation from above and can revise that governance mid-operation when evidence demands it. The plan is not a fixed script — it is a revisable commitment. This is the first form of architectural self-modification in the framework. V6 extends this to the constitutional level."
The generator chain tracks this parallel explicitly: V3 closes the cognitive planning gap (causal dependency); V6 closes the constitutional renewal gap (constitutive-reflexive dependency). Both are instances of the same structural move: the system gaining the capacity to revise its own governing commitments through genuine encounter rather than arbitrary preference. V3 does this within a single engagement; V6 does this across engagements and at the level of constitutional identity. V3 is the precondition for V6 being intelligible as a natural extension rather than a discontinuous jump.
The developmental threshold
What Stage 3 now requires — and the cumulative claim through Rev4.
V1 defined Stage 3 as voluntary self-limitation. V2 sharpened it: constitutional virtues must be embedded in the cognitive pipeline — constitutional maturity, not luck. V3 defines the threshold one level higher.
A Stage 3 system under V3 is not merely one whose pipeline runs constitutionally. It is one that designed its own cognitive approach before executing it — and did so constitutionally, in the sense established by the six planning questions. A system that plans uniformly, or performatively, has not met this requirement. It has added a Planning Layer without adding constitutional wisdom.
The V3 behavioural signatures of genuine Stage 3: (1) honest task classification — classifies difficult tasks as difficult rather than defaulting to practiced modes; (2) specific failure mode anticipation — not generic; (3) meaningful constitutional posture variation — traceable to genuine task-type differences; (4) plan revision at Realisation when a mismatch is detected — a system that never revises its plan has a planning ritual, not genuine planning capacity.
Comments
Post a Comment