TAKE BACK YOUR DATA
MACHINE · CICM · COGNITIVE INTELLIGENCE CORE MACHINE

An AI can provide proposals. Does that mean it may decide what counts as valid cognitive effect in your domain?

CICM is an SES machine type for explicitly bounded cognitive effect. A model, tool, human, or external service may observe, propose, compare, execute, or provide evidence. Normative meaning, rules, and the conditions for valid effect remain bound to the concrete operator instance.

Technical or probabilistic capability is not normative authority.

CICM is neither an AI model, an AI platform, nor an autonomous decision maker. The public C0 describes the machine type; it is not your concrete operator machine.

SOVEREIGNTY + FORMATION · A6 v1.2

Who operates? Who is sovereign? How is the instance formed?

Roles are bound separately. Operation, authority, sovereignty and formation are not the same thing.

1 · Operator

Concrete technical/organizational operator of the instance. Root sovereignty is separately bound as: Operator Sovereign; technical operation alone does not create it.

2 · Root sovereignty: O

  • Operator Sovereign: A model can answer persuasively, a human can make a recommendation, and a tool can rank candidates. None of that by itself means the output may count as a valid judgment, valid rule change, valid execution decision, or valid learning effect in the operator's effect space.

3 · Delegated sovereignty

No additional delegated sovereignty domain is claimed at C0 level.

4 · Authority-only roles

  • Delegated decision/analysis roles
  • AI/model/advisory carriers

5 · Target rule source

Target rules are defined and authorized in the machine’s own engineering/instance-binding path by its bound sovereign; no Factory authorship is claimed.

6 · Formation path

Self-build / Engineering

1 · WHAT PROBLEM DOES CICM ADDRESS?

CICM separates cognitive capability from valid cognitive effect.

A model can answer persuasively, a human can make a recommendation, and a tool can rank candidates. None of that by itself means the output may count as a valid judgment, valid rule change, valid execution decision, or valid learning effect in the operator's effect space.

CICM therefore binds every productive cognitive effect to exactly one Atomic Cognitive Claim, exactly one effect target, and exactly one immutable decision snapshot.

Judgment

one bound decision for exactly one cognitive claim

Rule candidate

an explicit, initially inactive rule proposal with provenance

Rule authorization / activation

two separate effects: authorize first, activate separately

Structure change

an authorized change to a governed cognitive structure

Execution decision

a prebound action or consciously governed non-action

Learning object

an explicit future-use object with authority and lineage; not hidden model memory

Revalidation

a new validity result after a material change

A productive CICM occurrence has exactly one primary effect class and exactly one PrimaryEffectResult. Multiple mixed effects may not be disguised as one PASS.
2 · WHO MAY DETERMINE?

A concrete CICM instance has exactly one Operator Sovereign.

The Operator Sovereign defines the meaning of goals, admissible rule classes, acceptance boundaries, and activation authority in the claimed cognitive effect space. Delegation is possible, but only when bound and without creating a second sovereign.

Operator Sovereign

defines goals, rule classes, and acceptance boundaries

Delegated Decision Authority

may make only prebound bounded judgments

Rule Authority

authorizes exact rule candidates

Evidence Authority

verifies evidence integrity and custody

Security Authority

may block, contain, and preserve evidence – but may not create positive cognitive validity

Carrier Authority

executes already valid technical instructions

A provider, model, administrator, reviewer, developer, or tool may not silently create new normative meaning, expand scope, or convert uncertainty into PASS.
3 · EFFECT SPACE AND SCOPE

CICM considers every path that can influence the claimed cognitive effect.

Scope is not a list of AI systems here either. It covers every effect-relevant path of the concrete claim.

Humans

inputs, manual intervention, delegations, and domain decisions

Models

model output, probabilistic proposals, and model-side state where they can influence effect

Retrieval and data

sources, data state, retrieval results, and bound revisions

Tools and processes

tool use, workflow, prompt context, time triggers, restore and replay paths

External systems

external services and provider outputs as inputs to the operator instance

Feedback

downstream feedback, memory/learning paths, and later state changes

BOUND

path is bound with identity, authority, state, evidence, and failure handling

COUNTERCLASSIFIED

proven not effect-relevant for the claimed effect

UNKNOWN

remains effect-relevant and prevents PASS

SHADOW

treated as an unclassified effect candidate and prevents productive progression

An unknown or shadow path is not tolerated residual risk inside a green claim. It prevents productive PASS until bound or validly counterclassified.
4 · THIRD SYSTEMS AND BLACK BOXES

A third system does not have to know CICM or SES.

An external system, model, or provider supplies input and receives the intended output. The operator instance determines what those inputs and outputs may mean for its own valid cognitive effect.

  • no SES-specific API, plugin, or handshake is required
  • no provider consent to the operator's normative order is required
  • provider output, a score, model answer, or retrieval result remains input first
  • black-box capability does not thereby become normative judgment
  • a third system can be effect-relevant and in scope without becoming sovereign or a cooperating SES participant
CICM does not physically control the provider. It controls what may count as valid cognitive effect inside the bound operator claim.
5 · HOW DOES VALID COGNITIVE EFFECT FORM?

CICM allows candidates to form – but forbids a free jump from candidate to valid result.

A productive occurrence works on one immutable snapshot. The machine binds context and authority, closes scope and evidence, detects conflicts, forms a complete finite candidate set, and compares it deterministically under active rules.

R1–R4

May the effect class occur, do required objects exist, and does current state permit evaluation?

R5–R7

Is the transition listed, are object/rule/authority/evidence bound, and is the effect inside authorized context?

R8–R10

Are identities/revisions integral, are rules/evidence coherent, and are all forbidden conditions absent?

R11

Which exact consequence follows from the completed evaluation?

If two valid outcomes remain and no active priority rule selects exactly one, the result is REQUIRE_DECISION – not “the AI chooses the most likely best one”.

Exactly one target-forming step may create the PrimaryEffectResult. Other cognitively capable steps may emit only CHECK_ONLY, NOT_APPLICABLE, or allowed SupportingTransitions.

The machine's validity decision is deterministic – the probabilistic carrier supplying candidates or inputs need not be.
6 · STATE, CONSEQUENCE, AND EVIDENCE

A valid result must be reconstructable as a bound effect path.

CICM binds not only an answer, but claim, scope, authority, source revision, context, rules, candidates, judgment, consequence, follow state, learning/revalidation objects, and the atomic occurrence identity.

claim and target identitycomplete scope classificationactive sovereign/delegation authoritybound source and context snapshotsexact active rule and meta-rule versionsconflict and candidate evidencedeterministic comparison and resultprecommit, attempt, receipt, and observed follow state when execution is relevantlearning lineage or revalidation result where relevantM01–M18 applicability and single-result proof
Evidence integrity does not automatically prove external truth. It proves that the machine used the exact bound evidence object it claims to have used.
7 · RELATION TO THE SEBG BASELINE

CICM is not a baseline machine, and `allow-as-before` is not a constitutive CICM rule.

For existing effect spaces, TBYD continues to recommend binding the real effect order first with a concrete SEBG baseline. If the operator then chooses to additionally govern a domain of explicitly bounded cognitive effect, CICM can serve as the domain-machine architecture.

Existing effect space

TBYD recommendation: SEBG baseline first; CICM afterwards only if the operator chooses to make this cognitive domain DOMAIN-RULED.

BASELINE-ONLY

The domain remains under the bound baseline and deliberately receives no CICM domain machine.

New effect space

CICM can directly serve as the machine type if scope, authority, and every later instance artifact are completely closed.

SEBG-first is a TBYD build recommendation for existing effect spaces. It is neither universal SES law nor part of the CICM C0.
8 · WHAT CHANGES IN PRACTICE?

The AI does not become more sovereign. Its effect becomes more tightly bound.

Model answer

often read as a “decision” → in CICM it begins as input/candidate; valid judgment requires the bound machine path

Rule proposal

does not become active automatically → candidate, authorization, and activation are separate effect occurrences

Non-determinism

is not hidden behind probability → REQUIRE_DECISION or another non-PASS state

Execution

a request is not treated as success → prebound consequence, attempt, receipt, and follow state remain separate

Learning

may not drift into future effect as hidden memory mutation → explicit learning object with authority and lineage

Change

does not retrospectively repair an old invalid effect → new snapshot, revalidation, and where required new machine identity

This effect order exists only in a concrete bound CICM instance. The public C0 alone changes no live operation.
9 · WHAT CICM DOES NOT CLAIM

CICM controls validity of cognitive effect – not the world and not the intelligence of a model.

no general intelligence or consciousness
no unrestricted autonomy
no guarantee of universal reasoning quality
no guarantee of factual truth
no automatic professional correctness
no legal validity, compliance, safety, or security certification
no economic-benefit guarantee
no human replacement
no C1, C2, C3, or C4 from the public C0
no PASS for a model or provider merely through participation
A probabilistic carrier may be capable, creative, or uncertain. CICM does not thereby give it free normative decision authority.
10 · WHO FORMS AND REVIEWS A CONCRETE CICM MACHINE?

The machine order is formed before model and carrier engineering.

1

Sovereign / Domain

defines effect space, goals, admissible rule classes, authority, and acceptance boundaries

2

Quality · Process · Systems Engineering

closes scope, rule space, state, failure, evidence, learning, and revalidation

3

Domain / Assurance

reviews domain meaning, risk boundaries, and audit/legal/safety context where relevant to the concrete claim

4

AI / Model / Technical Carrier Engineering

implements the already closed machine order or supplies bound cognitive/technical carrier functions

If the implementer, model, or prompt must determine a missing normative rule itself, the machine is not ready for implementation.
11 · PROFESSIONAL WORKING MATERIALS

For the concrete instance, use working material by review task – not a PDF collection.

Close scope and effect candidates

SES In-Scope Formation and Baseline Effect Binding for Engineers

Reviews technical, human, external, delayed, and hidden effect paths.

Route working material →

Review rule completeness

SES Effect Validity and Rule Completeness for Engineers

Reviews authority, rule, state, judgment, consequence, evidence, and binary validity.

Route working material →

Form the machine from C0 to C2

SES Machine Builder Method — Engineering and Quality Reader Edition

Moves from machine-type architecture to concrete artifact closure and machine binding.

Route working material →
Working materials support formation and review. They do not transfer PASS to your operator instance.
12 · C0 BOUNDARY

The public CICM C0 describes the machine-type architecture. The concrete MachineID is formed only at C2.

CICM C0 defines effect space, scope grammar, module order, authority, rule space, states, evidence, failure, and machine-type identity. The concrete operator instance must instantiate that architecture in C1-A through C1-I and then bind it into exactly one C2 machine.

C0

public machine architecture

C1

close concrete modules, scope, authority, rules, effects, states, evidence, failure, and revalidation

C2

bind exactly one no-bypass MachineID

Carrier

carry the closed machine technically

C3

prove activation readiness

C4

productively activate exactly that MachineID

C3 and C4 are explicitly NOT_CLAIMED by the CICM C0. A model, provider, or carrier inherits no CICM PASS through participation.
13 · VERSION AND INTEGRITY

Publication binding is closed separately and does not block understanding the machine.

The current CICM source is internally bound as C0 architecture PASS and its source-production quality is RELEASED. The website nevertheless publishes no CID or immutable publication PASS here yet: IPFS/CID bindings are being handled later as batch afterwork.

IPFS_PENDING · A7_NOT_YET_BOUND

Publication integrity is a separate claim. It proves neither C1/C2/C3/C4 nor semantic truth or productive operator operation.

14 · CICM IN EIGHT SENTENCES

If these eight statements remain stable, the machine is no longer being misread as an “AI control tool”.

  1. CICM is an SES machine type for explicitly bounded cognitive effect, not an AI model or AI platform.
  2. A concrete CICM instance has exactly one Operator Sovereign; models, providers, and tools gain no normative rule power from participation.
  3. Every productive occurrence binds one Atomic Cognitive Claim, one effect target, one immutable snapshot, and exactly one PrimaryEffectResult.
  4. Scope includes every technical, human, organizational, external, delayed, or hidden path that can influence cognitive effect.
  5. If multiple valid outcomes remain and no active priority rule decides, the machine does not guess; it returns REQUIRE_DECISION or another non-PASS state.
  6. Rule candidate, authorization, activation, execution decision, learning object, and revalidation are separate effects and may not silently collapse.
  7. CICM is not an SEBG baseline; for existing effect spaces, SEBG-first is a TBYD recommendation before the operator forms CICM as an additional domain machine.
  8. The public C0 is only the machine-type architecture; C1, C2, carrier, C3, and C4 remain concrete operator work.
15 · NEXT STEP

From here, return to operator formation, compare machines, or enter professional review.

Apply

Place CICM into the concrete operator path from claim and scope through C1/C2, carrier, C3, and C4.

Go to Apply →

Machines

Compare other public C0 reference architectures without treating them as transferable operator PASS.

Go to Machines →

Verify

Choose the appropriate Quality/Process/Systems working material for the concrete closure task.

Go to Verify →