Machine-assisted exploration
Components and authorised agents observe, preserve, translate, compare, relate and propose.
observe → translate → relate → proposeOUTPUTEvidence-backed proposal · non-authoritativeIntertween reference library Documentation, sources and later decisions →
HOW THE protocol PROTOCOL WORKS · 3 MIN
Understand the protocol in three minutes.
INTERTWEEN / protocol PROTOCOL
Turn knowledge into a governed object, make it understandable across contexts and preserve what happens when it is reused. Start with Wrapper, Mapper and Tracker. Explore the extended roles when you need more.
THE THREE-MINUTE PATH
Preserve. Translate. Remember.
Three roles create the core learning path. Knowledge keeps its source, becomes understandable in another context and returns with evidence of use, contribution and outcome.
01 / PRESERVE
<Wrapper>
What do we know?
Creates a governed Knowledge Object with source, context, rights and identity intact.02 / TRANSLATE<Mapper>
What does it mean here?
Makes knowledge understandable elsewhere without erasing local language or differences.03 / REMEMBER<Tracker>
What happened when it was used?
Traces contribution, reuse, change, outcomes and the learning returned.Source → Knowledge Object → New context → Reuse → Outcome → LearningWhat becomes possible next →
Fast exploration inside deliberate institutional learning.
Intertween connects two loops with different speeds and authority. They share evidence and provenance, but they do not share the power to adopt a decision.
Human and institutional learning
Eligible people and organisations review evidence, contest interpretations, decide, adopt responsibility, change work and assess consequences.
review → decide → adopt → change → assessOUTPUTAccountable action · tracked outcome · revised knowledgeTHE HUMAN-GOVERNED SEAM AI and components may surface a candidate. Adoption requires an authorised decision. AI proposes. People and institutions adopt. Provenance explains.
These are nested institutional learning loops, not reinforcement learning and not a claim of automatic convergence.
Read the formal adoption boundary →Agent-like in purpose. Explicit in accountability.
The components are not personalities and do not have to be autonomous. Each receives defined inputs, performs one kind of work, produces reviewable outputs and stops at a clear boundary. Implementations may change; the contract remains inspectable.
Seven roles. Three deployment responsibilities.
Client, Runtime and Gateway describe where responsibilities may be deployed in a protocol node. They do not add three more components or prescribe one product architecture.
<Client>
Presents a situated Agora view and lets people, applications and agents propose, inspect, request, validate and reuse governed knowledge.
USES · Wrapper · Finder · Assembler · Cultivator<Runtime>
Maintains object identity, representations, relations, indexes, validation state, local mappings and accountable event history.
USES · Wrapper · Mapper · Tracker · Assembler<Gateway>
Authenticates participants and enforces purpose, rights, projection, trust, exchange receipts, revocation and returned-learning terms.
USES · Mediator · TrackerComponents define the work. Client, Runtime and Gateway describe an implementation boundary. One service may contain all three, or existing systems may provide them separately.
[COMPONENT]bounded architectural unit
[ROLE]responsibility in the lifecycle
[CAPABILITY]what the role enables
[TASK]concrete work performed
[OPERATION]technical action invoked
[WORKFLOW]coordination across roles
Components take roles. Roles provide capabilities. Capabilities perform tasks. Tasks invoke operations. Workflows coordinate them.
The task is intertweening. The outcome is an intertween.
Wrapper makes objects addressable. Mapper makes contexts relatable. Tracker makes relations observable over time. Finder and Cultivator detect, propose, validate and maintain emergent relational knowledge. When materialised and governed, that knowledge becomes an intertween with its own identity, evidence and lifecycle.
Understand the Intertween →The core in three minutes. Four roles when you need more.
Wrapper, Mapper and Tracker define the current normative adoption core. Finder, Assembler, Cultivator and Mediator are informative roadmap roles whose contracts remain open to pilot evidence and governance design.
Wrapper
[PURPOSE] Protect source truthJOB TO BE DONE
“When information is represented through the protocol, preserve exactly what the source provided, declare its semantic role and optionally identify the persistent subject it concerns.”
[CAPABILITIES]
source captureintegrity preservationobject typingentity reference[TASKS]
capture payloaddeclare object typeattach about relationretain source identitycreate preservation envelope[OPERATIONS]
fetch()hash()wrap()linkEntity()recordCustody()[INPUT]
Source payload · source ID · optional EntityReference · observation time
[OUTPUT]
Typed Knowledge Object · source observation · content fingerprint
!BOUNDARY
Never rewrites source truth or treats the representation as identical to the entity. Entity references remain optional for simple objects.
{DONE_WHEN}
Payload, semantic role, identity, timestamp, hash and custody chain are recorded.
What did the source give us, and what does it concern?
Mapper
[PURPOSE] Make local meaning portableJOB TO BE DONE
“When local knowledge must be understood elsewhere, map schema, concepts, identity, participation roles, granularity and authority without erasing local language or forcing a universal record.”
[CAPABILITIES]
schema translationvocabulary alignmentidentity resolutionparticipation-role mappinggranularity mappingmapping provenance[TASKS]
read local fieldscompare identifiersmap contextual rolesexpress equivalence or differenceflag unresolved termsrecord reversible mappings[OPERATIONS]
resolveField()resolveIdentity()mapParticipationRole()mapGranularity()validateMapping()explainMapping()[INPUT]
Knowledge Objects · local EntityReferences · Participations · mapping profile · vocabularies
[OUTPUT]
Mapped properties · provisional identity relations · contextual role mappings · field provenance · unresolved mappings
!BOUNDARY
Mappings are declarative, reversible, evidence-backed and contestable. Similarity must not become silent merging, and a local role must not become a permanent entity type.
{DONE_WHEN}
Every transformed field, identity relation and participation role can explain source → rule → evidence → result.
What does it mean here, and how might the identities and roles relate?
Tracker
[PURPOSE] Give knowledge memoryJOB TO BE DONE
“When knowledge is created, resolved, used, disputed or changed, preserve an attributable event and consequence history so the full lifecycle remains reconstructable.”
[CAPABILITIES]
change detectionevent historyprovenance trackingconsequence tracking[TASKS]
compare observationsappend lifecycle or composition eventrecord use and outcomelink versionstrack Intertween lifecycle[OPERATIONS]
diff()appendEvent()linkVersion()recordReuse()recordOutcome()[INPUT]
Observations · assertions · participations · Knowledger resolutions · Intertweens · validation · policy decisions · reuse and outcome signals
[OUTPUT]
Relational Trace events · proposed integrity checkpoints · state transitions · contribution and consequence evidence
!BOUNDARY
Not observed does not mean deleted. Tracker appends attributable events to the local Relational Trace. The proposed integrity profile permits independently retained proofs; this is neither a central database nor a universal account of truth.
{DONE_WHEN}
Every material change, participation, resolution, use and consequence has a time, context, cause, source, authority and eligible responsible party where applicable.
What changed, what was done, and what resulted?
Finder
[PURPOSE] Discover possible connectionsJOB TO BE DONE
“When objects can be compared, find candidate relationships, situated participations and emergent phenomena that no single source contains.”
[CAPABILITIES]
entity matchingparticipation detectionrelation discoveryIntertween detectionevidence collection[TASKS]
compare identifiersassess contextual roles and effectsgather evidencepropose participation, relation or Intertween[OPERATIONS]
matchIdentifier()proposeParticipation()calculateConfidence()collectEvidence()proposeIntertween()[INPUT]
Knowledge Objects · EntityReferences · contexts · classifications · history
[OUTPUT]
#ParticipationProfile, #RelationAssertion or #IntertweenProfile {proposed} · evidence · confidence
!BOUNDARY
Never assigns permanent actorship, silently changes a source object or presents a human or machine detection as authoritative truth.
{DONE_WHEN}
Each candidate identifies its constituents, context, role or relation, observed effect, emergence method, evidence, confidence and state.
What became connected, and what made a difference?
Assembler
[PURPOSE] Compose governed knowledgeJOB TO BE DONE
“When evidence is fragmented, emit an immutable Knowledger manifest and resolve a neutral, rights-filtered Current Knowledge State that remains traceable to its members.”
[CAPABILITIES]
membership compositionmanifest versioningcurrent-state projection[TASKS]
assert membership rolesresolve member versionsemit manifestrecord inclusion and exclusion[OPERATIONS]
assertMembership()resolveMembers()emitManifest()projectCurrentState()[INPUT]
Knowledge Objects · EntityReferences · Relation Assertions · Intertweens · history · policy · validation
[OUTPUT]
#KnowledgerProfile manifest · Current Knowledge State projection · resolution evidence
!BOUNDARY
Assembly is intentional composition, not emergence. Membership transfers neither ownership nor truth; a projection never becomes a new source of truth.
{DONE_WHEN}
The immutable manifest and every resolved view identify member versions, policy decision and evidence basis.
What belongs together—and what may this viewer see now?
Cultivator
[PURPOSE] Turn use into governed learningJOB TO BE DONE
“When use, corrections and outcomes repeat, detect the pattern and create a reviewable proposal for institutional learning.”
[CAPABILITIES]
pattern detectionfeedback synthesisIntertween validationgoverned improvement[TASKS]
aggregate learning signalsdetect recurring patternsform Intertween or improvement proposalroute for validation[OPERATIONS]
aggregateSignals()detectPattern()proposeIntertween()requestValidation()[INPUT]
Tracked feedback · corrections · validation · outcomes · candidate Intertweens
[OUTPUT]
#IntertweenProfile · #LearningSignal · #CorrectionProposal · #MappingProposal · #ConfidenceUpdate
!BOUNDARY
An emergent pattern remains a non-authoritative proposal until governed validation promotes it. Contestation and supersession remain part of its history.
{DONE_WHEN}
The proposal is traceable, clearly labelled and ready for human or authorised validation.
What have we learned?
Mediator
[PURPOSE] Govern circulationJOB TO BE DONE
“When knowledge crosses contexts, decide what may move, to whom, for which purpose and under which conditions.”
[CAPABILITIES]
policy negotiationprojection controlfederated exchange[TASKS]
authenticate requesterevaluate purpose and rightsconstruct permitted projectionrecord exchange[OPERATIONS]
authenticate()evaluatePolicy()maskFields()authoriseExchange()[INPUT]
Identity · request · purpose · rights · access and learning profiles
[OUTPUT]
Policy decision · masked projection · audit event · feedback contract
!BOUNDARY
Access, action and learning are separate permissions. Federation is negotiated, not merely connected.
{DONE_WHEN}
Every exchange has explicit scope, terms, projection and audit trail.
What may circulate?
How people use the core.
These are possible products and experience layers built from the seven components. They are not additional core components and may be combined, replaced or implemented differently.
Components perform the knowledge work. Tools implement and coordinate it. Interfaces make the results understandable. Features describe what people can do.
[Explorer]
Navigate the Agora through a situated view of permitted relationships, evidence, timelines and projections.
USESFinder · Tracker · Assembler · Mediator
[Workbench]
Create, map, review and assemble Knowledge Objects.
USESWrapper · Mapper · Assembler
[Validator]
Test objects and profiles against schemas, rules and conformance requirements.
USESWrapper · Mapper · Mediator
[Node Console]
Manage local policies, trust, federation and operational state.
USESTracker · Mediator
[CLI / SDK / API]
Integrate protocol roles into existing systems and workflows.
USESAll seven components
EXPERIENCE FEATURES
Situated Agora viewKnowledge mapProvenance timelineEvidence inspectionObject comparisonRights and projection previewValidation resultsCitation and governed export
<Finder> makes candidate connections discoverable. [Explorer] makes permitted results understandable.
Three native functions across the roles.
These are not separate products and do not expand the stable core to eight components. They clarify how the existing roles support cross-boundary accountability and value.
Custody and governance
Keeps an observation visible, evidence-linked and routed to the mandates it touches. Custody never silently becomes ownership of the matter.
Attribution
Records who contributed, derived, validated, reused or corrected each object so credit and lineage travel across reuse.
Reciprocity
Records what value or contribution flows back on a reuse event. This is outside Level 0 and may be implemented inside a closed member network.
Knowledge
Federation