SYSTEM FRICTION INSTITUTE · CANONICAL PUBLIC PROFILE

SYSTEM FRICTION INSTITUTE

An evidence-governed instrument for observing a signal inside a changing field, proposing a minimal intervention and learning from the documented difference between prediction and outcome.

← FIELD

DEFINITION

System Friction Institute makes visible the friction that systems learn to normalize.

OPERATING PRINCIPLE

Observation, evidence, inference, authorization, execution, return and memory remain distinguishable and traceable.

CANONICAL QUESTION

What structural configuration of a signal, under what world state and field conditions, produces particular patterns of propagation, persistence, transformation or disappearance?

INSTRUMENTS

MIHM

Marco Integrado de Homeostasis Multivariable

Structural measurement and friction/homeostasis conditions with explicit evidence and missing-data handling.

MOP-H

Minimal Observation and Perturbation Protocol

Governed minimal-intervention hypothesis, return preservation and later comparison against T0.

WorldSpect

World Spectrum Observation

Persist world-state snapshots, source health, confidence and longitudinal domain observations.

World Vector

Represent current domain values, tensions, dominant signals and observed context.

AMV

Adaptive Meta-Observer

Accumulate signal measurements, evidence, provisional predictions, interventions, outcomes and governed learning.

Atlas

Atlas / SFI Reference Bank

Normalize observable cases and connect evidence, prediction-at-T0, outcomes and model versions.

Cognitive Twin

Governed longitudinal reconstruction, interpretation and proposal layer; not autonomous institutional authority.

Method Lab

Governed experimental and simulation environment that preserves SIMULATED/DERIVED apart from OBSERVED/VERIFIED_CONTRAST.

OPERATING LIFECYCLE

01 · OBSERVATION02 · EVIDENCE03 · INFERENCE04 · HYPOTHESIS / PREDICTION05 · METHOD LAB / MINIMAL INTERVENTION06 · PROPOSAL07 · ROOT AUTHORIZATION08 · EXECUTION09 · RETURN / OUTCOME10 · REALITY CALIBRATION11 · GOVERNED MEMORY

EPISTEMIC INVARIANTS

MISSING remains MISSING; absence is not converted to zero.

MODEL OUTPUT is not OBSERVATION.

SIMULATION is not OBSERVATION.

RELATION is not CAUSALITY.

PUBLICATION or ENGAGEMENT is not validation.

GOVERNANCE is not truth.

Cognitive Twin proposal is not institutional authorization.

Canonical promotion requires the applicable governed authority.

AI-NATIVE ACCESS

Discovery: /api/external/v1/manifest
Console: /api/external/v1/console
Auth: X-SFI-Token or Bearer token with explicit scopes

External agents may operate only inside their granted scopes; ROOT approval and canonical promotion remain separate governed decisions.

PUBLIC SURFACES

/Live institutional observation threshold / FIELD /fieldLive geospatial and multiscale field /systemsSystem boundaries, relations, exchange and persistence /archiveSource, archive, indexing, provenance and context loss /falsificationHypotheses, instruments, thresholds and rival explanations /optionalityReserve, redundancy, reversibility and open futures /governanceGovernance cycle from observation to memory /authorityAuthority, evidence and recovery as longitudinal variables /agentsAgent authority envelopes, identity, scopes, tools and consequences /identityTask, profession, identity, machine capability and context /modelsObservable generative-model processes /genaiOperational anatomy of governed GenAI applications /historyObserved institutional history with provenance /privacyPrivacy and external-agent data policy