Define evidence-context-object and non-linear journey modelling guardrails

[happygeneralist]

Context

The first bounded Ofsted ingestion slice has exposed an important modelling tension.

Bounded slices are useful for controlling evidence extraction, review size and provenance. However, a narrow extract set must not become the whole interpretive frame for formulating user needs, civic needs, pain points, insights or journey views.

Recent discussion has clarified several connected distinctions:

• need level and journey context are separate dimensions;
• a need may persist or recur across several lived-experience contexts;
• current policy and service mechanisms should not define enduring needs or pain points;
• mechanism context must still be preserved because it affects power, consequences, eligibility, remedies and burden;
• journey understanding should emerge from patterns in lived experience rather than be imposed as a linear path;
• evidence links should identify what aspect of an object they support rather than implying that a source validates the whole object;
• civic needs need a clearer normative basis and should not become well-written organisational preferences.

Goal

Define a small set of public-safe modelling guardrails for moving from source-specific evidence to durable knowledge objects and usable, non-linear journey-pattern views.

The goal is not to finalise a complete ontology. It is to make the next ingestion and formulation work safer while the deeper conceptual and protected-method work continues elsewhere.

Working principles to test

Bounded extraction, wider formulation context

A bounded slice controls what is extracted and reviewed in one work package. It does not, by itself, define the correct scope of a user need, civic need, pain point or insight.

Before creating or materially revising an interpreted object, review:

• the selected evidence;
• the wider source context;
• nearby existing repository objects;
• the relevant lived-experience contexts;
• the current mechanism through which the issue became visible;
• what the evidence does and does not support.

Separate enduring objects from mechanism-specific manifestations

Do not encode current policy or service mechanisms into enduring user needs or pain points unless the mechanism is genuinely part of the human condition being described.

Preserve mechanisms as evidence context, constraint, response, entitlement, funding, accountability or implementation context so their contribution to outcomes can be inspected.

Treat journeys as evidence-backed patterns and generated views

Do not assume a single linear pathway.

Build journey understanding from evidence about:

• contexts;
• events and triggers;
• transitions;
• waiting periods;
• decision points;
• responsibility changes;
• support mobilisation;
• escalation and recovery;
• recurring or looping experiences.

Journey views should help people design for recurring lived-experience patterns without becoming canonical containers that determine the meaning of needs or pain points.

Bound the claim made by each object

Wider source context may shape interpretation, but each object should still make a bounded claim.

Evidence-to-object relationships may need to distinguish support for:

• existence;
• local manifestation;
• actor experience;
• consequence;
• severity;
• prevalence;
• contextual relevance;
• association;
• source-asserted causation;
• analyst-inferred causal mechanism.

This distinction should be tested before any new relationship schema is implemented.

Preserve civic-need uncertainty

Do not create civic needs merely because an inspection report tells an institution to improve.

A civic need should eventually state:

• what public condition, protection, capability or obligation is at stake;
• who is affected;
• the normative basis for treating it as civic;
• how current mechanisms support, obstruct or redistribute that responsibility;
• what outcome or public value is at risk.

Until that conceptual work is mature, prefer candidate or needs_review treatment and avoid pretending the current administrative arrangement is itself the civic need.

Questions to resolve through real examples

• How much wider source context is enough before formulating an object?
• How should an object state which parts are supported by which evidence?
• When is mechanism context a field, a linked note, a constraint or a future first-class object?
• How can journey patterns remain usable for design without implying a standard linear route?
• How should recurring higher-level needs connect to more situated expressions without forcing a rigid hierarchy?
• How should pain-point levels or relationships distinguish experienced pain, pathway conditions, systemic causes and institutional risks?
• How should tensions, competing needs and distributional effects be represented?
• What minimum normative-basis information is needed for a candidate civic need?

Proposed work sequence

1. Capture the conceptual development plan in the Happygeneralist knowledge repository.
2. Develop private review-rubric candidates in design-intelligence-orchestration.
3. Use the current Essex Ofsted source and one later contrasting source to test the distinctions.
4. Record where the current object model is sufficient and where lightweight metadata or notes are needed.
5. Add a small public guidance note only after the principles survive real use.
6. Consider schema, relationship or validator changes only after repeated use shows they are necessary.

Out of scope

• Do not create a fixed journey-stage hierarchy.
• Do not implement a full mechanism ontology.
• Do not add new first-class object types yet.
• Do not rewrite existing user needs, civic needs or pain points in this issue.
• Do not create or upgrade civic needs from the first Ofsted slice automatically.
• Do not treat the current EHC framework or any other policy mechanism as a permanent fact of the problem space.
• Do not publish protected rubrics, prompts or ontology-navigation logic.
• Do not mark any object reviewed or validated.

Acceptance criteria

• Public-safe guardrail candidates are agreed or revised through examples.
• The distinction between bounded extraction and broader formulation context is explicit.
• Need level, journey context and mechanism context are kept separate.
• Journey views are framed as useful non-linear patterns over evidence-backed objects.
• Evidence-to-object claim boundaries are tested.
• Civic-need formulation is linked to dedicated conceptual work rather than improvised during ingestion.
• No schema or ontology expansion is implemented without a later focused issue.

Related work

• Develop safer repeatable non-linear ingestion workflow #86 Develop safer repeatable ingestion workflow after first bounded slice
• PR Create first Ofsted Essex communication evidence objects #85 Create first Ofsted Essex communication evidence objects
• docs/Safe_research_ingestion_MVP.md
• docs/User_needs_writing_rules.md
• Happygeneralist: 02_Projects/Civic_Design_Intelligence/Main_ontology_working_model.md
• Happygeneralist: 02_Projects/Civic_Needs/00_Meta/Roadmap.md
• Private orchestration issue Add research study scope metadata #8

[happygeneralist]

The candidate review for the next Ofsted Essex delay slice has exposed two concrete public guardrail questions, now documented in draft PR #94:

1. Source-led granularity: the report says some children and young people wait too long for appropriate provision and support. The planned analysis distinguished provision from support, but the sentence does not. The candidate evidence should remain one bundled unit rather than being atomised to match our categories.
2. Reuse of multi-claim evidence: EVID_003 already preserves a page 5 statement containing EHC needs-assessment timeliness, although the record was created and interpreted for the communication slice. The next slice should not duplicate the source statement, but it needs a lightweight way to reuse only the supported claim without implying that every part of the evidence record supports every linked object.

Neither finding currently justifies schema or relationship-type changes. The draft uses explicit notes and defers any evidence-maintenance decision until the candidate set is reviewed.