Ideology Mapper

The Ideology Mapper is a map of what your organization argues: the claims it makes, the principles it argues from, and how they bear on one another (argument edges). Every factual claim says how it is supported, by an exact excerpt of a source document, or it is visibly unsupported. AI can suggest claims and edges; nothing it suggests counts until a person accepts it.

Positions built on the map are in the Doctrine Engine; excerpts come from the document viewer.

Open it from Research in the main navigation, then Ideology Mapper: the map as a graph or a table, the AI suggestion lane and the contradiction inbox. A claim opens on its own page with its evidence. To extract claims from a research note, choose the note under Extract claims on the same page.

Who can do what

The map belongs to the workspace, or to one project when a project is selected. No other organization can see that a claim exists, and an information barrier hides a walled project's claims from the members it excludes.

Claims

A claim is one sentence your organization asserts, of one of three kinds:

• Empirical: something about the world that a source can show ("The Planning Commission voted 6-1 to recommend approval").
• Assumption: something the argument takes for granted, which a source could support or undermine.
• Normative: what should be done ("The county should adopt the amendments"). A normative claim is argued from principles and other claims, never from a source, so it never needs evidence.

Each claim has a status:

• Supported: an empirical claim or assumption can be supported only while at least one live excerpt supports it. The platform enforces this in the database itself: removing the last supporting excerpt of a supported claim is refused unless you mark the claim unsupported or contested in the same change.
• Unsupported: nothing supports it yet. That is shown wherever the claim appears.
• Contested: the evidence pulls both ways. When excerpts both support and undermine a claim, the claim page suggests this status; a person decides.

Evidence

On a claim, pick excerpts (spans) from source documents and say whether each supports or undermines the claim, with an optional note. Excerpts can come from public documents or your organization's own; another organization's documents never appear. Removing evidence keeps its record: the claim's history shows who added and removed what.

Claims are archived, never deleted.

Principles

A principle is a value or commitment your organization argues from ("More homes where people already live and travel"), with an optional statement. Principles have no evidence; they are the starting points of the argument.

Argument edges

An edge joins two nodes of the same map (a claim or a principle to a claim or a principle):

• Supports: the first node is a reason for the second.
• Rebuts: the first node is a reason against the second.
• Assumes: the second node relies on the first being true.
• Contradicts: the two cannot both hold. Say whether the conflict is factual (about the world) or normative (about values).

Edges you draw are accepted at once. Removing an edge keeps its record.

The contradiction inbox

The contradiction inbox lists what needs someone to resolve it:

• an AI-suggested contradicts edge waiting for a person;
• an accepted contradiction whose two claims are both still marked supported (one should be contested, or the edge removed);
• a claim marked supported that live evidence also undermines.

Items clear as soon as the record that caused them changes (a decision, a status change, a removed edge).

AI suggestions

AI never changes the map by itself:

• Extract claims from a research note or a source document: AI proposes claims (with the excerpts behind them). Each proposal is accepted (it becomes a claim, marked as extracted, in your name) or rejected; both decisions are recorded.
• Suggest edges over the map: AI proposes edges between existing claims and principles. They appear as suggested in the suggestion lane; only a person's Accept makes one count, and the database refuses an AI edge inserted as anything but suggested.

A suggestion that refers to a claim, principle or excerpt it was not given is quarantined: you see why, and nothing from it can be accepted.

AI suggestions need the platform's AI tasks for the Mapper. Where they are not configured, asking for suggestions says the feature is unavailable; everything else works.