How the record was read
This app reads a single cloud database built by the insight-bridge pipeline from the public submissions to the 2026 Murray–Darling Basin Plan Review: 999 submitting entities, 1,293 documents, 11,336 passages. Everything on every page is a live query against Cloudflare D1 (the relational residual) and Vectorize (the embeddings), through fully-typed Drizzle. The same data is exposed to agents over an MCP server at /mcp.
The pipeline
- Ingest
The published submissions to the 2026 Murray–Darling Basin Plan Review (consultation 5 Feb – 1 May 2026) are collected, split into documents, chunked and normalised. Submitting entities carry their published facets: sector, respondent type, named status, ICIP flag.
- Per-document extraction
Each document is read end-to-end by a language model: key points with verbatim supporting quotes; tags across four controlled vocabularies (Basin policy theme, Evidence type, Stakeholder interest or role, Geographic focus); and a recommendations-and-tradeoffs analysis under four fixed subtopics — concrete actions requested, conditions attached, responsible actors, outcomes sought or feared.
- Embedding & topic clustering
Topics are found from the passages themselves, bottom-up, by grouping the passages across the corpus that say the same thing; nothing supplies a list of themes beforehand. A topic has to be raised by more than one entity to form: one entity's volume cannot make a topic on its own, and the rare topic that only one entity carries is flagged as such. 88 cross-cutting topics resulted. Each passage, and each entity, is then placed against every topic at a graded strength (exemplar, high-value, member), so a entity that touches a topic in passing is recorded as well as the entities that define it, and a passage can belong to more than one topic. Each document is assigned its dominant topic.
- Perspective synthesis
Each topic is written up as a proposition with key points, and every exemplar and high-value entity's position is recorded against that proposition, from Supports to Opposes. Positions are relative to the topic's own framing, never absolute agreement, and every entity is measured against the same proposition. Each position carries its framing, analysis, key points and quotes, and a sector lens gives each sector's aggregate stance on the topic with its reasoning.
- Superclustering
The 88 topics are grouped one level up into 20 named supercluster groups — the map of what the consultation was actually about. The basin tree has a single grouping generation (unlike deeper corpora), so group and family coincide.
How to read the output
These rules apply to every Insight Bridge corpus. They are properties of the method, not caveats about a particular run.
- Positions are relative to a proposition. A source’s position records how it stands against that cluster’s particular framing — not whether it agrees with some absolute claim. The same source can support one cluster and redirect a neighbouring one that covers similar ground differently.
- Propositions are written from the cluster’s own members. Because the argument is built from the sources that were grouped together, and positions are read relative to it, a degree of agreement is built into the method. Comparisons between groups carry weight; a corpus-wide agreement rate does not.
- Counts describe the corpus, not the world. Every corpus here is curated. “N sources say X” measures what was collected and is never a measure of how common X is in the field.
- Clusters differ in how many distinct sources back them. A long document can fragment across many clusters, so weight a theme by the distinct sources beneath it rather than by how many clusters it contains.
- Every extraction and position is a model judgement. Key points, stances, propositions and syntheses are produced by a language model reading the source. They inherit its calibration and are not determinations of fact.
Exclusions and caveats in this corpus
426 no-content stubs, submission records with no retrievable text, were excluded at load, so document counts describe what was actually read. ICIP-flagged submissions (those declaring Indigenous Cultural and Intellectual Property) keep their entity and document metadata so the record stays honest, but their verbatim text — chunks, quotes, key points — is withheld from this deployment pending ICIP consent processes.
Where the disagreement is. The framings here are things the submissions actually argue — "halt water buybacks", "recover water for the environment" — and 88.5% of entity stances read supportive against them, in a corpus containing a deep two-bloc split. Dissent therefore surfaces between opposing framings rather than inside one topic's rows; the Fault lines view is built to show it.