How the record was read
This site reads a single database built by the insight-bridge pipeline from its corpus: 50 issuing bodies, 4,183 reports, 478,543 passages, 2,197 issuing body–topic memberships, 2,197 issuing body-level stances, 885 kind of body perspectives.
Everything on every page is a live query against Cloudflare D1 (the run's own tables, copied whole) and Vectorize (the embeddings), through fully-typed Drizzle. The same data is exposed to agents over an MCP server at /mcp.
What was done, in order
Each step below is a stage this run actually executed, read from the run's own record of itself; a stage the run was configured to skip is not listed.
- Collection
Published Australian public-sector accountability reports, 2006–2026, from oversight bodies across all nine jurisdictions — auditors-general, ombudsmen, anti-corruption and integrity commissions, royal commissions, the Productivity Commission, inspectors-general and independent reviews. Each body carries its kind and its jurisdiction, and each report its year and its type, as the collection recorded them; the time range is the year placed in one of four periods. Partial-year 2026 is included as-is.
- Carried onto the current record
The corpus was read in July 2026 by an earlier version of the pipeline, and its store was later carried onto the current run schema by the pipeline itself: the analysis as it was written, no step re-run and no text changed, and every quote checked again against its report by the current rule. That earlier pipeline recorded nothing of its settings beyond the run’s name, and ran no review of empty topics, so every topic found stands.
- Reading in the reports
Every report was stored whole under its issuing body, its text cut into passages small enough to compare — 478,543 of them across 4,183 reports from 50 issuing bodies — and each passage embedded, which is what makes a search by meaning, and the grouping below, possible at all.
- Reading each report end to end
A language model read every report in full against this run's profile and recorded what it found as key points, each with the verbatim quotes that support it: 38,874 key points and 111,196 quotes. Every quotation is verbatim from the source report: each is verified against the report's text before it is stored, and one that cannot be found there is never presented as verified. 98.6% verify, and a quote that does not is never shown here. The same reading graded each report on two 1–5 rubrics — finding severity and systemic extent — and tagged it for theme and recommendation acceptance.
- Grouping the passages into topics
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 issuing body to form: one issuing body's volume cannot make a topic on its own, and the rare topic that only one issuing body carries is flagged as such. The run did not record the settings it clustered with. Each passage, and each issuing body, is then placed against every topic at a graded strength (exemplar, high-value, member), so an issuing body that touches a topic in passing is recorded as well as the issuing bodies that define it, and a passage can belong to more than one topic. 276 topics resulted, over 441,236 passages.
- Writing each topic up and reading positions against it
Each topic is written up as a proposition with key points, and every exemplar and high-value issuing body's position is recorded against that proposition, on this run's own scale, Assurance, Minor, Significant and Systemic. Positions are relative to the topic's own framing, never absolute agreement, and every issuing body is measured against the same proposition. 2,197 positions were written, and the same was done for each value of type of oversight body and time period: 885 perspectives. A position here records how an issuing body stands against one topic's particular framing, and is never a verdict on the subject in general.
- Offering the topics upward into themes
The topics were grouped again, generation by generation, each generation clustering the one below it; the settings it ran with were not recorded. This site reads the tree as the run left it — here, 70 families above the topics, each named by the model from its members.
- Placing the topics in a sky
Each topic's place in the galaxy was computed once and stored with the run, seeded so it is the same sky every time. That is what lets these pages draw the constellation as plain graphics, with no WebGPU and nothing solved in your browser.
- Loading this site
Outside the pipeline, before loading: the run database was copied whole to the cloud — every table as the pipeline wrote it, with the passage embeddings moved to a vector index — and two facts were derived from it for these pages, each report’s and each passage’s dominant topic. Nothing was flattened, re-verified or re-classified here; every figure on every page is a live query over the run’s own tables.
What each issuing body was classified by
Every facet the run's profile declares. A facet marked compared is one the analysis wrote a perspective for, value by value — the lens you can read a topic through. Facets that are not compared ride beside it as facets of the record.
What a position means here
This run's scale does not measure agreement. Read the labels as what they say and not as a spectrum from support to opposition.
Choose the level that best reflects the SEVERITY of the findings on the cluster's theme; this is an assessment of finding severity, not tone.
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.
In this run
Severity and systemic reach are graded on two 1–5 rubrics, so they rank reports against each other and never against an official standard. Cross-body divergence reflects both genuine differences of mandate and differences in what each body was examining, and averages combine very different report genres, so they describe the corpus rather than evaluate the bodies in it. Recommendation-acceptance rates cover only reports that recorded a response; many genres (royal commissions, Productivity Commission inquiries) structurally do not, and are excluded rather than read as evasion. The family tier is read through each supercluster's primary family, while the topic↔supercluster tier is genuine multi-membership. Partial-year 2026 is included as-is.
What else was asked of every report
Beyond the standard reading, this run declared its own questions and put each of them to every report.
This run kept no record of the models or settings it ran with, so this page names none of them. Everything else here is read from the run itself.