About

Just the source, made accessible.

Important public reports — government inquiries, investigations, official findings — contain some of the most careful research ever produced on matters of public importance.

But they are buried. Scattered across broken government websites, locked inside enormous PDFs, impossible to link to at the level that actually matters: the specific finding, the key paragraph, the actual evidence.

Reports that Matter makes these reports usable on the web. Searchable. Readable. Linkable at the paragraph level.

The report's own words, as published. Our introductions and highlights are labelled as ours.

Faithful text

Reports are converted from their original PDFs by a deterministic pipeline, with automated fidelity checks against the source.

Paragraph permalinks

Every paragraph has its own address. Highlight any passage to copy a link straight to it, or save it to your own list.

Stable hosting

Published as plain web pages that search engines can index and that will not rot behind a broken departmental site.

The report text is the official text. It is taken from the published document by ordinary, deterministic software and checked against the source; no language model wrote, summarised or rewrote it, and nothing an AI wrote is presented as part of a report. For one scanned report, the Challenger report, a vision model's reading of the page images helps find where paragraphs and headings break, used only where its reading closely matches the document's own text; the words are always the document's.

AI coding agents wrote much of that software, and this website, under Rufus Pollock's direction; the code is public. The introduction at the top of each report is ours, not the report's: drafted with AI assistance, published under our editorial responsibility, and every excerpt it shows from the report is checked automatically to be the report's exact words.

If you are named in or affected by a report, you can ask us to change or remove anything we have made from it: a quote card, a post, an excerpt or highlight, a line of our introduction. We will do so within a day, and you do not need to say why. Write to contact@reportsthatmatter.org; it goes to the editor, Rufus Pollock. Or, if you would rather do it in public, ask for a correction or removal on GitHub.

If something in a report reads wrongly here, a sentence cut in half or a stray number, write to the same address or report the problem. Where our copy and the official document differ, the official document is right.

The GitHub forms need a free account, and what you write there is public. Email is not. For anything else, write to the same address or open an issue.

There are no accounts and no tracking cookies. When you save or share a passage, your browser sends us its words and a few words either side, with the report, paragraph and page they come from, whether you saved or shared it, and the time. We keep them alongside a code made from your network address and browser that changes every day, used only to count how many different readers marked a passage and to limit how many passages one reader can mark in a day; the address and browser themselves are not stored.

Saves and shares are both counted, anonymously. When the words are found in the report, the passage can appear on Highlights about half an hour later, with the report's own text, page and a time rounded to the hour, and nothing about who chose it. Your own list of saved passages stays in your browser, and you can turn counting off on Yours.

The code is open source under the MIT licence. Each report's text keeps the status it has at source, which is not ours to change: most are works of the US government, in the public domain, or UK Crown copyright reused under the Open Government Licence. A few are neither, or we could not confirm their status, and we say so. Our own markup and editorial writing are CC BY 4.0. The exact terms, with the evidence and anything we could not confirm, are in the LICENSE file of each report's repository.