About

A research record you can check: every document kept whole, every quote traceable to its source.

1Who writes this

Chris Telford commissions deep research on whatever question he is working through — science, media, technology, the products he is building — and keeps every document whole here.

He writes from it: articles that argue from the documents, papers that cite the literature, and names for patterns nobody had named.

2What you will find

Article

Articles

Short pieces that argue one idea from documents in the record. Every quote links back to where it came from.

Paper

Papers

Longer, original work that cites the published literature, with numbered references.

Research

Research documents

The documents themselves, kept whole: deep-research reports, evidence reviews, project briefs.

Concept

Concepts

Patterns the research found that had no name, so they were given one, after checking nobody already had.

Open problem

Open problems

Questions nobody has answered, or things nobody has built, each linked to the document that found the gap.

3How it stays honest

Documents stay whole

A document goes up exactly as it arrived, at one permanent address. New versions keep the address.

Quotes are word for word

Every figure and quotation is copied from a stored document. Software checks each one; anything that does not match is dropped.

The software does not judge

It finds and links. It never rates sources, settles disagreements or summarises. Where two documents disagree, both are shown.

Only the author publishes

Chris writes with help from Claude. Nothing goes live until he publishes it.

4Reading a research page

Each research document opens with the parts most people want — the summary, the key figures, the main findings — pulled out of the document in its own words. Below them come the claims it makes and how strong it says each one is, the questions it answers, its caveats, and then the full text.

Key findings12SupportedContestedVendor claim

Green pills link a quote to the part of the document it came from. Blue numbers link to a source. Coloured pills show how strong a claim is.

5For machines

For readers who build tools: every page is plain HTML, and each document also comes as Markdown and JSON.

/p/<slug>The page. Complete HTML, no JavaScript required to read it. Article and Person markup with the author's sameAs; a Dataset block describing the key-figures table; every named source as a citation. FAQ markup is not emitted — Google retired FAQ rich results for most sites in 2023, so the question-and-answer shape is carried in headings and the answering sentence instead. Indexed. /p/<slug>.mdThe document, byte-exact, with its front matter. For agents and for you. Not indexed; canonical is the page. /p/<slug>.jsonEverything found in the document, typed, with receipts. Not indexed. /q/<id>A receipt: the passage, its location, its check. Not indexed; canonical is the document at the anchor. /llms.txt · /llms-full.txtOne line per document, grouped by project, in the document's own words. Kept because it is cheap; its fetches are counted rather than assumed. /sitemap.xml · /feed.jsonDocuments and articles with the date they were last read. A new version pings IndexNow. citation_* metaHighwire tags on every research document — title, author, date, the .md twin as full text — so Google Scholar indexes it as a document, not a web page. #:~:text=Every receipt is a text-fragment link that highlights its passage. Deep links that survive re-reads, because they are made of the words. /searchPlain full-text search over passages and documents. Not indexed. AI crawlers are allowed at the edge and the allowance is checked weekly.