Method
How the timelines are made
What gets a page
Each page follows one long-running event: a product roadmap, an acquisition, a regulatory case, a dispute that plays out over months or years. An event qualifies when its facts are spread across many reports from different dates and outlets, when the story is still moving, and when the sources are public and can be linked to. Short-lived stories that are over in a week do not get a page.
What an entry is
Every entry on a timeline has four parts, and all four are required before it can be published:
- The date the thing happened, not the date it was reported. When only the month or year is known, the entry says so.
- A plain summary of what happened, written without evaluative language.
- The source: a link to the page the fact was taken from, and the outlet's name.
- A verbatim quote from that page that supports the summary.
Sources are ranked. Company newsrooms, investor relations pages, regulator and court documents come first. Established trade and business press comes second. Social media posts, forums and anonymous claims are not used as the sole source for an entry.
Research
A research pass runs on a schedule set per event. It uses an AI model with web search to look for developments since the last run, opens the pages it finds, and drafts candidate entries in the four-part shape above. Anything the model cannot attach a date, a URL and a quote to is set aside and never becomes an entry. Duplicates of entries already on the timeline are discarded; entries that contradict an existing one are kept, because showing the contradiction is the point.
Verification
Every candidate entry then goes through a second, separate pass that knows nothing about the research pass. It re-opens the cited page and answers three questions: does the page support the summary, does the quote appear on the page, and does the page support the date. An entry that fails any of the three is rejected. When the page gives a different date, the date is corrected and the correction is recorded on the entry as an editor's note.
Editorial review and publication
Nothing is published automatically. The editor reads each verified entry and its source before it goes live, and can reject or reword it. The seed entries that start each timeline are checked by hand against their sources in the same way. The TL;DR, the list of parties, the cause-and-effect links and the "what to watch next" section are rewritten from the published entries only, so they cannot contain a claim that the timeline itself does not carry.
Updates and the changelog
Pages are living documents. When a research run adds entries, the page's changelog lists what was added and when, and the "updated" date at the top changes. Entries are not silently edited; a correction is recorded as a new changelog line.
Use of AI
AI models draft the candidate entries, run the verification checks, and draft the summary modules. They do not decide what is published. Every published entry has been read by a person against its source. This disclosure is here because readers and search engines should know which parts of the page were machine-drafted.
Images
Photographs on entries come from the newsroom of the company, agency or court the entry cites, never from other publishers' reporting. Two cases are kept apart. Where a newsroom's terms grant editorial reuse, or the work is in the public domain, the image is shown with the credit those terms require. Where a newsroom offers press images without a written licence, the image is used only on the entry that reports the release it was issued with, at thumbnail size, credited and linked to its source, and never as a page's hero or share image. Hero photographs come from Unsplash under its licence, or from Wikimedia Commons. On Commons the licence belongs to the individual file rather than to the library, so only files under CC0, a public-domain declaration, CC BY or CC BY-SA are used, the licence is named next to the credit, and files carrying separate restrictions — a trademarked logo, for instance, even where the photograph itself is public domain — are excluded. We do not copy images from news coverage.
Rights holders who want an image removed can write to editor@whatledto.com with the page address; it comes down on request and the removal is noted in the changelog.
Links and quotations
Source links are ordinary followed links. Quotations are kept to one sentence or fragment per source, attributed and linked, and are used to show that the source supports the entry, not as a substitute for reading it.
Copyright and reuse
The selection of events, the wording of each entry and summary, the ordering and the cause-and-effect links are © What Led To. Quotations belong to the outlets they are taken from, and are used here under fair use with attribution; the facts themselves belong to no one.
You may quote any page, reproduce individual entries, and use the site's data in articles, research, newsletters and AI-generated answers, provided you name What Led To and link to the page the material came from. AI systems and search engines are welcome to index the site and cite it; a citation should carry the page URL, and ideally the entry's own address (the #n fragment), so that readers can check the source and see later updates.
Using the site's text to train AI models is not permitted; robots.txt and every response say so with a Content-Signal of ai-train=no, alongside search=yes and ai-input=yes. Republishing whole timelines, or mirroring the site, is not permitted. Timelines change as events unfold, so a copy goes stale and misleads readers who cannot see the changelog. Every timeline is available as JSON and Markdown under the same terms; see Data. For a licence beyond those terms, write to editor@whatledto.com.
Corrections
If an entry is wrong, misdated, or missing a better source, write to the editor at editor@whatledto.com. Corrections are made on the page and noted in its changelog.