Sourcing
Not a confidence score we made up. A description of how directly each claim traces to a document you can open yourself.
Why not a score
A percentage or a letter grade on our own work would be a number we invented, and the first person who wants to dismiss this site would be right to ask how we calculated it. So we do something narrower and checkable instead.
Every figure and claim on this site carries a label describing how directly it traces to its source. That is a fact about provenance, not a judgment about truth. A label tells you what kind of document is behind a number and how many hands it passed through on the way to us, so you can decide for yourself how much weight it carries.
The rule we follow: the further down this list a claim sits, the more carefully we word it, and nothing below the last tier goes on the site at all.
The labels
We read the actual source and you can open the same file. Google's Project Raptor FAQ, the ERCOT market notice, the Archer County datacenter page. Nothing between us and the original. Most of the hard numbers on this site sit here.
Commissioners court minutes, posted agendas, PUCT filings and orders, agency releases. Authoritative on what a body did, though a set of minutes is a summary and not a transcript.
Quotes from meetings, vote descriptions, attendance counts. Local outlets covering our own county. Reliable, but reporters compress and paraphrase, and two outlets covering one meeting sometimes differ in detail. Where they conflict we say so.
Most commonly, a study or news analysis compiling company disclosures. Useful, and one step removed. When a claim is labeled this way we are telling you to check the underlying disclosure before quoting it at a podium.
National laboratory and research institute projections. Legitimate work with stated methods, but these are calculations about a range of facilities, not readings from a meter. Useful for scale, never for precision about our site.
What does not appear
Anything without a published method. There are industry rules of thumb circulating that would put a campus this size far above the figures Google has stated. Several of them are on sites that look authoritative. None of them show their work, so none of them are here.
Anything we could not open ourselves. If a figure appears in a summary and we cannot reach the document behind it, we either label it as secondary and say so plainly, or we leave it out.
Property value claims. The research on large industrial neighbors is mixed and location-specific, and nobody has studied Archer County. We could find numbers that support us. They would not survive a question from someone who reads carefully, and one collapsed claim would cost us every claim that is solid.
Health effects. We take noise seriously and we say so in engineering terms. We do not make health claims we cannot source.
Corrections
If a number here is wrong, we want to know, and we will fix it in public rather than quietly. Send it through the contact form with the document you are looking at. No name or email required. That includes people who support the project, especially people who support the project.
Primary Every source on this site is listed and, where possible, mirrored on the documents page.