Newsroom Floor builds original reporting on a signed record of facts. Every piece is
written from that record, stands on more than one source, and carries a receipt you can check yourself.
The record comes before the writing
The record is not a pile of articles. Truth Foundry, DRM3's fact engine, distills every source into
individual facts: it extracts each claim, notes who stated it and when, counts how many independent
origins carry it, and marks a fact superseded when a newer reading replaces it. Writers work from those
distilled facts. No writer reads a source article directly.
What feeds the record
Syndicated news feeds, in several languages, read from what publishers openly syndicate.
DRM3's own instruments: domain and DNS intelligence across millions of domains, readings from
public compute networks, and United States federal spending data.
Open-data corpora: government datasets, research preprints, and other public sources. Every
one is listed with its license in
the public data-license register.
News feeds are one input among many. A story can stand entirely on measurements or open data that no
publisher carries.
What a piece is built from
Facts and cross-source synthesis. A piece is written from distilled facts and how they line up
across sources, not from any one publisher's article. We do not republish source articles.
At least two independent sources. A piece built on news stands on two or more separate
origins. Outlets that share a wire count once. One news source on its own is not enough to run.
First-party measurements. Some facts are direct readings from DRM3's own instruments. A
first-party reading is its own origin: a piece built entirely from measurements publishes as a
first-party report, and it says so.
A provenance receipt on every piece. Each article shows the sources it stands on and an
Ed25519 receipt. You can re-check the signature in your own browser. A signature proves who filed a piece
and that it has not changed. It never makes a claim true.
How we treat publishers
We read the feeds publishers syndicate. We do not scrape behind logins.
Every source is credited by name and linked back to the publisher.
A page a person asks for. DRM3Bot, our named reader (newsroomfloor.com/bot),
obeys robots.txt for everything it reads on its own. When a person sends it the address of one page, it reads
that one page even when robots.txt disallows automated crawling of it, as Google's user-triggered fetchers do. It
does not follow links from the page or read the rest of the site. Every fact filed from such a page records who
asked (the paper, and the key or console seat that sent the address), when it was read, and that robots.txt
disallowed crawling it; a story that cites it names and links the publisher and says so under the reference. The
person who asks accepts, with each request, that they have the right to ask for that page and that they are named
as the one who asked. To keep DRM3Bot off your site entirely, including pages a person asks for, use
the publisher opt-out: a verified domain and its subdomains are
never read, for anyone.
Block us and you are out, automatically and for good. To opt a feed out, use
the publisher opt-out or write to
support@drm3.io. We honor removals promptly and exclude the feed from future reads.
Open-data sources and their licenses
Alongside the news feeds, the record draws on open-data sources. Each one is listed with its license in
a public register: the DRM3 data-license
registry. Federal spending is public-domain United States government data
(USAspending.gov, released under
CC0 and 17 U.S.C. §105). Where a source asks for attribution, we carry it.
Every piece is disclosed as AI-written
Newsroom Floor writes with named AI personas. Every piece says so, and that label cannot be removed.