How to reach them

Twenty-three people wrote these pieces, at 26 organisations. If you want to answer them — and this section exists partly because we do — you need a route. The corpus gives you none. This page is what going to the organisations’ own sites produced, under rules published before the looking started.

The rules, published before the looking

A contacts list is exactly where a publication does damage if it is careless, so the method is fixed in data/contact-rules.json and enforced by the gate:

1. Organisations, not people. Every address here is an organisation’s own published role address, from that organisation’s own site. No personal email, no direct line, for anybody. Individual routing is via the organisation.
2. A role address is a published formula. An address is published only if its local part is one of these, or is the site’s own name: admin, ask, comms, communication, communications, contact, contactez, contacto, contacts, contato, correspondence, desk, editor, editorial, editors, email, enquiries, enquiry, general, hello, help, hey, hi, imprensa, info, information, inquiries, kontakt, letters, mail, media, media-enquiries, mediaenquiries, news, newsroom, office, pitches, prensa, press, press-office, presse, pressoffice, reachus, redacao, redaccion, redaccion, redaktion, secretaria, secretariat, submissions, support, team, tips, webmaster. Everything else found is counted and dropped. 32 were dropped to publish 11.
3. One hop. The home page, then the one link the site itself labels as contact. Both frozen and hashed. No directory, no third-party database, no guessed address.
4. The domain has to be established. Four published tests; a page that passes none is kept as a candidate and nothing is published from it. News Corp Australasia fails all four because the site says News Corp Australia — close is not the same, and a contacts list is where close is dangerous.
5. A page that is a staff directory is not retained. Ten or more addresses of named people on one page and we keep the URL, the hash and the counts but not the bytes. Why, and what it costs →

The organisations

The authors

Route only. No personal contact detail for any named person appears anywhere in this section’s data, and the gate refuses one where the data is parsed rather than hiding it where the page is rendered.

If you are one of the twenty-one

Then you are the reader this page was built for, and three things are worth saying plainly.

The finding is not a criticism of your piece. It is about the infrastructure under it: a permission with no licence name, terms stated three different ways, and a corpus with no route back to its authors. None of that is your doing and all of it is fixable in an afternoon.

Everything here is checkable, and wrong things get corrected in public. Every number walks back to bytes we froze and hashed; the vault is one download. This section has already corrected itself once, in public, about the publisher of this very corpus — what we got wrong →.

If something here is wrong about you or your organisation, say so and it changes. The repository is open and every page is generated from data files you can read. Corrections are recorded with the version that carried the error, not quietly edited away.