Sourcing policy

The hierarchy

Sources are ranked by how close they are to the thing being claimed. A factual claim should rest as high on this list as the subject allows:

1. Government bodies and public authorities.
2. Building-code and standards bodies, where their material is publicly accessible.
3. Official research institutions.
4. Peer-reviewed literature.
5. Universities.
6. Recognised architecture and construction institutes.
7. Official manufacturer technical documentation — for product-specific facts only.
8. Industry organisations.
9. First-party company announcements — for company and product news only.

Every citation on this site records which of these categories it belongs to, and that category is displayed next to the reference. A reader should be able to see at a glance whether a sentence rests on peer-reviewed research or on a manufacturer’s own datasheet, because those are very different warrants for the same words.

What does not count

Factual claims about construction are not based on SEO content, anonymous trade blogs, affiliate sites, scraped summaries, AI-generated summaries, image boards or content farms. Secondary journalism is useful for finding out that something happened; it is not what the final factual claim rests on. Where a secondary source is the only available one, it is cited as secondary analysis and the limitation is stated.

Paywalled and restricted standards

Much of the standards literature that governs construction is not publicly readable. Where a standard is not accessible, the Journal says so and describes what is publicly documented about it. It does not paraphrase a standard it has not read, and it does not imply that it has reviewed a full text that sits behind a paywall. A standard is cited by its number and issuing body so a reader with access can check it directly.

Access dates

Where an access date is recorded, it is the real date the source was consulted. Access dates are never approximated or back-filled to look thorough. An article with no access date on a reference is telling the truth about what was recorded; an article with an invented one would make every other date on the page worth less.

How citations appear

Numbered markers in the text link to a reference list at the foot of the article, and each reference links back to the sentence it supports. The list is in the initial HTML — it is not loaded afterwards, hidden behind a control, or rendered only for readers running JavaScript — because a citation that a machine cannot see is not doing the job a citation exists to do.

When the evidence is thin

Some questions in construction do not have a good published answer. The Journal reports that rather than filling the gap. “There is no public dataset on this” is a finding; inventing a plausible figure is not a smaller problem than getting one wrong, it is a larger one, because a fabricated number cannot be corrected by evidence.

See also the editorial policy and image policy. Questions about sourcing: info@helperg.com.