Direct answer: across the 36 article pages, using four signal types, 22 pages (61.1%) have no author/source hit at all; 14 have at least one. The most common hit is JSON-LD: 9 pages carry an author field, but every one is an Organization, not a Person. article:author appears on 0 pages; the only non-empty meta author holds a website vendor string, not an article author.
Five pages contain one of the visible byline/source words (Chinese terms for author, editor, responsible editor, source, contribution or writing), but that is a keyword hit, not a verified byline. This page only measures whether a marker exists; it does not judge whether the author is real, the content is trustworthy, or the page is good.
Method: four separate signals
We reused the 36 cached article pages from study 83 (10 sites, fetched 2026-11-03), with no network. Each page is checked for a non-empty meta name=author; a non-empty article:author; an author key in any JSON-LD object, with its type; and visible text containing one of six Chinese byline/source words (author, editor, responsible editor, source, contribution, writing) after removing script/style/comments/tags.
| Signal | Rule | Boundary |
|---|---|---|
| meta author | Non-empty content | May be the website vendor, not the author |
| article:author | Non-empty content | May be a URL or a person's name |
| JSON-LD author | Any object with an author key | Type recorded as Organization/Person/Other |
| Visible byline/source words | Visible text contains one of six words | Keyword hit, not a verified byline |
Result: 14 pages have a signal, 22 do not
| Signal | Pages | Of 36 | Note |
|---|---|---|---|
| meta author | 1 | 2.8% | Holds a website vendor string |
| article:author | 0 | 0% | — |
| JSON-LD author | 9 | 25.0% | 9 organizations, 0 persons |
| Visible byline/source words | 5 | 13.9% | Keyword hits, may be false positives |
| Any signal | 14 | 38.9% | Union, pages may appear in several rows |
| No signal | 22 | 61.1% | No hit under this method |
All nine JSON-LD authors are company entities, not people. Even when a search engine reads the author field, it gets an organization rather than a person. The single meta author hit is also not an article author; it is website vendor information.
By site: only two sites carry a JSON-LD author
| Site | Pages | Any signal | meta author | JSON-LD author | Visible words |
|---|---|---|---|---|---|
| E03 | 5 | 0 | 0 | 0 | 0 |
| E06 | 5 | 1 | 0 | 0 | 1 |
| E09 | 5 | 5 | 0 | 5 | 0 |
| E14 | 5 | 1 | 0 | 0 | 1 |
| E16 | 3 | 1 | 0 | 0 | 1 |
| E19 | 1 | 0 | 0 | 0 | 0 |
| E22 | 5 | 4 | 0 | 4 | 0 |
| E23 | 5 | 0 | 0 | 0 | 0 |
| E26 | 1 | 1 | 0 | 0 | 1 |
| E28 | 1 | 1 | 1 | 0 | 1 |
E09 and E22 are the only sites with a JSON-LD author, and even there it is not on every page (E09 5/5, E22 4/5). E03 and E23 have no hit across 10 pages; E19 has one page and no hit.
"An author field" is not "a person's byline"
The key distinction is type: all nine JSON-LD authors are Organization. For a reader, a company name is not a personal byline; for a search or AI system, the article is attributed to an organization, not a person. Conversely, a page without an author field is not necessarily poor — corporate news and product announcements often have no individual author. This study only measures marker presence, not value.
What this means for acceptance
- Separate author from publisher. An Organization is the publisher/company and should not be passed off as a Person author; add an individual byline only when one exists.
- Add at least one machine-readable author signal. meta author, article:author or JSON-LD author; the value must match what the page shows.
- Keep visible and structured data consistent. JSON-LD only, with nothing visible, cannot be checked by readers; visible only, with nothing machine-readable, cannot be used by retrieval systems.
- A source word is not a byline. The Chinese words for source or editor may sit in the footer; sample and verify whether it points to this article's author or source.
- Do not invent authors to fill a field. When there is no individual author, use an organization byline and state editorial responsibility rather than fabricating a name.
Limits
- Keyword hits: visible words may be footer/navigation, not this article's byline; 5 is a hit count, not a verified count.
- Missed cases: a page showing only a person's name, without any of the six words, is not captured by the visible signal.
- No verification: we do not check whether the author is real, matches the article, or is authorised.
- meta author is ambiguous: it may be the author, editor, website vendor or a CMS default; only presence is recorded.
- Cached snapshot: fetched 2026-11-03; no redesign recheck.
- Pilot: 10 sites, 36 pages; not an industry estimate; sites anonymised; no author values published.
Data and reproduction
Anonymised data: equipment-article-author-signals-2026.csv (36 rows × 8 columns: site, page, four signal flags, JSON-LD author type, any signal). In-repo copy: docs/research/2026-11-article-author-signals.csv. No author names, company names or meta author values are published.
Probe: scripts/article_author_signal_probe.py; tests: scripts/test_article_author_signal_probe.py (no network, 15 tests: both meta orders, article:author, JSON-LD person/organization/mixed/none/bad JSON, visible words, script exclusion, any signal). Run:
python -m unittest scripts.test_article_author_signal_probe
Related: the article citation-link pilot, the article visible-text repetition pilot, and the article HTML weight composition pilot.
Sources and definitions
- Page cache: study 83, 10 sites and 36 pages (2026-11-03 08:40–08:41).
- Signals: meta author, article:author, JSON-LD author, visible words; one boolean per page.
- Measured 2026-11-06 with no network requests.