Field notesContent trustFundamentals
Authorship, sources, and dates: the three signals thin content never has
Trust signals are not a checklist to satisfy. They work because they make a claim accountable, checkable, and current — which is precisely what mass-produced content cannot do.
AI Detective7 min read
Ask why authorship matters and you will usually get a circular answer: because it is a trust signal, and trust signals help. That framing leads to bylines attached to nothing, dates that update themselves nightly, and links added to hit a quota. None of it works, and understanding why explains what does.
What these signals actually do
Each of the three answers a specific question a reader — human or machine — has about a claim.
| Signal | Question it answers | Why it resists faking |
|---|---|---|
| Authorship | Who is accountable if this is wrong? | A named person or organisation with a track record has something to lose. Mass-produced content has nobody to name. |
| Sources | Can I verify this without trusting you? | Linking to a checkable source invites verification. Content built from nothing has nothing to link. |
| Dates | Is this still true? | A visible date is a falsifiable claim about currency. Undated content avoids the question. |
Notice the pattern: each signal is cheap for genuine content and awkward for content produced at volume. That asymmetry is the whole reason they carry weight, and it is why implementing them cosmetically produces nothing.
Authorship that means something
A byline reading a name and nothing else is barely better than none. What makes attribution useful is that it connects to something:
- A page about the author explaining why they can speak to this subject
- Consistent attribution across their work, so a body of writing accumulates
- Author markup in structured data so the connection is machine-readable rather than inferred
- For organisational content, a named team or editorial function that genuinely exists
Attributing to an organisation rather than an individual is entirely legitimate, and often more honest for content produced collaboratively. What matters is that the accountable party is real and identifiable.
Sources, and the tempting shortcut
Citing sources means linking to what a claim rests on: the specification, the study, the documentation, the dataset. It is straightforward and most sites do it far less than they think.
The shortcut to avoid is treating outbound links as the goal rather than the by-product. Links to generic homepages, to sources that do not support the specific claim, or to whatever ranked well for the topic add nothing and are reasonably transparent to anyone who follows them.
Dates, honestly
Show when content was published, and when it was substantively updated. Both, distinctly. Two common failures:
Automatically bumping the date
Some systems update the modified date on every deployment or minor edit. This makes the date meaningless, and it is worse than no date because it asserts currency that was never established. Update dates when content actually changes in substance.
Hiding the date because content is old
Understandable and counterproductive. Undated content cannot be assessed for currency at all, so it tends to be treated cautiously or skipped in favour of something dated. An honest old date is more useful than none, and it signals where a genuine update would pay off.
The part that is not a technique
These signals are difficult to fake because they describe how content was actually made. A page written by someone accountable, resting on checkable sources, and maintained over time will carry them naturally. A page produced in bulk to occupy a topic will not, and adding the markers afterwards does not change what the page is.
This is inconvenient advice, because it cannot be implemented in an afternoon. It is also the reason the signals continue to work while every purely technical tactic gets arbitraged away.