What a documentation-based record cannot establish
Everything in a register like this rests on pages somebody chose to write. What a product actually does, what a file actually carries and what a regulator will actually accept are all outside that. As of 2026-09-12.
| Question | Answerable from published pages |
|---|---|
| What has a vendor stated, and when | Yes |
| What does a product do | Only where a page describes it |
| What does this delivered file carry | No; that needs an inspection |
| Is this arrangement compliant | No, and not by any register |
Inclusion rule. Question types a register of published vendor material is asked. Order. From the answerable to the unanswerable.
1An undocumented behaviour is real and unusable
A vendor may write something into every file and never mention it. That is a real mark. It is also useless to anybody who needs to cite it, plan around it or show it to somebody who is asking.
So a register recording nothing documented is making a claim about the published record rather than about the file. Both are worth knowing and they are not interchangeable.
2Compliance is not a documentation question
Whether an arrangement satisfies an obligation depends on the arrangement, the jurisdiction, the role somebody occupies and usually on advice. No amount of vendor reading produces that answer.
What a register can do is put the published wording beside the published product material with dates on both, so the person who does have to decide is deciding from sources rather than from summaries.
3Counting is only as good as the inclusion rule
Any statement about how many entries do something requires a rule about what counts. Counting inferences alongside quotations produces a larger number and a less useful one.
Which is why a count without its rule beside it is not a finding. The rule is the part somebody can disagree with, and being disagreeable-with is what makes a number worth publishing.
4The most useful thing a register does is keep dates
Pages are rewritten without notice. A quotation with a reading date stays usable afterwards, because a reader can compare it with the current page and see that something moved.
That is a modest function and it is the one that survives. Everything else here is an argument about how to read a document, and arguments age faster than dated quotations do.
Background on mechanism and practice. Nothing here is attributed to a product, and nothing here is a reading of any instrument. The sourced material is on the generator table. Related: Comparing unlike evidence, When a cell earns a page.