About
Corrections
If something here is wrong, we want to know. This page sets out how to report it, what happens next, and what CiderHQ commits to doing about it.
A reference work that cannot be corrected is a reference work that is slowly getting worse. Cider is a subject where errors propagate unusually well: a figure quoted without its context, a classification asserted without its system, a date attached to an invention that was actually gradual — each gets copied, and each copy makes the next one look better sourced than it is.
Reporting an error is therefore a genuine contribution, and it is treated as one.
How to report something
Write to privacy@hekno.co.uk. There is no form: the site posts nothing anywhere, holds no accounts and collects nothing about you, and adding a form purely to receive corrections would undo that.
What to include
- The page. Its address, or enough of the title to find it.
- The statement. Quote the sentence or figure you think is wrong, so there is no ambiguity about which claim is at issue.
- What it should say, as far as you can tell.
- A source, if you have one. This is the part that decides how quickly anything can change. A named document, a section or page reference, and a link if it is online. “Everyone in the trade knows this” is a real signal that something needs checking, and it is not something CiderHQ can publish.
Reports about how something is written are equally welcome — a page that reads as an encouragement to drink is an editorial error under the editorial standards, and so is a superlative or a recommendation that has crept in. Accessibility problems have their own route, on the accessibility statement.
What happens to it
- The claim is located in the data. Every statement on the site comes from a record, and the record names the sources it rests on. That is the first thing checked: what does the claim actually cite, and does the cited source support what the page says?
- The source is checked within its competence. A source may only be cited for the subjects it is registered as competent for. A surprising number of errors turn out to be this rather than a wrong fact — a claim resting on a source that is authoritative about something adjacent.
- One of four things then happens. The claim is corrected in the data; or it is re-tiered, if the evidence turns out to be weaker than recorded; or it is recorded as disputed, where credible sources genuinely disagree; or it is removed outright, if nothing adequate supports it.
- Everything built from that claim changes with it. Because pages are generated from the records, a correction reaches every page that draws on the same record — including derived material, which is regenerated rather than edited.
| What is found | What is done |
|---|---|
| The claim is wrong and a better source exists | The claim is corrected in the data, the source register updated, and the as-of date reset where the claim is time-sensitive. |
| The claim is right but overstated | It is re-tiered or re-typed — commonly from a flat statement to a hedged typical characteristic — so the phrasing matches the evidence. |
| Credible sources genuinely disagree | It is recorded as disputed, and the page explains the disagreement instead of picking a side. |
| Nothing adequate supports it | It is removed. A claim with no source is not softened into a hedge; it goes. |
| The report is right about a gap rather than an error | The gap is recorded against the source, with what is published instead — which is usually nothing. |
The standing commitment
This is the difference between a site that can be corrected and one that merely appears to have been. If a wrong figure is edited out of one paragraph but left in the record, the same figure comes back on the next page generated from it, in a related list, in a comparison, in a derived result — and it comes back looking freshly written rather than like the error it is.
It also means a correction is auditable. The record changed, the sources it cites changed, and the automated checks that run before anything is published — data plausibility, source competence, claim publishability, editorial neutrality — run against the corrected record rather than against a sentence.
Two smaller commitments follow from the same principle. A claim that turns out to be unsupported is removed rather than downgraded into a vague hedge, because vagueness is how an unsupported claim survives. And where an error was material and someone might have relied on it, the corrected page says what changed rather than quietly presenting the new version as though it had always said that.
What is not a correction
- Disagreeing with a sensory description. Trained tasters legitimately disagree about the same glass, which is why sensory statements are framed as perception rather than as properties of the liquid. If a descriptor is attributed to a mechanism that does not produce it, though, that is a correction.
- A measurement differing from your own. Two analyses of the same cultivar from different sites and decades are two measurements, not a contradiction. If you have a sourced analysis CiderHQ does not hold, that is an addition and it is welcome.
- Objecting that a producer or brand is missing. Producers appear only where historically, legally or technically necessary. There is no listings process, because there are no listings.
- Asking for a ranking or a recommendation. The site does not hold one and will not add one.
What we can and cannot promise
CiderHQ is a small operation. There is no service-level commitment on replies, and a report that needs a document to be obtained and read may take a while, because that is what obtaining and reading a document takes.
What is promised is narrower and more useful: a report that identifies a real error will result in the data changing, and a report that cannot be acted on will not be quietly ignored while the claim stands. Where a claim cannot be verified either way, the honest outcome is that it is re-tiered or removed rather than left at a confidence it has not earned.
Reports go to privacy@hekno.co.uk. The grading model a correction is assessed against is at the evidence policy; the register it resolves to is at sources.