Policy
Evidence and sourcing
Wine attracts confident assertion. The purpose of this policy is to make it structurally difficult for WineHQ to state something more firmly than the evidence supports.
Why claims are typed
“Barolo must be aged for at least 38 months” and “Barolo tastes of tar and roses” are different kinds of statement. The first is a regulation you can look up and check. The second is a widely shared impression that admits enormous variation and cannot be checked at all. Wine writing routinely presents them in the same voice, which is how readers end up mistrusting both.
So every claim in the system carries a type, and the type dictates how it may be phrased. The rules below are not editorial suggestions — they are attached to the data.
| Claim type | What it is | How it must be phrased |
|---|---|---|
| Fact | A verifiable statement that is not seriously contested by authoritative sources. | May be stated plainly. Must still carry a source. |
| Typical characteristiccaveat required | A tendency observed across a category of wines, with real variation between producers, regions and vintages. | Must be hedged ("commonly", "typically", "most often") and must never be written as a universal property. |
| Regulated requirement | A requirement imposed by an appellation rule, wine law or labelling regulation. | Must name the jurisdiction and the regulating body, and must carry an as-of date because rules change. |
| Historical claimcaveat required | A statement about past events, people or origins. | Must avoid "invented by" formulations where the historical record is gradual or contested. |
| Recommendationcaveat required | WineHQ advising a course of action, such as a serving temperature. | Must give a range and state its assumptions rather than a single exact figure. |
| Subjective descriptorcaveat required | A sensory impression that different tasters will legitimately disagree about. | Must be framed as perception, not as a property of the liquid. |
| Structured inferencecaveat required | Generated by WineHQ's own deterministic rules from other structured data, such as a pairing suggestion. | Must be explainable — the reasons that produced it have to be displayable to the reader. |
Evidence tiers
Separately from what kind of claim it is, each claim records how good the evidence behind it is. The bottom two tiers exist so that weak material can be recorded honestly rather than either laundered into confident prose or silently dropped.
| Tier | Meaning | Published? |
|---|---|---|
| Primary authority | Stated directly by the body that defines or regulates the subject — for example the OIV, the European Commission GI register, the US TTB, or an appellation’s own consortium. | Yes |
| Corroborated | Agreed independently by two or more credible sources, without a single defining authority. | Yes |
| Expert reference | Supported by an established educational body or standard reference work, such as WSET teaching material or a recognised ampelographic reference. | Yes |
| Derived by WineHQ | Produced by WineHQ’s own documented, deterministic rules from better-evidenced inputs. Always explainable. | Yes, marked |
| Common practice | Widely accepted practice among people who work with wine, without a formal authority behind it. | Yes, marked |
| Needs verification | Recorded but not yet checked against a source of adequate authority. | No |
| Disputed or variable | Credible sources genuinely disagree, or the correct answer legitimately varies by context. Published only when the disagreement is itself explained. | Yes, marked |
| Do not publish | Known to be inaccurate, or barred by WineHQ content policy. | No |
Freshness
A statistic is true on a date, not permanently. Every claim is classed by how quickly it goes out of date, and volatile claims must carry the date they were checked — which is why the UK industry figures on the England and Wales pages are rendered with their dates attached rather than as plain prose.
Stable
no reviewNot expected to change. New scholarship can still revise it, but no routine review is scheduled.
For example: Established grape parentage; Dated historical events; The mechanism of alcoholic fermentation.
Slow-changing
review every 24 monthsChanges occasionally, typically through regulation or long-term planting trends.
For example: Appellation rules; Permitted grape varieties; Classification systems.
Volatile
review every 12 monthsChanges at least annually. Must always be published with an as-of date and is flagged stale once the review window passes.
For example: Harvest and production volumes; Planted vineyard area; Current legislation and duty rates; Retail availability.
What is never published
- Invented numbers. No fabricated prices, vintage ratings, production volumes or search volumes. Where a figure is not known, the field says unknown rather than guessing — the demand registry has thousands of such fields and every one of them is honest.
- Other people’s work. No critic scores, no reproduced tasting notes, no scraped databases, no copied maps or photography. Facts are extracted from sources and explanations are written here.
- False precision. No decanting time to the minute, no serving temperature to the degree, no confidence about parentage where the DNA work is unresolved.
- Health benefits for alcohol. Covered separately in the alcohol and health policy, and enforced by an automated screen rather than by care alone.
123 authorities are registered, of which 37 record a date on which they were actually consulted. The remainder are listed as the appropriate authority for their subject with no consultation date, because claiming to have read something is exactly the kind of small dishonesty this policy exists to prevent. See the full register →