Special cases: encyclopedias, error pages and forums

What the document says

Encyclopedias (9.1)

High and Highest only for encyclopedias with very good reputations for accuracy and an article that is well-researched with proper references. On Wikipedia, the guidance is nuanced: the site has a good general reputation, but nobody vouches for each article’s accuracy and quality varies — so you evaluate the article, not the domain. A non-YMYL article with satisfying information and trustworthy references: High range. Thinner MC and references: Medium. Factual inaccuracy: Low or Lowest.

Error pages and pages with no MC (9.2)

The question is whether the site owner made an effort to help the user even in failure. A custom 404 that explains the problem and helps navigation: typically Medium; with effort and original content, High — and the document cites a 404 with an original cartoon and integrated search as “High+ to Highest,” one of the most creative on the web. A broken page: Low if isolated and temporary; Lowest if the problem is systemic — the test is opening other pages on the site and seeing whether the pattern repeats.

Forums and Q&A (9.3)

The MC is the question + answers + discussion, evaluated from the perspective of someone arriving from search, not a participant. Usernames suffice even for Highest. E-E-A-T is judged from the posts themselves — experience and expertise become visible in the conversation, and other participants tend to point them out. Highest: extremely satisfying conversations with depth, unique insights or exchanges of experience inaccessible in the reader’s real-world community. Low: little participation, surface-level discussion, mild inaccuracies, or a lack of respect and decorum that would deter others from joining. A new question with no answers: Medium by default — the page needs time. A discussion that derailed but keeps valuable stretches: Medium, crediting what’s worthwhile.

How to rate without stalling (10.0–11.0)

The operational instructions carry two useful tiebreakers: torn between two ratings, use the lower one; between three, the middle one. The page’s country is not a factor (a UK site doesn’t lose points with a US rater). And the guidelines admit their own incompleteness: if the rater genuinely believes a page is High or Low for a reason not covered, they should rate it so and explain. The FAQ (11.0) closes recurring points: the steps exist precisely to prevent the superficial “does it look good?” judgment; personal opinion and experience are beneficial purposes even when they offend; alternative niches have their own experts (the acupuncture site by a renowned practitioner vs. one by someone with no experience — E-E-A-T distinguishes the two); no category is doomed — there is High and Low celebrity gossip; and the scale applies even to page types with no High examples yet: “we need a uniform set of standards that apply to all pages, even for pages that have not yet been created.”

In practice

“Evaluate the article, not the domain” cuts both ways. The same principle that limits Wikipedia protects you: your excellent content on a modest site can be High, and weak content on a strong domain can be Low. Domain authority as a blank check is a simplification the document explicitly declines to endorse.

The systemic-breakage test is a technical argument for site hygiene. One isolated error costs a Low on that URL; a pattern of errors reclassifies the site as neglected. Monitoring 404s, broken links and orphan pages isn’t perfectionism — it’s preventing isolated issues from becoming a systemic diagnosis.

For community operators, moderation is a ranking factor in writing. Lack of respect among participants is a named Low criterion — not for the content itself, but because it deters future participants. Investing in community culture and moderation is investing in the score of the thousands of pages the community generates.

The pessimistic tiebreaker favors whoever leaves no doubt. “When in doubt, lower rating” means ambiguity is expensive: the page that almost demonstrates expertise, that almost makes the owner clear, loses the benefit of the doubt by rule. Clarity of signals isn’t aesthetics — it’s the difference between the higher and lower score in a tie.