Voidly’s category breakdowns — “Iran blocks 110 news sites,” “Russia leads on human-rights blocking” — are real, but they answer a narrower question than they appear to. They describe what gets blocked among the sites that are tested, and the sites that are tested are a deliberately curated, politically-weighted list. An honest observatory has to say so, because it changes how the counts should be read.

The tell is in what’s accessible

If censorship measurement sampled the whole web, the sites confirmed accessible in a country would look like the web: shopping, banking, health, entertainment. They don’t. In Iran, the domains Voidly confirms accessible are themselves dominated by news (94) and human rights (57), with commerce, health and government barely represented (1–2 each). The accessible set and the blocked set are made of the same kind of site — because both are drawn from the same test list. Iran isn’t leaving banking and shopping conspicuously open; those categories are simply not what gets measured.

Why: the test lists are built to find censorship

Voidly aggregates OONI measurements, which test the Citizen Lab test lists — community-maintained lists of URLs selected for their likelihood of being censored: news, human rights, political criticism, LGBTQ+, religion, circumvention tools, plus gambling and adult content. That is exactly the right design for catching censorship, but it means the category mix of the corpus is set partly by the list authors, not only by the censor. “NEWS is the most-blocked category” reflects, in part, that news is one of the most heavily tested categories.

What this does and doesn’t change

It does not make any individual finding wrong: a domain confirmed blocked across three independent networks is genuinely blocked, whoever put it on the list. The intent-profile, category-leaders and co-blocking results all hold as statements about the tested universe. What it changes is the denominator: read “Iran blocks 110 news domains” as “of the news domains on the test list, Iran blocks 110” — a lens, not a census of the entire internet. Cross- country comparisons stay fair (everyone is measured against largely the same global list); absolute “fraction of the web” claims are the ones the data cannot support. This is the same honesty the observatory applies elsewhere: the confirmed-block map is a measurement-density map, the category counts are a test-list-shaped map, and saying so is the point. Live: /v1/measurement/category-leaders and /v1/measurement/intent-profile.