Frequently asked questions
How this site tests, what its verdicts mean, and what they cannot tell you. For questions about GreatFire itself, see greatfire.org.
What is this site?
It tests whether individual web addresses can be reached from inside mainland China, and has done so without a break since 2011. Every page here is built from those measurements: the verdict for one URL, the evidence behind it, and how it has changed over the years. Nothing is crowdsourced or reported from your browser — probes inside mainland China and a control probe outside fetch the same address at the same moment, and the difference between them is the finding.
How do you test?
One test fetches one exact URL from several probes inside mainland China and from a control probe outside it, simultaneously. Each probe records how the request ended, what the domain resolved to, how many bytes arrived and where redirects led. Four detectors read those rows — connection reset, DNS poisoning, redirection to a server inside mainland China, and failure to load — and the most severe signal decides the test.
What do the verdicts mean?
Blocked: a detector proved interference. Redirected: the mainland China probes ended on a server inside the country. Intermittent: probes disagreed, or the address alternates between reachable and blocked. Not blocked: the probes matched the control. Not enough data: the test could not tell. Unresolvable: nothing anywhere resolves the domain, so there is nothing left to block — usually an expired or parked domain.
It says blocked, but the site opens for me
Three ordinary explanations. Your network may not be ours: a VPN, a corporate line, Hong Kong or Macau all reach the internet differently from our probes. Blocking is per address rather than per site, so a home page can be reachable while an article is not. And it moves by the hour — the headline figure is the share of conclusive tests over the last 90 days, not a claim about this second. The calendar on each URL page shows precisely when each test ran.
Why do http and https give different answers?
Because they are different requests and the Great Firewall treats them differently: plain http exposes the path and the page to inspection, while https hides the path but still reveals the hostname as the connection opens. We test them as separate addresses and never merge the results. Where both exist, each URL page links to its twin.
What does “not enough data” mean?
That we could not tell — not that the address is fine. It happens when the control fetch failed too, so there was nothing to compare against; when a test left no measurement rows; or when signals contradict each other without any mechanism proving itself. Detectors that need no control still convict when they see their evidence; it is the “not blocked” verdict that requires a healthy control.
Why does a verdict say “as of” an earlier year?
Because nothing conclusive came out of the last 90 days, so the newest conclusive test stands in and is dated. Most of these are addresses that stopped answering anywhere years ago. Read a dated verdict as history rather than as the present state, and use Test now if you need today's answer.
How often is an address tested?
It varies. Addresses in demand are tested many times a day; the long tail comes round over weeks. Every URL page shows its last test and its full history, and carries a Test now button that runs a fresh measurement on the spot, usually in under a minute.
Can I add an address?
Yes. Search for it, and if we have never tested it the page offers a live test. From then on it stays in the rotation and builds a history. Test the exact address you care about, including the scheme and the path — the results genuinely differ.
What are connection reset, DNS poisoning and the others?
Four ways a request can be interfered with. A connection reset tears the connection down mid-request, so nothing arrives. DNS poisoning answers the name lookup with an address that is not the site's, so the request never reaches it. A redirect inside mainland China lands the request on a server in the country instead of the real one. Failure to load is the residual case: the control downloads the page, the mainland China probes cannot. Each is a filter on the Explore page, so you can see which addresses are affected by which.
What about mobile apps?
Where app-store entries match the domain you are looking at, the URL page lists them and whether they are still available in mainland China. Store removal and website blocking are separate acts: an app can disappear while its site stays up, and the reverse happens too. That data comes from GreatFire's AppleCensorship and GoogleCensorship projects.
How far back does the data go?
To 16 February 2011, when the first test ran. GreatFire published that data as GreatFirewall.biz and took the GreatFire.org name in October 2011. Nothing is deleted since: the record now holds 23,245,058 tests across 708,309 addresses, and each URL's calendar reaches back as far as that address does.
Can I use the data?
Freely, and citation is welcome. Every URL page has a JSON view; there are endpoints for verdicts, search, domains and single tests, an OONI-compatible export of any measurement, and RSS and JSON feeds. If you cite a figure, cite its date — a verdict is a statement about a moment.
How do I get around the block?
That is Circumvention Central's question: it ranks VPNs and other tools on speed and stability as measured by users inside mainland China. What this site can tell you is whether a given tool's own website is reachable — often the real obstacle, since a provider's download and signup pages tend to be blocked well before its protocol is.
Do you test other countries?
Not here. Every verdict on this site compares mainland China against an outside control, so the method only speaks about the Great Firewall. Circumvention Central covers Russia, Iran and others through user-run tests, and OONI publishes global measurements — our OONI export exists so the two can be compared directly.
Who runs this?
GreatFire, a non-profit that has documented online censorship in mainland China since 2011. For the organisation itself — the team, the other projects, how to support or contact us — see greatfire.org.