Net Neutrality Monitor analyses how Internet Service Providers resolve, block and inject DNS traffic. The platform tracks servers under examination, surfaces country-level reports, and publishes a live blacklist of injected addresses.
View Country ReportsWhat the platform does
Net Neutrality Monitor provides real-time analysis of the censorship systems used by Internet Service Providers. It tracks DNS servers currently under examination, lists known DNS servers that respond correctly to specific tests, and produces country reports on the types of blocking detected — from gambling and file-sharing to streaming and image hosting.
- DNS probes
- Country reports
- Injected addresses
- ISP scoring
Coverage across the monitored regions
Country reports are available for China, Colombia, Denmark, Estonia, Finland, Italy, Korea Republic of, Sweden, Switzerland, Thailand, Turkey and additional regions as new probes are activated.
- 11+Countries with published reports
- LiveDNS server list under examination
- CC BY 2.5 IT / BY-SA 3.0Content licensing applied
- OpenDonations and probe submissions
How CDN Architecture Shapes the Censorship Users See
When a website sits behind a content delivery network, its visitors may never connect to the server that actually hosts the content. A request from Brisbane could be answered by an edge node in Sydney, while another request from Darwin may travel through a different route, resolver, or security layer. That distributed design improves speed and resilience, but it also changes how filtering is experienced. Learn more about How To Test Your Isp S Dns For Censorship Without Any Technical Skills.
Censorship can therefore look like a broken website, a DNS error, a timeout, a certificate warning, or an ordinary-looking refusal page. The visible result depends on where interference occurs and how the CDN handles the request. A blocked domain may still resolve correctly, while a shared CDN address may be restricted even though many unrelated sites use it.
For researchers, administrators, and everyday users in Australia, this makes simple tests less reliable. An NBN connection in Melbourne, a Telstra mobile connection near Cairns, and public Wi-Fi in Perth can produce different results for the same URL. Comparing the path, DNS answer, HTTP response, and timing gives a much clearer picture than treating every failure as the same kind of block.
| CDN or filtering condition | What the user may see | Likely interpretation |
|---|---|---|
| DNS resolver returns no address | “Server not found” or DNS failure | Domain-level DNS interference or resolver malfunction |
| CDN edge refuses the request | Branded denial page or HTTP error | Filtering at the edge, WAF, or hosting layer |
| Shared IP address is blocked | Several unrelated sites fail | IP-based overblocking |
| Origin is unreachable | Timeout, reset, or 5xx error | Routing problem, outage, or origin restriction |
| TLS or SNI is inspected | Handshake failure or connection reset | Filtering based on encrypted connection metadata |
| Cached block response is served | Fast, consistent refusal | Edge cache or intermediary returning a stored decision |
Why CDN Design Changes Visible Symptoms
A CDN places reverse proxies between visitors and an origin server. These proxies terminate connections, select nearby points of presence, cache files, apply bot controls, and forward selected requests to the origin. The user normally sees the CDN’s IP address and certificate, not the origin’s network location.
That arrangement creates several possible enforcement points. A DNS provider can prevent the hostname from resolving; an ISP can restrict an address; a CDN can deny a path after inspecting the hostname and request; or the origin can reject traffic from particular countries. Each point produces a different external symptom, even when the user describes all of them as “the site being blocked.”
CDN behaviour can also resemble the way digital services hide complex back-end systems behind a simple login. An overview of digital ID access shows how a single user-facing action can rely on several interconnected services. Censorship analysis has a similar problem: the browser’s error message is only the final visible step in a longer chain.
Where Filtering Can Enter The Request
DNS is often the first point of observation. When a browser asks a resolver for example.com, the resolver may return the genuine CDN address, a synthetic address, no answer, or an answer that varies by location. Australian ISPs may use filtering systems associated with court-ordered site blocking, while public or workplace networks can apply their own policies. A resolver response alone cannot establish who made the decision.
The next layer is routing and IP access. Blocking a CDN address is technically easy, but a single address can serve hundreds or thousands of domains. This creates collateral damage, especially when small publishers share infrastructure with large services. Anycast routing adds another variable: the same IP may lead to different facilities depending on the network, geography, and current routing conditions.
HTTP and TLS introduce further distinctions. A CDN may identify the requested hostname through TLS’s SNI field, then return a refusal for one domain while serving another on the same address. A web application firewall can block a URL pattern, user agent, country, or suspected automated request. The result might be a 403 page, a branded “access denied” screen, or a connection reset rather than a conventional censorship notice.
Why Australian Tests Can Disagree
Australia’s network geography matters. Sydney and Melbourne host dense concentrations of cloud and CDN infrastructure, so users there may reach an edge location quickly. Someone in regional Queensland, Western Australia, or the Northern Territory may be routed through a different metropolitan hub. A website can therefore appear available in one state and unreachable in another without any change at the origin.
The access method matters as well. NBN providers, mobile networks operated by Telstra, Optus, or Vodafone, and smaller resellers can use different recursive DNS services and traffic policies. A person testing on home broadband in Adelaide might receive a normal answer, while the same phone tested on mobile data in Hobart receives a refusal. Switching browsers is less informative than switching the network or resolver.
Caching can preserve a misleading result. If an edge node stores an error response, a user may repeatedly receive the same block page after the underlying condition has changed. Conversely, a request sent to a fresh edge may reach the origin successfully. This is why a single screenshot from a café in Sydney cannot represent national availability.
Reading Evidence Without Overclaiming
A useful investigation records more than whether a page loaded. Note the queried hostname, resolver, returned addresses, time, network type, HTTP status, response headers, certificate details, and whether the content came from an identifiable CDN. Repeating the test from several Australian locations helps separate a local configuration issue from a broad access restriction.
A “resolved” result is also easy to misunderstand. It generally means that a monitoring probe received a DNS answer or otherwise completed the resolution step; it does not guarantee that the page loaded, that the address was reachable, or that the content was permitted. The explanation of resolved status is useful when interpreting monitoring records without treating one field as a final verdict.
The most valuable evidence is often longitudinal. An isolated failure could be a CDN outage, a broken origin, or a transient route problem. Repeated measurements across providers and dates show whether a pattern is stable. Historical records also help researchers distinguish policy changes from ordinary operational faults; this is the wider value described in historical censorship data.
Signals Worth Recording
- DNS answer type and returned IP addresses
- Resolver, ISP, access method, and approximate location
- HTTP status, block-page wording, and response headers
- Timing, repetition count, and test date
Clues That Need Caution
- A timeout may indicate congestion or an unavailable origin
- A shared IP failure may affect unrelated domains
- A successful DNS answer does not prove unrestricted access
- A CDN error page may reflect security policy rather than censorship
Building A More Reliable Monitoring Picture
A sound monitoring system treats the CDN as part of the measurement environment. It should compare multiple resolvers, networks, and probes, while preserving raw responses instead of recording only “blocked” or “accessible.” This approach is particularly important for volunteer projects, where a small number of probes can otherwise make local behaviour look like a country-wide result.
User-submitted URLs also require care. A URL may redirect to another hostname, load some assets while blocking the main document, or behave differently when a query string is present. Services such as Net Neutrality Monitor can help organise DNS, blacklist, probe, and statistical observations, but results remain indicative and should be read alongside network context.
The human impact is easy to miss in technical data. A blocked health, education, community, or support resource can affect someone who has few alternative connections. Organisations such as Girl Force illustrate why access to online communities and information matters, particularly when users depend on a small number of trusted websites. A CDN-level decision that looks minor in a log can become a serious barrier for a person on limited mobile data.
CDN architecture does not create censorship by itself, but it changes its shape. Distributed edges, shared IPs, encrypted hostnames, cached responses, and location-aware routing can make the same policy appear inconsistent or ambiguous. The practical way to investigate a suspected block is to repeat the request across networks, record DNS and HTTP evidence separately, compare the CDN path, and treat each observation as part of a wider pattern rather than a standalone verdict.
Transparent, analytical, community-run
The platform documents how ISPs handle neutrality on the wire, with method notes, country breakdowns and a public blacklist of injected addresses. — Project methodology






