Measure internet censorship
Internet censorship measurement tools like OONI probe network-level blocking of websites and services, helping map how states restrict access at the IP and DNS layers. Commenters contrast this with harder-to-measure forms of control such as social media moderation, platform bans, and legal pressure in both authoritarian regimes and democracies, debating what should count as “censorship” and how biased test lists can skew results. There is broad agreement that even necessary or popular restrictions—whether for copyright, child safety, or national security—ought to be transparently measured and documented.
Platform vs. Network-Level Censorship
- Many argue the most pervasive censorship is inside platforms (Twitter, Reddit, Facebook, etc.), not at the ISP/firewall level.
- Examples cited: old Twitter blocking a major newspaper story (even in DMs), Reddit mods deleting/shadowbanning posts, and broad moderation of certain political viewpoints.
- Others respond that this project focuses on network-layer blocking (IP/DNS/transport), not application-layer/platform decisions.
Moderation vs. Censorship
- One camp: platform moderation is not “censorship” but editorial choice; users can go elsewhere.
- Opposing view: deleting posts for political reasons is censorship, even if not by governments; users experience it as suppression of viewpoints.
- Some distinguish “good” moderation (spam, CSAM, threats) from “bad” moderation (removing disfavored news or opinions).
- Several note that essentially all public spaces online are moderated; truly unmoderated spaces are undesirable or unusable.
Authoritarian vs. Democratic Contexts
- Some stress that those in democracies underestimate what pervasive, Iran/China-style censorship feels like.
- Others see trends of increasing authoritarianism (e.g., press access limits, state pressure on media and critics).
- Counterpoint: some perceive censorship declining on major platforms after policy changes.
Bias, Scope, and Measurement Limits
- Concern that OONI’s test lists skew toward “dictatorship” blocks while under-testing sites blocked in democracies (e.g., copyright-violating libraries).
- Disagreement over definitions: some say copyright and geofencing are not censorship; others say any content-based blocking is.
- Several note that platform-level suppression and self-censorship are hard to measure; OONI is explicitly about measuring blocking, not upstream pressure.
Safety, Participation, and Technical Aspects
- Some are reluctant to run probes in heavily censored countries (e.g., China) due to perceived risk or perceived obviousness of results.
- Others argue that documenting which domains are blocked, where, and how often is precisely the point.
- Suggestions include testing for net neutrality violations (latency/throughput by protocol/port, traceroute per-port) and referencing related measurement projects (RIPE Atlas, Quack, DNS manipulation studies, circumvention-tool reachability, Psiphon).