Sampling 1-in-10,000 TCP connections at Cloudflare's 285+ PoPs (serving ~17–20% of the Internet's websites, handling 45M HTTP requests/second at average load) over two weeks in January 2023 revealed that 25.7% of all sampled connections were 'possibly tampered.' The passive technique requires no vantage points inside censored networks, covering cellular, enterprise, and low-penetration-country networks that active measurement cannot reach.

Manual analysis of 700+ unique packet groupings from possibly tampered connections yielded 19 high-confidence tampering signatures — up from 6 in prior work — covering 86.9% of all possibly tampered connections. Post-SYN signatures account for 43.2% of possibly tampered connections (99.5% matching a known signature), post-ACK for 16.1% (98.7%), and post-first-data-packet (PSH+ACK) for 5.3% (97.9%), with 19 signatures described as flag-sequence patterns of the form ⟨X→Y⟩ in Table 1.

§4.1, Table 1 detection

Post-handshake tampering signatures (⟨SYN;ACK→RST⟩ and ⟨SYN;ACK→RST+ACK⟩) constitute 34.4% of tampered connections from Iranian networks, but over 70% from Sri Lanka networks and over 81% from Turkmenistan networks, suggesting that censors in the latter two countries disproportionately block at the IP/TCP-handshake level before any application-layer content is visible — consistent with IP-list-based blocking rather than SNI-based DPI.

§4.1 detection

Censoring middleboxes predominantly use RST injection rather than in-path packet dropping because injecting forged RST/RST+ACK packets does not require the middlebox to sit in the data path — off-path copies of packets suffice. The GFW specifically injects both RST and RST+ACK packets simultaneously after an offending PSH, a known idiosyncratic signature, while Iran's censor uses post-handshake RST injection (⟨SYN;ACK→RST⟩) and packet drops at the same stage.

§2.1, §4.1 detection

Passive measurement of real user connections demonstrates that published active-measurement test lists (Citizen Lab, Herdict, GreatFire, Berkman Klein, and top-K lists) miss a considerable fraction of domains that are actively being tampered with, as confirmed in §5.5. Because passive measurement is driven by real user requests rather than an a priori domain list, it can discover blocked domains that were never included in any test list and has no dependency on volunteers providing ground truth.

§5.5, §2.2 evaluation

The largest single source of censored domains in the GNL is MESA lab's SNI monitoring dataset (E21-SNI-Top200w.txt) containing 57,362 censored domains, and E21-SNI-Top120W-20221020.txt with 36,467 domains—totaling over 93K domains from network tap data alone for a single country (E21 = Ethiopia per InterSecLab attribution). A separate Xinjiang dataset (XJ-CUCC-SNI-Top200w.txt) contains 13,604 domains. These datasets "do not seem to come from popular domain lists, and instead appear to be gathered from network taps," confirming that Geedge builds censorship target lists directly from passive traffic observation.

2026-sheffey-geedge §4.2, Table 3 dpisni-blockingmeasurement-platform

Of 6,915,266 domains extracted from the 572 GiB Geedge Networks Leak (GNL), 298,955 censored domains (93.7% of all GNL-censored domains) appear in neither Tranco top-1M nor CitizenLab test lists. Measurements across China (Guangzhou/Nanjing), Myanmar, Pakistan, and Algeria confirmed censorship via DNS injection and SNI-based TLS connection termination. The GNL covers 25–62% of Tranco-censored domains across countries, showing substantial but incomplete overlap. This vendor-side ground truth reveals a censorship surface roughly two orders of magnitude larger than curated academic test lists.

2026-sheffey-geedge §4.1, Table 2 dpisni-blockingdns-poisoningmeasurement-platform

CensorAlert aggregates censorship signals from heterogeneous open sources — including OONI, Cloudflare Radar, NetBlocks, GitHub Issues of circumvention tools (Hysteria, Xray), Net4People BBS, NTC Party forum, Mastodon, X, Telegram channels, and arXiv — normalizing each item into a common schema preserving timestamp, source, raw data, and provenance with a link to the original content.

2026-zohaib-extended §2 (Data Sources) measurement-platform

Systematic measurement platforms (OONI, Censored Planet, Cloudflare Radar, NetBlocks) have inherent blind spots due to geographic coverage and protocol-specific test constraints; critical censorship discoveries — including GFW fully-encrypted protocol blocking and regional Chinese censorship — were first surfaced by user reports on forums and GitHub issue pages of circumvention tools, not by automated measurement infrastructure.

2026-zohaib-extended §1 measurement-platformfully-encrypted-detect

CensorAlert generates text embeddings (OpenAI text-embedding-3-small) from each item's summary, title, and tags to cluster semantically similar reports within configurable time windows; near-duplicates such as reposts, copied headlines, and translations are collapsed into a single canonical post that preserves all original source URLs and metadata.

2026-zohaib-extended §2 (AI-based Scoring) measurement-platform

CensorAlert's LLM agent scores each ingested item 0–5 on five independent dimensions — credibility, novelty, impact, timeliness, and verifiability — then computes a normalized significance score (0–10) from these components; items are processed every two hours using OpenAI GPT-5 Thinking (hosted on Azure) constrained to return structured JSON output.

2026-zohaib-extended §2 (AI-based Scoring) measurement-platform