Skip to content
PALIMPSEST / CHINA CENSORSHIP ANALYSIS

A current reading across ten declared instruments

China censorship today: the board synthesis is unavailable

7 of 10 selected censorship instruments are current. The evidence spans attention, silence, network interference, erasure, and app distribution, but their denominators remain separate.

Structured article
  • 218directive terms rankeddocumented and collected provenance only
  • withheldblackout topicsunder the predeclared rule
  • 28.6%fused network indexmethod-specific, not national traffic
  • 41.6cross-layer erasure indexdirectional co-movement

Availability warning. board-alarm, coverage-guard, silence-index did not publish a current finding in this edition. The article reports those gaps instead of retained values.

China's information controls must be read as separate content, network, erasure, and distribution measurements rather than one seductive national score.

01

The board detects movement, not motive

No current finding is published for Board alarm because the source status is stale. The board statistic tests whether named signals moved beyond their own histories; it does not identify one cause for the movement. No current finding is published for Coverage guard because the source status is stale.Receipts chinaevidence-886fc4c9f43bd80ad6a1, chinaevidence-aa296dabcfde04b7539f

02

Attention, silence, and erasure answer different questions

The current deletion-directive index ranks 218 terms with documented censorship provenance. No current finding is published for Silence index because the source status is stale. The current cross-layer erasure index is 41.6. The directive index counts provenance-bound terms, the silence index tests a predeclared blackout rule, and the erasure index summarizes heterogeneous layers; placing them together does not create one censorship rate.Receipts chinaevidence-4bef7722a9f13ac6e0b1, chinaevidence-f39c5d3e9e1e0539cf8e, chinaevidence-f58d7a794c4a86527e7e

03

Network measurements agree on scope, not on one denominator

The current coverage-weighted fusion of reporting vantages is 28.6%. The current OONI aggregate reports a 58.3% anomaly index across 272,935 completed China measurements. The current controlled panel reports 100% blocked responses across 5 tested domains. The fused index, volunteer-probe aggregate, and fixed inside-China panel use different samples and protocols, so none can be substituted for a national share of traffic or users.Receipts chinaevidence-144772da33bec275904d, chinaevidence-4e5f972900017ee51297, chinaevidence-ec56f38af9bcc0320ec8

04

App availability needs both a fixed panel and a broad corpus

The current panel finds 60% of 10 tracked apps unavailable in the China storefront. The current AppleCensorship corpus reports 28.13% unavailable among 110,432 tested apps for mainland China. The fixed panel supports controlled same-round comparison, while the broad corpus improves catalogue coverage; neither percentage identifies why an individual app is absent.Receipts chinaevidence-a3c1c782d89f5c385775, chinaevidence-d003e34c274e42b955d1

05

What today's reading can support

At the edition clock, 7 of 10 selected instruments publish current claims and 3 publish availability warnings instead. The defensible conclusion is a layered description of current observations, not a single estimate of censorship prevalence or intent.Receipts chinaevidence-144772da33bec275904d, chinaevidence-4bef7722a9f13ac6e0b1, chinaevidence-4e5f972900017ee51297, chinaevidence-886fc4c9f43bd80ad6a1, chinaevidence-a3c1c782d89f5c385775, chinaevidence-aa296dabcfde04b7539f, chinaevidence-d003e34c274e42b955d1, chinaevidence-ec56f38af9bcc0320ec8, chinaevidence-f39c5d3e9e1e0539cf8e, chinaevidence-f58d7a794c4a86527e7e

Adversarial reading

What else could explain the same observations?

Counterreadings

Limits that stay attached

Evidence ledger

Every sentence routes back to a current aggregate receipt.

board-alarmstale

Board alarm: no current finding

No current finding is published for Board alarm because the source status is stale.
Source clock
24 Aug 2026 · 06:38 UTC
Input SHA-256
d88df12f13c3186c723cab50c0e97b23e1863fa87e0917e0a403087ea02dc4b1

Interpretation limit. Current finding withheld: freshness deadline has passed

Open exact reading

coverage-guardstale

Coverage guard: no current finding

No current finding is published for Coverage guard because the source status is stale.
Source clock
24 Aug 2026 · 06:38 UTC
Input SHA-256
94ddabab22d710823ebd216ab693bae01ac32c6647408f64964968c2b5b9752f

Interpretation limit. Current finding withheld: freshness deadline has passed

Open exact reading

ddtilive

The directive index now tracks 218 censorship terms

The current deletion-directive index ranks 218 terms with documented censorship provenance.
Source clock
24 Aug 2026 · 04:07 UTC
Input SHA-256
39563b64c12df0bec07120b57c2204fa1320c42c4c47c1ecfee9f9566c6916d3

Interpretation limit. The index reflects documented and collected directives; it is not a census of every censored term.

Open exact reading

silence-indexstale

Silence index: no current finding

No current finding is published for Silence index because the source status is stale.
Source clock
24 Aug 2026 · 06:04 UTC
Input SHA-256
dd14507946d943cbe74123145c5e5ea8a1fb940bbcd815530a29ca83981de586

Interpretation limit. Current finding withheld: freshness deadline has passed

Open exact reading

vantage-fusionlive

Reporting network vantages combine to a 28.6% interference index

The current coverage-weighted fusion of reporting vantages is 28.6%.
Source clock
24 Aug 2026 · 02:27 UTC
Input SHA-256
9fb06968248eb124c92f21306f1db07a05fa41d4e596433c5ffde2743fb56a6b

Interpretation limit. The fused index summarizes differently sampled methods and is not a direct percentage of China's internet traffic.

Open exact reading

ooni-gfwlive

OONI's China anomaly index is 58.3% across 272,935 completed measurements

The current OONI aggregate reports a 58.3% anomaly index across 272,935 completed China measurements.
Source clock
24 Aug 2026 · 02:00 UTC
Input SHA-256
90accfb51b97976a774e922f9874ab3af2d4fe9a81b934354dfa714981bf1009

Interpretation limit. Volunteer-probe coverage is uneven across place, network and test family; anomalies are not all attributable to censorship.

Open exact reading

inside-viewlive

100% of the 5-domain mainland panel is blocked

The current controlled panel reports 100% blocked responses across 5 tested domains.
Source clock
24 Aug 2026 · 02:10 UTC
Input SHA-256
9fe7b2157fc21a8b26214153b614b0c01fe9f8b5e9839f45e753aa6546831291

Interpretation limit. The volunteer panel is small and availability varies by region; it does not represent every mainland network.

Open exact reading

erasure-observatorylive

The composite erasure index stands at 41.6

The current cross-layer erasure index is 41.6.
Source clock
24 Aug 2026 · 06:31 UTC
Input SHA-256
c3bb372cc4d6bff859ff39f73ade3a53dc95909b13894b75ed5f5504b1b45622

Interpretation limit. The index combines heterogeneous erasure layers and indicates co-movement, not a count of deleted items.

Open exact reading

app-storefrontlive

60% of the 10-app panel is unavailable in China's storefront

The current panel finds 60% of 10 tracked apps unavailable in the China storefront.
Source clock
24 Aug 2026 · 02:13 UTC
Input SHA-256
901254f8220d2c9ddea70d229be60dffb3bd3cab5bbbb3518119f918120e7ba0

Interpretation limit. A fixed app panel is transparent but not representative of the entire App Store catalogue.

Open exact reading

apple-censorshiplive

28.13% of 110,432 tested apps are unavailable in mainland China

The current AppleCensorship corpus reports 28.13% unavailable among 110,432 tested apps for mainland China.
Source clock
24 Aug 2026 · 06:06 UTC
Input SHA-256
5967cfb9da07bcd3996168364d7e5e3f106a4611a965970535490d50942a568f

Interpretation limit. Storefront unavailability can have policy, licensing, commercial or technical causes; the percentage alone does not identify cause.

Open exact reading

Method

A closed path from instrument to article.

  1. 01

    Validate

    Accept only the closed aggregate newsroom feed and its current availability semantics.

    chinaevidence-886fc4c9f43bd80ad6a1
  2. 02

    Project

    Copy each selected claim, metric, timestamp, input hash, and first interpretation limit without rewriting the evidence.

    chinaevidence-aa296dabcfde04b7539f, chinaevidence-886fc4c9f43bd80ad6a1, chinaevidence-4bef7722a9f13ac6e0b1, chinaevidence-f58d7a794c4a86527e7e, chinaevidence-ec56f38af9bcc0320ec8, chinaevidence-144772da33bec275904d, chinaevidence-4e5f972900017ee51297, chinaevidence-f39c5d3e9e1e0539cf8e, chinaevidence-a3c1c782d89f5c385775, chinaevidence-d003e34c274e42b955d1
  3. 03

    Compare

    Place instruments beside one another only when the prose preserves their separate samples, units, and denominators.

    chinaevidence-4bef7722a9f13ac6e0b1, chinaevidence-ec56f38af9bcc0320ec8, chinaevidence-f39c5d3e9e1e0539cf8e, chinaevidence-d003e34c274e42b955d1
  4. 04

    Withhold

    Turn every non-live story into an explicit availability warning and never reuse its retained metric as current.

    chinaevidence-886fc4c9f43bd80ad6a1

Quality gate

Why this edition was allowed to publish.

  • Every analytical input is a declared aggregate newsroom story

    10 of 10 required signal projections are present.

    closed-source-set
  • Non-live instruments publish availability, not retained findings

    3 selected instruments currently carry availability warnings.

    availability-honesty
  • Every analytical sentence names exact evidence receipts

    16 of 16 analytical sentences carry citations.

    sentence-citations
  • Incompatible instruments are not collapsed into one censorship rate

    Attention, silence, network, erasure, and distribution retain their own claim boundaries.

    denominators-separated
  • No interviews or free-form model prose are represented as reporting

    Generated from validated aggregate Palimpsest newsroom stories with a deterministic editorial template. No interviews and no free-form model prose were used.

    bounded-authorship