Signal · App layer
What the Chinese App Store will not offer you
Two instruments on the same layer. One is ours: a small panel of apps we check every few hours against both the Chinese and the American storefront, built to catch the day a specific app disappears. The other is GreatFire's census of the whole catalogue, which is the only way to see what the removals are about. They answer different questions, and their denominators are not the same number.
Open source, MIT licensed every finding ships its raw evidence Data and methods Fund this work
Read this before any number below
Delisting is not network blocking
These are two separate controls, applied by two separate parties, and they do not line up. Apple removes an app from a storefront. The network blocks traffic. An app can be subject to either, both, or neither, so a storefront reading tells you nothing about whether anything is reachable.
Listed, and blocked anyway
Facebook, X and YouTube remain listed in the Chinese App Store. You can download all three. None of them reaches its servers from a Chinese network. The store says yes and the network says no.
Present, and useless
The converse also holds. An app can be sitting in the storefront, installable and current, and still be unusable because the thing it talks to is unreachable, throttled, or has been made to answer differently.
This page measures the app layer only. It is a record of what a company was compelled to withdraw from a catalogue. It is not a measure of what a person in China can reach. For the network layer, see the Great Firewall reading. Reading one as if it were the other is the single most common mistake made with this kind of data, and we would rather lose the cleaner headline than let it happen here.
Two instruments, two denominators
Ours · first party · every 6 hours
Live storefront panel
A hand-curated panel of apps checked against Apple's public lookup endpoint in the Chinese storefront and the American one, with control apps in the panel to prove the method is working before any reading is published. Its job is to notice a single delisting on the day it happens, and to name it.
Denominator: the tracked apps in our own panelThird party · GreatFire · daily
Whole-corpus census
GreatFire tests the catalogue at a scale we could not gather ourselves, across every surveyed storefront, and tags a portion of what is missing. Its job is scale and composition: not which app went, but what the removals are for, and how China compares with everywhere else.
Denominator: the apps GreatFire testedThe two are deliberately kept apart. The census does not feed the panel's rate, and the panel does not adjust the census. The feed states the relationship itself, and it is rendered verbatim with the census below.
Our live panel · CN storefront against US
Reading
How much of the panel the Chinese storefront will not carry
The rate below is the share of tracked, comparable apps that Apple offers in the American storefront and does not offer in the Chinese one. Control apps sit in the panel to catch a broken method, and are excluded from the rate.
Live panel
Loading the storefront reading from
/readings/app-storefront-latest.json.
Corpus scale · GreatFire's census
Reading
How much of the whole catalogue is missing, and what it is about
A share of unavailable apps says how much is gone. The category tags say what the removals were for, and the ordering of those tags is the finding. The peer ranking is there because a share is only meaningful against the spread.
Census
Loading the corpus reading from
/readings/apple-censorship-latest.json.
What Palimpsest is
A censorship observatory
Public reads only, on a schedule, with the raw file published beside every number. Watch the censor, never the censored.
A verifiable eval registry
Pre-registered model evaluations, hash-chained, with chain roots anchored externally so no result can be quietly revised.
A funded public good
MIT licensed and developed in the open. Never a commercial product, and it never monetizes the people or topics it observes.