@let@token how censorship in china allows government ...censorship is not “leaky &...

How Censorship in China Allows Government Criticismbut Silences Collective Expression

Gary King, Jennifer Pan, Molly Roberts

Institute for Quantitative Social ScienceHarvard University

(talk at Emory University, 10/12/2012)

Gary King, Jennifer Pan, & Molly Roberts () Chinese Censorship 1 / 34

Chinese Censorship

The largest selective suppression of human expression in history

Content filtering: Removing individual posts by hand

Estimated numbers of censors: ≈200,000

Internal censors at individual sites like Sina.com: up to ≈1,000Internet police: 20,000–50,000

Gary King, Jennifer Pan, & Molly Roberts () Chinese Censorship 12 October 2012 2 / 34

Chinese Censorship

Internal censors at individual sites like Sina.com: up to ≈1,000

Internet police: 20,000–50,000

Chinese Censorship

Our Main Point

Prevailing Wisdom: the goal of censorship: to stop government criticism

A BBC Reporter thought he found a contradiction in Chinese Premier WenJiabao’s fairwell speech:

“[Premier Wen] claimed that China’s leaders welcomed criticism expressed onthe internet. It is ‘natural for that to happen’, he said.

And yet he presides over a government that routinely censors theinternet — and has imprisoned many who have given voice to their criticism.”

We show:

There is no contradiction; Wen Jiabao was sincereChina allows plenty of criticism in social mediaCensorship is not “leaky & imperfect”; our understanding of it isThe goal: stopping collective action, clipping social ties(They also censor: pornography & criticism of the censors)

Our Main Point

We show:

Our Main Point

We show:

Our Main Point

We show:

Our Main Point

We show:

Our Main Point

We show:

There is no contradiction; Wen Jiabao was sincere

China allows plenty of criticism in social mediaCensorship is not “leaky & imperfect”; our understanding of it isThe goal: stopping collective action, clipping social ties(They also censor: pornography & criticism of the censors)

Our Main Point

We show:

There is no contradiction; Wen Jiabao was sincereChina allows plenty of criticism in social media

Censorship is not “leaky & imperfect”; our understanding of it isThe goal: stopping collective action, clipping social ties(They also censor: pornography & criticism of the censors)

Our Main Point

We show:

There is no contradiction; Wen Jiabao was sincereChina allows plenty of criticism in social mediaCensorship is not “leaky & imperfect”; our understanding of it is

The goal: stopping collective action, clipping social ties(They also censor: pornography & criticism of the censors)

Our Main Point

We show:

There is no contradiction; Wen Jiabao was sincereChina allows plenty of criticism in social mediaCensorship is not “leaky & imperfect”; our understanding of it isThe goal: stopping collective action, clipping social ties

(They also censor: pornography & criticism of the censors)

Our Main Point

We show:

Our Data Source

Chinese Social Media: Fractured over 1,400+ sites

hi.baidu 12%

bbs.voc.com.cn 5%

bbs.m4.cn 4%tianya 3%bbs.beijingww.com 2%

Share of Posts in Our Analysis by Internet Content Provider (ICP)

(Plus 59% from Sina Blog)Gary King, Jennifer Pan, & Molly Roberts () Chinese Censorship 12 October 2012 6 / 34

Posts Collected and Analyzed

Collect 3,674,698 social media posts in 85 topic areas over 6 months

Random sample: 127,283

(Repeat design; Total analyzed: 11,382,221)

For each post (on a timeline in one of 85 content areas):

Download content the instant it appears(Carefully) revisit each later to determine if it was censoredUse computer-assisted methods of text analysis

Download content the instant it appears

(Carefully) revisit each later to determine if it was censoredUse computer-assisted methods of text analysis

Download content the instant it appears(Carefully) revisit each later to determine if it was censored

Use computer-assisted methods of text analysis

Censorship is not Ambiguous: BBS Error Page

The Censors are Fast; Our Automated Methods are Faster

Days Until Censorship, Shanghai Subway Analysis

Days After Post Was Written

0 1 2 3 4 5 6 7

The Censors are Fast; Our Automated Methods are FasterExample: Shanghai Subway Crash

0 1 2 3 4 5 6 7

The Censors are Fast; Our Automated Methods are FasterExample: Shanghai Subway Crash

0 1 2 3 4 5 6 7

For Rare Topics, Censorship Effort Appears Constant

Jan Mar Apr May Jun

Count PublishedCount Censored

Jan Feb Mar Apr May Jun Jul

Students Throw Shoes at

Fang BinXing

Pornography Criticism of the Censors

For Rare Topics, Censorship Effort Appears Constant0

Jan Mar Apr May Jun

Students Throw Shoes at

Fang BinXing

Pornography Criticism of the Censors

For most topics: Censorship & Post Volume are “Bursty”

Count PublishedCount Censored Riots in

Zengcheng

Unit of analysis:

volume burst(3 SDs greater thanbaseline volume)

We monitored 85 topicareas (Jan–July 2011)

Found 87 volume burstsin total

Identified real worldevents associated witheach burst

For most topics: Censorship & Post Volume are “Bursty”0

Zengcheng

Unit of analysis:

Zengcheng

Unit of analysis:

Zengcheng

Unit of analysis:

volume burst

(3 SDs greater thanbaseline volume)

Zengcheng

Unit of analysis:

Zengcheng

Unit of analysis:

Zengcheng

Unit of analysis:

Zengcheng

Unit of analysis:

The Goals of Chinese Censorship

Theories. The Purpose of Censorship is to:

1 Stop collective action

Either or both could be right or wrong.

1 Stop criticism of the state

1 Stop criticism of the state Wrong

2 Stop collective action Right

Our Central Hypothesis: The Rules Censors Follow

1 Monitor many broad topic areas

2 Detect volume bursts; identify associated real world events

3 Choose events that could (or did) generate collective action

4 Censor all posts within bursts associated with these events

5 Judge individual posts for association with events, not for collectiveaction potential (or for criticism of the state)

The government censors all posts (regardless of how critical orsupportive of the state) in volume bursts associated with events withcollective action potential

Our Evidence

3 increasingly specific tests based on

1 Post volume

2 Nature of the event generating each volume burst

3 Content of the posts

Our Evidence

1 Post volume

Our Evidence

1 Post volume

Our Evidence

1 Post volume

Our Evidence

1 Post volume

Test 1: Post Volume

Begin with our 87 volume bursts in 85 topics areas

For each burst, calculate change in % censorship inside to outsideeach volume burst within topic areas – censorship magnitude

If goal of censorship is to stop collective action, we expect:

1 On average, % censoredshould increase duringvolume bursts

2 Some bursts (associatedwith politically relevantevents) should havemuch higher censorship -0.2 0.0 0.2 0.4 0.6 0.8

Censorship Magnitude

Density

Test 1: Post Volume

Density

Test 1: Post Volume

Density

Test 1: Post Volume

Density

Test 1: Post Volume

Density

Test 1: Post Volume

2 Some bursts (associatedwith politically relevantevents) should havemuch higher censorship

-0.2 0.0 0.2 0.4 0.6 0.8

Density

Test 1: Post Volume

Density

Test 2: Nature of Event Generating Increases in Volume

Event classification (each of which could be negative, positive, or neutralcomments about the state)

protest or organized crowd formation outside the Internetindividuals who have organized or incited collective action on theground in the past;topics related to nationalism or nationalist sentiment that have incitedprotest or collective action in the past.

(Other) News

Government Policies

(Other) News

Government Policies

Collective Action Potential

(Other) News

Government Policies

protest or organized crowd formation outside the Internet

individuals who have organized or incited collective action on theground in the past;topics related to nationalism or nationalist sentiment that have incitedprotest or collective action in the past.

(Other) News

Government Policies

protest or organized crowd formation outside the Internetindividuals who have organized or incited collective action on theground in the past;

topics related to nationalism or nationalist sentiment that have incitedprotest or collective action in the past.

(Other) News

Government Policies

(Other) News

Government Policies

Criticism of censors

(Other) News

Government Policies

Pornography

(Other) News

Government Policies

Pornography

(Other) News

Government Policies

Pornography

(Other) News

Government Policies

Pornography

(Other) News

Government Policies

Pornography

(Other) News

Government Policies

Pornography

(Other) News

Government Policies

Pornography

(Other) News

Government Policies

Pornography

(Other) News

Government Policies

What Types of Events Are Censored?

Density

-0.2 -0.1 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8

Collective ActionCriticism of CensorsPornography

PolicyNews

What Types of Events Are Censored?

Popular Book Published in Audio FormatDisney Announced Theme ParkEPA Issues New Rules on LeadChinese Solar Company Announces EarningsChina Puts Nuclear Program on HoldGov't Increases Power PricesJon Hunstman Steps Down as Ambassador to ChinaNews About Iran Nuclear ProgramIndoor Smoking Ban Takes EffectPopular Video Game ReleasedEducation Reform for Migrant ChildrenFood Prices RiseU.S. Military Intervention in Libya

New Laws on Fifty Cent PartyRush to Buy Salt After Earthquake

Students Throw Shoes at Fang BinXingFuzhou Bombing

Localized Advocacy for Environment LotteryGoogle is Hacked

Collective Anger At Lead Poisoning in JiangsuAi Weiwei Arrested

Pornography Mentioning Popular BookZengcheng Protests

Baidu Copyright LawsuitPornography Disguised as News

Protests in Inner Mongolia

PoliciesNews

Collective ActionCriticism of CensorsPornography

-0.2 0 0.1 0.3 0.5 0.7

Censoring Collective Action: Protests in Inner Mongolia

Protests in Inner Mongolia

Censoring Collective Action: Ai Weiwei’s Arrest

Jan Mar Apr May Jun Jul

Ai Weiwei arrested

Low Censorship on One Child Policy

Speculation of

Policy Reversal at NPC

Low Censorship on News: Power Prices

Power shortages Gov't raises power prices

to curb demand

Test 3: Content of Posts

We adapted “ReadMe” (Hopkins-King) methodology to the content ofcensored posts in Chinese

Standardize encoding of Chinese text

Eliminate punctuation and stop words

Experiment with segmenting characters into words

Calculate new the quantity of interest (% censored by category)

Code a validation setReadMe estimates: % of posts in each categoryQuantity of interest: % of posts censored in a category

P(Censored |Category) =P(Category |Censored)P(Censored)

P(Category)

Code a validation set

ReadMe estimates: % of posts in each categoryQuantity of interest: % of posts censored in a category

P(Category)

Code a validation setReadMe estimates: % of posts in each category

Quantity of interest: % of posts censored in a category

P(Category)

“ReadMe” Algorithm Validated in Chinese

“ReadMe” Algorithm Validated in ChineseExample: Labor Strikes, 2010 (Training set: 100; Test set: 900)

ReadMe Results for Chinese Sampled, Not Segmented

Probability

Facts Supporting Employers

Facts Supporting Workers

Opinions Supporting Workers

Opinions Supporting Employers or Irrelevant

ReadMe Results for Chinese Sampled, Not Segmented

Probability

Facts Supporting Employers

Facts Supporting Workers

Opinions Supporting Workers

Opinions Supporting Employers or Irrelevant

ReadMeTrue

Test 3: Content of Censored and Uncensored Posts

Predictions of

Test 3: Content of Censored and Uncensored PostsPredictions from Two Theories of Censorship

Predictions of State Criticism Theory

Collective Action

yes no

State Critiqueyes

Test 3: Content of Censored and Uncensored PostsPredictions from Two Theories of Censorship

Predictions of Collective Action Theory

Collective Action

yes no

State Critiqueyes

With Collective Action, Posts that Support or Oppose theState are Censored

Criticize Support Criticize Support Criticize Support

Ai Weiwei Inner MongoliaFuzhou Bombing

With Collective Action, Posts that Support or Oppose theState are Censored

Ai Weiwei Inner MongoliaFuzhou Bombing

Without Collective Action, Posts that Support or Opposethe State are NOT Censored

One Child Policy Corruption Policy Food Prices Rise

Without Collective Action, Posts that Support or Opposethe State are NOT Censored

One Child Policy Corruption Policy Food Prices Rise

Uncensored Posts (w/o Collective Action Potential)Critical of the State

This is a city government [Yulin City,Shaanxi] that treats life with contempt, thisis government officials run amuck, a citygovernment without justice, a city govern-ment that delights in that which is vul-gar, a place where officials all have mis-tresses, a city government that is shamelesswith greed, a government that trades dig-nity for power, a government without hu-manity, a government that has no limits onimmorality, a government that goes back onits word, a government that treats kindnesswith ingratitude, a government that caresnothing for posterity. . .

Censored Post (with Collective Action Potential)Supporting the State

The bombing led not only to the tragedy ofhis death but the death of many governmentworkers. Even if we can verify what QianMingqi said on Weibo that the building de-molition caused a great deal of personal dam-age, we should still condemn his extreme actof retribution. . . . The government has con-tinually put forth measures and laws to pro-tect the interests of citizens in building de-molition. And the media has called attentionto the plight of those experiencing housingdemolition. The rate at which compensationfor housing demolition has increased exceedsinflation. In many places, this compensationcan change the fate of an entire family.

Censorship Pre-empting Collective Action: Ai Weiwei’sArrest

Mar 19 Mar 29 Apr 08 Apr 18

Actual %censorship

Predicted %censor trend basedon 3/19−3/29 data

Mar. 29,5 days prior

Apr. 3,Ai Weiwei Arrested

Placebo Test: Mostextreme of all effects

Actual %censorship

Placebo Test:

Mostextreme of all effects

Actual %censorship

Predicting the South China Sea Peace Agreement

Jun 12 Jun 22 Jul 02

Predicted % censortrend basedon 6/10−6/20 data

Actual %censorship

Jun. 20,5 days prior

Jun. 25,PeaceAgreement

Actual %censorship

Placebo Test:

Actual %censorship

Censorship Pre-empting Collective Action: Wang Lijun’sDemotion

Jan 23 Jan 30 Feb 06 Feb 13

Jan. 28,5 days prior

Feb. 2,Wang Lijundemoted

Actual %censorship

Placebo Test:

Actual %censorship

Concluding Remarks

The Chinese Censorship Program:

Stops collective action not criticismReveals government intentions; predicts actions

Our methodology:

Automated text analysis methods validated in Chinese (withouttranslation)Enables continuous time measurement of Chinese governmentcensorship activitiesReveals information about state action when traditional media is silentMight be applicable to other countries

Concluding Remarks

Stops collective action not criticism

Reveals government intentions; predicts actions

Our methodology:

Concluding Remarks

Our methodology:

Concluding Remarks

Our methodology:

Concluding Remarks

Our methodology:

Automated text analysis methods validated in Chinese (withouttranslation)

Enables continuous time measurement of Chinese governmentcensorship activitiesReveals information about state action when traditional media is silentMight be applicable to other countries

Concluding Remarks

Our methodology:

Automated text analysis methods validated in Chinese (withouttranslation)Enables continuous time measurement of Chinese governmentcensorship activities

Reveals information about state action when traditional media is silentMight be applicable to other countries

Concluding Remarks

Our methodology:

Automated text analysis methods validated in Chinese (withouttranslation)Enables continuous time measurement of Chinese governmentcensorship activitiesReveals information about state action when traditional media is silent

Might be applicable to other countries

Concluding Remarks

Our methodology:

For more information

gking.harvard.edu

people.fas.harvard.edu/˜jjpan

scholar.harvard.edu/mroberts

@let@token how censorship in china allows government ...censorship is not “leaky &...

Documents