AI summary by TwinMind
How Data-Driven Journalism Illuminates Patterns of Injustice | Alison Killing | TED
TED
12m 43s
Introduction and Background
- Speaker (architect) explored Kashgar, Xinjiang through YouTube videos/Instagram posts from October 2017
- Documented visual signs of crackdown: checkpoints with metal detectors, ID checks, iris scans, CCTV cameras, riot police
- Open-source investigations have grown in journalism/human rights using photos, videos, digital traces
- Combined with satellite imagery, 3D modeling, traditional journalism techniques
- Brought new professionals to journalism: software developers, animators, archaeologists, architects
Xinjiang Campaign Context
- China conducting oppression campaign against Turkic Muslims, primarily Uyghurs
- Part of forcible assimilation campaign, described as genocide by several nations
- Estimated over 1 million people disappeared into detention camps
- Chinese government claims benign re-education program
- Former detainees report torture, abuse, forced sterilization of women
- Limited information due to Chinese internet control, journalist restrictions
- Journalists followed/detained
- Fake roadworks/staged car crashes to prevent access
- Locals face detention risk for speaking to journalists
Investigation Partnership and Challenges
- Speaker met Megha Rajagopalan (American journalist) summer 2018
- Megha first journalist to visit camp, visa declined after publication, forced to leave
- Other journalists visited handful of camps but fraction of believed network
- Challenge: Xinjiang size - 4x California size
- Satellite imagery solution: uncontrolled by Chinese government (US/European owned)
- Question: where to look in huge satellite imagery area
Baidu Total View Discovery
- Photographer Jonathan Browning discovered buildings photoshopped out of Baidu Total View (Chinese Google Street View equivalent)
- Industrial estates clumsily removed from ground level imagery
- Speaker realized same likely happening with Xinjiang detention camps
- Investigated known journalist-visited camp locations in Baidu
- No street level imagery available
- Light gray squares appeared/disappeared when zooming on satellite images
- Same phenomenon at all known camp locations
- Realized technique to find entire network - blank spots inadvertently revealed locations
Mapping Operation
- Worked with developer Christo Buschek (human rights documentation specialist)
- Built tools for open-source researchers to map masked-tile locations
- Worked quickly/secretly before discovery and removal
- Cross-referenced masked locations with unaltered satellite imagery
Camp Characteristics and Evolution
- Example: Former Kashgar high school became "Kashgar Vocational Skills Education and Training Center"
- Satellite imagery showed barbed wire courtyards creating exercise pens
- People in red uniforms lined up in courtyards
- Program evolved from makeshift camps in schools/hospitals to permanent, larger, higher-security, purpose-built facilities
Largest Camp - Dabancheng
- 2 miles long complex
- Covers quarter of New York's Central Park
- Thick perimeter walls, guard towers, blue factory buildings
- Estimated capacity: 40,000+ people without overcrowding
Corroboration Methods
- Government documents mentioning camp addresses
- Media reports on camps
- Interviews with former detainees in Kazakhstan, Turkey, Europe
Final Results
- Found 348 locations with camp/prison hallmarks
- Believed to be near-complete network
- Estimated capacity: 1+ million people
- Enough space to detain 1 in 25 Xinjiang residents
- Number potentially higher due to overcrowding
Ground-Level Verification
- Chinese vlogger "Guanguan" used published map to travel Xinjiang
- YouTube video showed ground-level views of facilities
- Drove past compound with barbed wire, barred windows
- Filmed "13th Division Detention Center" sign
- Walked past huge Ürümqi prison complex with hidden camera
- Climbed earth berm overlooking Dabancheng facility
- Recklessly brave - no plausible deniability for being there
- First ground-level image of new Dabancheng camp
- Corroborated satellite imagery interpretations
- Facility signs provided names and types
Broader Open-Source Applications
- Syria civil war: most documented conflict ever
- People filmed bombings, uploaded to social media
- Bellingcat investigated war crimes allegations (chlorine gas use)
- Current Russian invasion of Ukraine monitoring
- Google Maps traffic jam from Russian artillery crossing border
- TikTok videos revealing Russian troop movements
- Real-time war crimes investigation and fact-checking
Technical Requirements and Challenges
- Satellite imagery essential: high-resolution, up-to-date, monthly updates, free access
- Need affordable imagery access for all investigation locations
- Require people to take photos/videos and upload to accessible platforms
- Material preservation crucial
- Social media platforms remove violence content, even human rights evidence
- Civil society actors like Syrian Archive download/preserve material
Impact and Significance
- Provides evidence of human rights abuses previously impossible
- Shows scale beyond individual violations
- Corroborates eyewitness testimony
- Provides additional proof for stories
- Builds detailed pictures for policymakers
- Provides court-presentable evidence
- Enables accountability and action
