We spent months digging through a 700 gigabyte cache from within Russia's de facto internet intel agency Roskomnadzor.
It gave us one of the most complete pictures yet of Putin's efforts to control the internet. It is at turns farcical and terrifying. nytimes.com/interactive/20…
As Russia's invasion of Ukraine proceeded, local officials in Bashkortostan, a republic east of Moscow where the files are from, noted down in detail online discontent and protests. They tallied views, likes, specific criticism of Putin, and updated dossiers of worst offenders.
The attention to detail from one small team in one tiny part of Russia is startling. They chronicle anti-war walks and complaints about inflation. At times the reports sound like weather forecasts. “Calm with separate minor pockets of tension,” one said after a dissident arrest.
They worked closely w/ security forces. Charts mapped how often they should report back to KGB successor, the FSB. In an older letter to FSB they worried Navalny's supporters were uniting "various small oppositional regional communities into a ‘united front.’” Part of the letter:
Yet there was also a doofus element to their work. In China much of this is part automated. In Russia, much was done by hand. Elaborate flow charts mapped the social media accounts of key figures and media sites, describing them as oppositional, neutral or pro government.
The censors emerge as bureaucrats dutifully doing a grinding job. Often they took screen caps of their own screens with the clock open on the bottom left to show when they had identified a particular criticism or curse word.
Other times they recorded their screen to report back videos. Some were clips of gangs or police bribery. Others were more mundane. This is of a young rapper who went viral. Bashkortostan is a center for Russian rap. Watch the censor's mouse below, sound on for a ripping rap.
The censors are the security state middle class of Russia. As they monitored dissidents and helped with arrests they also joked around. Here's a goofy video they filmed, where they joke about accidentally blocking the Kremlin website.
The video, made for women's day, is special. At one point the lawyers get too drunk to work and ask a judge to win the case for them, offering a bribe of alcohol and chocolate. In another they send porn off to hq accidentally. One source we showed it to joked some had "FSB faces"
Yet even as they recorded these, they were the tip of the online spear for a security state busy crushing any resistance to Russia's invasion of Ukraine. When a one-woman anti-war protest appeared, they collected records. The woman was arrested, given a drug test, and fined.
The system is sweeping into action now to crush opposition to Russia's mobilization. Powered as much by determination as sophistication, it shows old-fashioned bureaucracy, mixed with brutality, can be a powerful autocratic weapon, even in the digital age. nytimes.com/interactive/20…
• • •
Missing some Tweet in this thread? You can try to
force a refresh
As Russia's invasion of Ukraine stalled, its television propaganda fired on all cylinders. In recent months we went through 1000s of emails from Russia's biggest state broadcaster to learn how. It was a master class in constructing an alternative reality. nytimes.com/2022/12/15/tec…
Each day Russian producers sent incredibly detailed lists of memes and media that could be used to undercut the West. Often that included right-wing cable TV and misleading memes, like this one, showing empty grocery shelves in the US:
Tucker Carlson was a favorite. “Don’t forget to take Tucker” producers wrote to a state-media journalist in the US, pointing to a clip where Tucker warned how opposing Russia and China could end the dollar’s status as a world reserve currency. They often used other clips too:
It’s day 3 and still videos of protests spread on Chinese social media. We see some familiar tricks, people use filters and take videos of videos to bypass AI models designed to identify sensitive videos. Still, the scale of sharing is likely overwhelming censorship manpower.
It’s difficult to censor video. AI trained software can find specific videos, or things in videos (like candles and tanks). But it’s costly to train new algos and the diversity of protest videos makes it very tough. Thus you need people to check, that takes time.
Tactics like the above, video of a video, trip up the algos too. Ultimately, it’s not that the censorship apparatus is failing, it’s just hit it’s natural limit. When you have this many people posting this much and being creative, the world’s best internet control regime loses.
Videos of protests across China are still visible on WeChat, even as they quickly become unplayable. This is definitely another Li Wenliang moment, when the full power of the world’s best censorship system battles the full fury of many Chinese. For now, the censors are struggling
It’s worth recalling since Li’s death, we’ve seen with increasing frequency online outrage that even China’s internet controls have struggled to contain. Shanghai lockdowns, Guizhou bus crash, now the Xinjiang fire. Each was a massive censorship event all its own.
It’s hard to know how large each given outpouring was. But the increasing frequency with which you see these mega censorship events is certainly a trend line that gets us to where we are now.
Get ready for China state-affiliated commentators to raise objections/kiss up to Elon. It would be surprising if Musk doesn’t get lobbied by Beijing to cut Twitter’s labeling of China media/officials. We’ll see how he responds. Obvi big q is also how he handles China disinfo.
Exhibit two in the trend. Again China has been massively active on Twitter with bot nets boosting state media/diplomat disinfo efforts on Covid origins, mass internments in Xinjiang and Russian propaganda like Ukraine bio labs. Musk will have a test in how to deal with it.
For examples about how this works, last year we showed how Chinese state media and embassy accounts on twitter spread YouTube videos in which minority Uyghurs read rote scripts attesting to how free they were and how great their life was: nytimes.com/interactive/20…
Awash in a sea of data, China authorities are trying to police the future.
It's not sci fi. Using vast data records on citizens, new software uses scoring and AI to predict crime and protest before they happen. Often the result is automated prejudice. nytimes.com/2022/06/25/tec…
It works like this: Police make blacklists of people they believe are suspicious: drug users, protesters, the mentally ill. Then they aim the surveillance system at those groups, collecting huge amounts of data on activity, location, relationships: nytimes.com/video/world/as…
Specialized software helps them program digital tripwires on predetermined behaviors they believe could signal a crime. If someone goes to a train station daily, but doesn't ride, they might be a pickpocket. An alarm instructs police to check on them. No warrant necessary.
China is building a new modern marvel. It's not a dam or a high speed rail, it's the most sophisticated domestic surveillance system in the world.
The scale of data collection is staggering. No biometric frontier is neglected. This is how it works: nytimes.com/video/world/as…
Our video is based on 1000s of government bidding documents. There are many takeaways, I'll run through a few. The first, is just how meticulous police are in deciding what goes where. Camera placements are thought out meticulously. Police point precisely to locations and angles.
Cameras go in hotel and hospital lobbies. In Fujian a Sheraton gave police access to the cameras INSIDE its hotel. Likely many other international hotels do too. With China boasting half of the world's one billion cameras, data collection can be immense. nytimes.com/2022/06/21/wor…