A Russian pro-government political think tank has launched what they term a “predictive and monitoring system” to monitor unsanctioned rallies and protest actions on social networks. While similar systems have been developed around the world, experts doubt this particular system is as dangerous as it sounds.
The software began a test run on May 18, crawling a section of an unidentified social network, Yevgeny Venediktov, director of Center for Research in Legitimacy and Political Protest, told pro-Kremlin daily Izvestia. Venediktov said the system would only monitor key targeted segments of networks, like groups or user profiles, collected by volunteers into a database and filtered by “sociologists and politologists.” Those collecting targets for monitoring would pay special attention to politically oriented groups, civic protest communities, and local discussion forums, watching likes and retweets of content posted by “extremist groups.”
Venediktov claims the new software, called “Laplace's Demon,” (a name borrowed from a mathematical thought experiment by 19th-century French scholar Pierre-Simon Laplace describing an omniscient “demon”) will be able to spot preparations for protests long before they happen, and could supply that information to law enforcement, academics and state officials.
How does predictive monitoring work?
Jay Ulfelder, a political scientist and research affiliate with Data-Pop Alliance, told RuNet Echo it is possible to develop systems that use social-media posts and other raw content to anticipate protest activity “with some accuracy, especially close the time of the event.”
The simplest version of this is old-fashioned eavesdropping—watch specific social media feeds or group forums for planning activity—and that probably still works in many cases. More sophisticated versions would watch those spaces and possibly many other forms of unstructured data, convert those streams to structured data through things like natural language processing and imagery analysis, and then use statistical models or machine learning to generate forecasts about the activity of interest.
Ulfelder cites the examples of EMBERS (short for Early Model Based Event Recognition using Surrogates), a predictive system funded by the US government, and the Chinese government's monitoring of social media, which anticipates collective action and censors discussions that might trigger it. Egyptian law enforcement has also attempted to build comprehensive surveillance systems to monitor social media for expressions of dissent and content “harmful to public security.” While the predictive algorithms in each case are different and not always obvious, Ulfelder says there is at least “circumstantial evidence that this can be done reasonably well.”
What will be monitored?
Although it's unclear which social networks the software is able to monitor at the moment and what the systems’ capabilities are, Venediktov told Izvestia they were planning to start monitoring Twitter in September, and called the platform one of the worst offenders in terms of hosting “extremist content.”
Мы провели исследование и выяснили, что именно эта соцсеть является не только лидером среди других социальных медиа по числу размещенных на нем ссылок на экстремистский контент, но и к тому же не удаляет их по требованию Роскомнадзора.
We did some research and found that this particular network is not only the leader among other social media in terms of the number of links to extremist content, but it also doesn't delete these links at the request of Roscomnadzor.
Twitter, Facebook, and other online social networks have been used by activists and opposition members in Russia as an alternative to mainstream media, largely co-opted by the Kremlin, to organize protest rallies and facilitate discussion of key social and political issues. While Russian media watchdog Roscomnadzor routinely demands these networks take down content related to political activity that it classifies as “extremist,” these demands often go unfulfilled and only some pages are taken down. Roscomnadzor has been especially unhappy with Twitter as the company refuses most of their demands to remove problematic content.
How serious is the threat?
Although the new software is presented as a sparkling innovation, Russian security services already have tools in their arsenal to monitor public expression on the web, like the Semantic Archive software, used by the security services and law enforcement to monitor mainstream media and the Internet, including blogs and social networks.
Andrei Soldatov, editor of the Russian intelligence watchdog site Agentura.ru, told RuNet Echo that Russian monitoring systems were first developed for monitoring structured information, such as media reports, so making them work with unstructured data such as social media content requires them to be relatively small and tuned to specific tasks. Another issue, according to Soldatov, is that these systems are only able to work with open accounts, (not private profile pages or locked Twitter accounts), and even then they often have trouble accessing content and metadata, as is the case with Facebook.
With the multitude of monitoring tools already in place at various government agencies, Soldatov believes the new Laplace's Demon software “doesn’t look very serious,” and says the think tank and Venediktov, whom he dubs “a low-level spin doctor from Novgorod” are simply looking to cash in on a lucrative trend.
This Center [for Research in Legitimacy and Political Protest] is not a big thing. It seems to me that it’s an effort to raise the public profile of the Center and find a way to government funding. Russian secret services have their own systems, like Semantic Archive, the [Presidential] Administration and the ministries use their own systems, and this Venediktov is merely trying to find his way to this profitable market.
While we wait for the details about the new predictive monitoring system to emerge, there is plenty to worry about with existing cases of social media users sanctioned for posting undesirable content and “extremist materials.” With the Kremlin pressing on social media and other websites to store users’ data inside Russia and asking them to take down increasingly more politically charged content, it seems predictive monitoring might be only one of the things Russian activists might have to contend with as they struggle with finding a place for their work and opinions online.