Propaganda鈥檚 latest twisted form was laid bare in last week鈥檚 FBI indictment alleging that a Russian cabal tried to game the 2016 US election with an industrial-scale fake news operation.
Misinformation, and the war against it, is, of course, nothing new. I recall how, as a kid in the 1970s, I鈥檇 hit the wrong frequency on our kitchen radio and a deep voice would emerge from the static, explaining how badly people in the capitalist West were oppressed. It seemed hilarious. It was Radio Moscow, the voice of the Soviet Union.
Long before Facebook and Twitter delivered bogus news, doctored video and divisive social media posts to smartphones, the airwaves fizzed with such political propaganda 鈥� or fake news 鈥� from state-controlled broadcasters.
Advertisement
The difference was that back then engineers on all sides had devised methods to jam each other鈥檚 signal with random noise, speech or music during periods of international tension. Today, with Russian internet troll farms, clickbaiting Macedonian teenagers and the US extreme right pumping out propaganda on the internet, democracies need new versions of those jammers.
That will be a lot harder, but the should spark a technological fightback. It , working mainly through a St Petersburg company called the Internet Research Agency, ran bogus grassroots activist groups on Facebook 鈥� with hundreds of thousands of followers 鈥� and also ran social media ad campaigns. These sowed social division on many levels, strongly criticising Hillary Clinton, for instance, whilst also politicking against Muslims and encouraging minorities not to vote.
Unreal Americans
To run those social media accounts, and to buy ad slots, the Russians had to appear to be US citizens. So, alleges the FBI, they stole social security numbers and IDs so they could open PayPal accounts that seemed legitimate. Such activity should have been a red flag, detected by PayPal and the social media firms. Now, at least, it gives people fighting fake news operations an idea of what to look for and ways to counter it 鈥� improved user authentication, for example.
To maintain the charade of being Americans, the FBI says the Russians also 鈥減urchased space on computer servers located inside the US鈥�. That meant they could, via an encrypted network connection to Russia, work from a US IP address. Such connections could also raise a red flag. And if, as is likely, this server capability was inside a cloud data centre 鈥� where cybercriminals have been known to rent servers for a short time before moving to another 鈥� it gives tech firms another telltale sign to look for.
If 鈥渏amming鈥� of such fake news operations by detecting fake IDs, or the use of hired servers, is not enough, some are pinning their hopes on machine learning. Earlier this month, for instance, one of , Christopher 鈥淏iz鈥� Stone, was amongst investors pumping cash into Factmata, a aiming to highlight deceptive online content with the help of AI.
By training a machine-learning algorithm on large datasets of malicious, hyper-partisan or hate content, with learning supervised by humans, Factmata says it is able to spot key features that correlate with fake news 鈥� and give readers an idea of how credible a source鈥檚 content is. It focuses on a credibility score because software simply cannot decide if an article is categorically true or false unless the subject matter is very simple.
This seems preferable to the notion that governments should run fake news detection operations, as plans, for example. In an age of spin, political involvement in the process will leave many feeling uneasy, if not downright queasy. Let鈥檚 leave it to the tech sector.
