Online propaganda: how safe were the European parliament elections?
The FT's Martin Sandbu looks at the threat of online disinformation around the European Parliament elections and asks what can be done to increase transparency and protect the democratic processes in the future.
Produced by Margot Tiounine; filmed by Petros Gioumpasis; edited by Richard Topping
You can enable subtitles (captions) in the video player
It's the new threat to national integrity. Misinformation spread through social media. And it's disrupted elections around the world.
Recent European Parliament polls were billed as the most hackable ever. But were they? And is enough being done to fight back?
To be able to detect patterns of malicious disinformation, journalists and analysts work with data sets made available to them by social media companies.
So when we talk about accessing data programmatically, what I'm referring to is writing a small piece of software called a script.
The aim is to detect abnormal activity, such as social media bots spreading fake news or computational techniques like political advertising using microtargeting.
We're interested in who is advertising most, who is spending the most on advertising.
For the first time, Facebook has published a database of political and issue ads that could help identify the top spenders in 30 European countries.
I would start by accessing data from the API. API stands for Application Programming interface.
So you extract this data through the API and then what?
And then you have, if you've written a certain amount of data to your computer, you can then analyse it as you see fit depending on what you're looking for.
But that data you have in here now, do you?
Yes, actually. That's what we're looking at here. So here you can see on the left-hand side the lighter blue dots represent lower bounds, the darker blue dots the upper bounds. Lots of the ads are repeated many, many times.
My guess would be they run a campaign with the ads, they look at the responses, they perhaps adjust the targeting.
So they optimise...
As an advertiser, it would make sense to do that.
Who is the top spender here?
Well currently, we see that it appears to be the People's Vote Campaign.
The level of concern that preceded the EU parliamentary election was absolutely warranted.
Social media companies like Facebook are under pressure to increase this kind of transparency. Fears of a foreign interference were fueled by accusations of Russian meddling in the 2016 US presidential campaign. But the European elections were uniquely vulnerable. Low turnout and protest voting means propaganda campaigns can reap high rewards. Nahema Marchal investigated social media misinformation in the run-up to the European election.
What we did is that we looked at the most important sources of junk news and professional news that were being shared during that one-month period before the election. And we looked at what the public engagement with these sources was on Facebook. So, how many shares, and likes, and comments, and eyeballs sort of ended up on these stories.
And by comparing the two, one of the things that was quite striking is that junk news stories, any given individual story by a junk news outlet, tends to be a lot more engaging. Junk news is emotive. It tends to rely on click bait, often taps into base emotions, like fear or outrage. And that's something that creates engagement on social media.
You looked at half a million tweets in seven different languages. What did you find?
Despite all of the concerns that were raised before the election, we find very, very low levels of junk news overall. And, even more strikingly, very few of these links actually redirected to known Russian sources of misinformation. While the attention, especially in the context of this election, was on foreign interference and potentially on Russian interference, that was not reflected in our data.
The findings and the data...
Can the social media companies themselves be trusted to take this seriously enough?
We're not yet at the stage, essentially, where any researcher can go and have comprehensive or meaningful access to social media data. Researchers still have to be extremely creative about the ways in which they analyse and examine the conversations that are happening on social media.
Public policymakers are also pushing for more transparency.
We haven't yet seen enough, in my view, engagement from the platforms. And we're going to keep pressing on these issues because, after all, the European parliamentary elections were very important, but elections don't stop. There's an election somewhere in Europe every week.
We didn't have..
Who is behind online propaganda? Who pays for it? Transparency on these questions is only a first step in the fight against fake news.