Read in Catalan

Russia's disinformation machinery, which includes Julian Assange and Wikileaks and Russia Today, has got involved in the campaign in favour of the 1st October Catalan referendum, filling social media with fake messages aimed at influencing the vote, as detailed by Spanish newspapers El País and La Vanguardia (both in Spanish) at the same time this Saturday. The reality, however, is much more sophisticated that the amateur analysis of the two newspapers, especially El País', signed by its deputy editor.

The online independence debate is a real, active, but also strange conversation, but not in the way El País would like it to be. This Saturday they opened their front page with the topic and, because why not, mixed it with the social media manipulation during the Brexit campaign, Marine Le Pen's campaign, Donald Trump's campaign... carried out by quasi-official Russian organisations. With enough nothing you get something.

The "evidence" offered by El País is the percentage of fake followers of Julian Assange: 59% according to TwitterAudit, a simple tool that measures this and other variables. The report's author deduce that Assange has an army of "bots" behind him, fake profiles programmed to automatically retweet what he says. Fine. El País, according to the same tool, has a following that is 44% fake. By the same logic, it's worth wondering if they rent networks of bots to spread their content on Twitter on a large scale.

(For comparison, El Nacional has only 10% fake followers).

The second bluff

The other "piece of evidence" is the speed with which a tweet in English by Julian Assange sent at 18:46 on 15th September (below) earned more than 12,000 retweets and 16,000 favourites (24 hours). The newspaper deduces that this is only possible if there was an extensive networks of bots behind it.

That might be true, but they don't offer any data to confirm it. El País gets surprised by whatever is in their interest. More retweeted than Assange's pro-independence tweet was the below pro-union one with a link to the Spanish newspaper ABC:

This tweet has almost 16,000 retweets and 76 favourites. Almost 4,000 retweets more than Assange's.

Why does this tweet not call the attention of El País (nor of La Vanguardia)? At least it has two characteristics that jump out at a user with the least experience. One, the profile has no biography. Two, the enormous difference between the number of retweets and number of favourites, a typical indicator that it has gone viral thanks to a network of bots. The difference between the two signals a certain automation of the spread, as does the speed at which it happens.

Unionist bots

The internet analytics companies Websays and Sibilare have investigated the most powerful users in the spread of this tweet. The study identifies a network of bots spread around different countries, from Brazil to Indonesia, which show clearly automated and coordinated activity. As well as this article from ABC, they also shared a greeting in Arabic with the Saudi Arabian flag on the day of their national festival.

These profiles are very varied. There's even a porn one. For anyone curious, these are the profiles: @WhatSincero, @HBK_DB, @gaypotters, @homehoneyhoney, @gabzxaviorz, @Meki_Susan, @kennedyso.

In other words, currently, what we can confirm are some anomalous moves in the spread of information against the referendum. Maybe the better question would be who is behind @marilena_madrid and who hired their services. But this fits badly with the El País' story of Russian spies.

A converse scene|trouble

More clues that the story is doubtful. Websays and Sibilare have analysed a sample of the conversation on Twitter about the independence of Catalonia mentioned more than two million times. They deduce, as both newspapers do, that "there are suspicious moves aimed at altering public perception of the conversation through coordinated actions". At the same time, however, "there's not sufficient evidence that these actions can be attributed exclusively to one of the parties".

They do find "conventional activism on both sides, by real people (under their real names or pseudonyms) using the networks as they used to use traditional communication and debate channels". In other words, the conversation about the referendum on Twitter have many more participants, but the majority are flesh and blood, not bots.

In the two last weeks, the analysis adds, users with fewer than 25 followers have increased their presence in the conversation by more than 140%. In the next band, between 25 and 100 followers, the increase is 50%.

Taking into account that, globally, participants in the online conversation about independence have increased some 25% since the start of September, these anomalous increases, the analysis continues, can be attributed to three types of user: recent arrivals; users who simply work on coordinated rebroadcasting (a form of activism seen in previous election campaigns, for example the Popular Party's campaign in the last general election); abundant trolling.

What cannot be concluded, says the analysis, "is that these increases come from one of the two sides exclusively", as El País and La Vanguardia do.

Bad Assange

Another piece of confirmation of the analysis is that the participation of Julian Assange “has influence primarily among a pro-independence audience, much more than it does outside of Catalonia, or in other idiomatic environments”.

On the other hand, the Russian campaign manipulation apparatus works via advertising on Facebook. This has been known for months, but it has been headline news this week in the United States as Mark Zuckerberg himself recognised that they had been misled by Russian organisations. Twitter, for now, has remained above suspicion in terms of propaganda operations on such scales. But El País only talks about Twitter.

How has it become a headline that it was a "penetration" by the "Russian interference machinery"? Well, by applying a technique recently subscribed to to discuss the Catalonia-Spain conflict: reports that hardly inform of anything but which make you think even worse, like those published (in Catalan) on the recent Law of Transitional Jurisprudence, among others (in Spanish).

It's delicate cooking: it requires a skillful composition of insinuations, double meanings, half-truths, anonymous sources, etc. A rudimentary way of doing so was what they used this Saturday. It consists of associating the referendum with entities, countries or people who have been stigmatised by social consensus or by prejudice as "bad". Say, Assange (who used to be so good! - in Spanish), Russia Today, Trump, Le Pen, Disobedient Media and the whole catalogue of bogeymen. Automatically, the referendum is contaminated by their bad reputations.

It's sowing suspicions without evidence, with low-quality or distorted sources and few or no details. If they would like to stain the referendum by associating it with Russian manipulation, they haven't proved it. Anyway. We all have bad days.