Swedes eat 'junk news' diet ahead of vote

The ratio of professional news to junk news shared on Swedish social media was 2:1 in a sample period of 8 August to 17 August, the report said.

This was the same as in the US election in 2016, when the populist Donald Trump won power, but the junk news ratios in recent UK (4:1), French (5:1), and German elections (7:1), were much lower.

"For every two links of professional news content shared, Swedish users shared one junk news story ... this was the largest proportion of junk news across all the European elections we have studied," the Oxford Internet Institute said.

It classified "junk news" as content that lacked professionalism and credibility, used emotionally-driven language, showed political bias, or mimicked normal news outlets via counterfeit methods.

"These sources deliberately publish misleading, deceptive or incorrect information purporting to be real news about politics, economics or culture. This content includes various forms of propaganda and ideologically extreme, hyper-partisan or conspiratorial news and information," it said.

Meanwhile, eight out of the top-10 most shared junk sites were native Swedish ones.

Three of these - Samhallsnytt, Nyheteridag, and Fria Tider - accounted for 86 percent of junk content shared, while Russian sites accounted for just 0.2 percent.

The findings come despite warnings by Swedish politicians and security services that Moscow had planned to interfere in the vote, as it did in the US, France, and Germany, to help divisive parties, like the far-right Sweden Democrats, do well in order to weaken Europe.

"We see that Russia has an intention to influence individual issues that are of strategic importance ... we can expect attempts at Russian influence," Daniel Stenling, a Swedish counter-intelligence officer said in February, for instance.

Native problem

The native junk news, which focused on "polarising" content, might still help the Sweden Democrats, which, in any case, produced more actively shared content on social media than any other Swedish party ahead of Sunday's vote.

The party's own material accounted for 14 percent of "high frequency tweets" in the sample period, compared to 11 percent by the centre-left Social Democrats and just 5 percent by the centre-right Moderate party.

One reason for the popularity of extremist content, the Oxford institute said, was the nature of the artificial intelligence or software used by Twitter.

"During times of heightened public interest, social media algorithms repeatedly promote conspiracy content over accurate information," they said.

In what amounted to better news for Swedish democracy, the study also found that 47 percent of the material shared "was general content, as opposed to party specific content".

Real debate

This meant that social media debate in Sweden was more about issues than about political blocs talking internally to reinforce their own views.

The proportion of general content shared in the recent French elections was 26 percent and in the German ones 29 percent.

At the same time, if the Twitter conversation was an index of the likely outcome of the vote, then a centre-left coalition is more likely to emerge than a centre-right one.

The left coalition (the Left Party, the Green Party, and the Social Democratic Party) dominated the sample Twitter conversation with 67,907 "relevant mentions and hashtags" out of the 274,953 tweets in the data set.

The right coalition (the Centre Party, the Moderate Party, the Liberal Party, and the Christian Democratic Party) had just 40,871 tweets, however.