(Loic Venance/AFP/Getty Images)

In June 2018, the newly elected Italian government refused to allow the Aquarius, a ship carrying 600 refugees, to dock in an Italian port. This story was news, and Corriere della Sera, Italy’s paper of record, covered it accordingly, with articles written from different angles and with different headlines. But the article that really affected readers on Corriere’s Facebook page had a headline featuring the name of Italy’s then-new interior minister, Matteo Salvini, as well as his declaration: “Our ports are closed to Aquarius!” A new study — carried out by my London School of Economics Arena disinformation project , the University of Venice and Corriere della Sera itself — found that, in 2018, this was the second-most “engaged with” post on the topic of migration on Corriere’s Facebook page, the one that attracted the most clicks, likes, dislikes, comments and shares.

Pleased with the reaction, Salvini repeated the trick a few weeks later, banning another boat from landing on Italian soil. When challenged on this decision by a prosecutor from Catania, Salvini posted a 20-minute rant on Facebook. When Corriere linked to this post on its own Facebook page, it was so heavily debated that we found it became the most “engaged with” post on the topic of migration for the same one-year period.

These two events were not, of course, unique. Since his election in 2018 as the head of the Northern League — now known just as the League, or Lega — Salvini has used deliberately controversial language as well as deliberately staged “happenings” to force his way into the headlines. Recently, he did it again, not only by banning a boat — this one a rescue ship that had picked up people who would have otherwise drowned — but calling for the arrest of the 31-year-old German captain, Carola Rackete.

Quite a lot has been written, including by me, about the effect of social media on politics, and in particular the way in which the algorithms built into Facebook and YouTube are more likely to spread angry, extremist and deliberately provocative political language. Less is known about the way mainstream media copes with the same kind of language. At the moment, there is a kind of vicious spiral: Many newspapers, especially those without large subscription bases, are dependent on an ad-tech market that delivers more revenue to pages or posts that generate more clicks — and more controversy — regardless of their quality. Politicians such as Salvini, sometimes backed by cheaply made, partisan websites — also funded by advertising — as well as armies of real and fake followers, have learned to create this controversy. “Mainstream” papers feel they have to cover what they say, regardless of whether it is important or true, both because they need the ad revenue and because they will seem irrelevant if they don’t. This creates more incentive for more outrageous statements and staged events, and so on.

We found that the effects on the public of this vicious spiral are stark. In 2018, the immigration crisis actually ended in Italy. In real life, the numbers of refugees coming to the country plunged by about 87 percent in 2018 compared with 2015 thanks, mostly, to the efforts of the previous Italian government as well as the European Union, an institution Salvini often attacks. But even as migration fell, the number of stories written about migration rose dramatically. The “Salvini effect” actually distorted reality, creating new levels of controversy and anger around an issue that was actually disappearing.

Is there a solution to this? Our project tried to find one, partly by seeking to measure not just “clicks” on a particular article but also the quality of the conversation it creates and the trust it engenders. This was an experiment, and our answers might not be conclusive. But we did find, for example, that narrative features intended to humanize the migrant story did not, as might have been expected, create more positive feedback, especially when they featured a group of people. Instead, they inspired high levels of attacks on the reliability of the newspaper, as well as anger about migration. Old-fashioned reporters will be delighted to hear that straight, plain vanilla news reporting created the fewest attacks on the newspaper itself, suggesting that impartial reporting is best at creating trust.

There are others seeking to create new kinds of metrics, with the ambition, eventually, of getting advertisers to go along with them. The Global Disinformation Index, a group created with this goal in mind, is planning to use some automated techniques to measure the reliability of online domains, and will alert advertisers who want to bypass the ones that are most likely to produce manipulated information. People selling soap or shampoo may not want their ads appearing next to false or sensationalized articles whose readers are not likely to be drawn to them.

But advertisers alone are not responsible for creating a more civilized, less polarized discourse, and it’s not clear who else is. Journalists can rightly argue that their job is to report what they see; publishers are responsible for making money. Maybe there’s a role for a different kind of institution, a “public service-spirited” social media whose algorithms favor material that creates useful conversations and trust, not discord. We aren’t anywhere near that yet, but we can imagine what it might resemble, and how revolutionary it could be.