Twitter needs more filters to protect free speech

Heiko Maas may be about to learn where the road being paved with his good intentions will lead. A Social Democrat and the current federal justice minister, he has announced ambitious plans to rid the internet of abusive and offensive language. His plans have incited concern in the German offices of Twitter and Facebook and may ensure that he goes into history books as the politician who brought the curtain down on free speech on social media in Germany.

Maas’s plans, which centre on legislation allowing legal actions against online insults, libel and sedition, take aim at several real problems, including a sharp increase in hate crimes against more than one million new migrants and refugees to Germany and the spread of “fake news” that, in his view, helped Donald Trump win the American election. And they come after his own failed attempt to get social media companies operating in Germany to agree to self-regulation. Despite promises by Facebook to crack down on harmful speech, Maas says Facebook still deletes only 39 per cent of punishable content, and Twitter only 1 per cent. After years of negotiations the minister, quite understandably, has run out of patience.

Recently he proposed a law obliging social media to erase “obviously illegal” content within 24 hours after a complaint. In less obvious cases, the deadline is one week. If the networks don’t comply, they face fines of up to €50 million, or $55 million. The law is expected to be ratified by Parliament before it goes on its summer break. But even as a proposal, its chilling effect on freedom of expression can already be felt; Twitter is now blocking accounts in Germany that have even the slightest whiff of hate speech.

One such account is called @einzel fallinfos — roughly, “individual case reports.” The account’s name mocks the mainstream narrative in Germany that crimes committed by refugees and migrants are “individual cases” — something the account’s operators clearly dispute. Instead, they see a “recurring pattern” of sexual assaults against women perpetrated by young men of mostly Arab origin. So they persistently post official police reports about, as they put it, “crimes committed by refugees, migrants, and presumed migrants.”

As of May 15, the @einzel fallinfos account is no longer accessible in Germany. When I asked Twitter why it was being blocked, and how many other accounts are being blocked in Germany, a press officer said, “We do not comment on individual accounts for privacy and security reasons.” What were these privacy and security reasons, I asked? “Longstanding company policy — nothing further to add.”

Of course, as a private company, Twitter isn’t obliged to give any reasons for blocking users from its platform. In theory, it could do so because it doesn’t like someone’s face. But for a business that thrives on giving a voice to as many people as possible, such arbitrariness and opacity may be self-defeating.

Though narrower than in the United States, freedom of speech in postwar Germany has been broadly interpreted by the courts. In 2009, the Constitutional Court, Germany’s highest legal body, ruled that disseminating radical-right-wing and Nazi views is not per se unconstitutional. On the contrary, the judges ruled that the Constitution “relies on the power of free confrontation as the most effective weapon” against “totalitarian and inhuman ideologies.”

Twitter is less sure. Faced with the threat of being held responsible for offensive and illegal content, the company instead relies on the power of the algorithm. In March, the service announced that it had updated its software to restrict accounts that engage in “abusive behaviour.” We don’t know how Twitter reaches these decisions, because it doesn’t disclose the process.

One possibility is that after an account attracts a certain number of complaints, algorithms intervene to silence it. That’s bad news for Twitter, as users are likely to flock to alternative platforms like gab. ai. And it’s bad for the rest of us, because it creates yet another bubble in our already filtered public discourse.

The truth about free speech in Germany is that its limits depend heavily on context, in ways that are far too complex for any algorithm to sort out, let alone to discern “obvious illegality.” An utterance that is punishable as an insult in a normal exchange can be perfectly legitimate if done in a satirical context.

So what is Twitter supposed to do when caught between its users’ interests in a broad debate and an ambitious leftist minister with ideological guidelines that, if in doubt, rule against free speech?

Rather than pull up the drawbridge, or fall back on algorithms, Twitter should hire a corps of well-trained personnel to deal with hundreds of thousands of contested cases. There will always be public courts to turn to as a last resort. But if Twitter wants to maintain its unique position as a leading marketplace for opinions and ideas, it needs to invest in the personnel to keep it there. Only then can it offer the most orderly and open debate possible.

So what is Twitter supposed to do when caught between its users’ interests in a broad debate and an ambitious leftist minister with ideological guidelines that, if in doubt, rule against free speech?