The Streisand effect and social media extremism

A few years ago, my old school magazine featured a pupil’s brief account of a geography field trip. Before the magazine was mailed out, someone had noticed a jokey reference to a minibus being driven erratically after the teacher had visited the local pub, and worried that this might be libelous. The school could have reprinted the magazine, or else thought “publish and be damned.” Alas they did neither: they stuck a white sticker over the offending paragraph in every copy. This led to the piece receiving perhaps 1,000 times more attention than it would otherwise have done, the attempt at censorship spotlighting a harmless joke that most people wouldn’t have noticed anyway.

There’s a name for this phenomenon, where attempts at suppressing information are counter-productive and make matters worse: the Streisand effect. It’s so named because in 2003, a project recording coastal erosion released aerial photographs of the Californian shoreline on the internet, including the stretch of seafront containing Barbra Streisand’s clifftop mansion. She (not unreasonably) instructed her lawyers to have her name removed from the photograph, citing security concerns. The lawyers unfortunately overplayed their hand, demanding the photograph itself be removed and seeking millions in compensation. Before the trial, the photograph had been viewed only six times. The case — which Streisand lost — then gained huge media attention. By the end, half a million people had downloaded the photograph.

The Streisand effect is a kind of obverse of Aesop’s “Boy Who Cried Wolf,” where repeated spurious attempts to attract attention will lead people to ignore you. What you have instead is “the boy who denies the existence of wolves, despite considerable evidence to the contrary.”

Once people believe information is being suppressed, all kinds of fantastical conspiracy theories may fill the vacuum. This effect, I think, partly lay behind the riots following the child murders in Southport. For the first day, the news media kept reiterating that the perpetrator had been born in Cardiff and would say nothing else about him, even though they clearly knew more. What may have been a well-intentioned attempt to convey the fact that Axel Rudakubana was not an illegal immigrant was so palpably an attempt at deflection that conspiratorially minded people thought they were being lied to, and confected bogus explanations of their own.

The problem with information is that we have not evolved to process it literally. What is unsaid may have a greater effect on us than what is said. And, having evolved to live in small groups, we cannot cope with the modern glut of opinion. Research at Stanford shows only a small minority of people hold deranged opinions; those who do, however, spend an insanely disproportionate amount of time consuming and propagating them.

In a group of pub friends, it is perfectly acceptable to be a Marxist or a train-spotter. What is not acceptable is to talk about Marxism or trains to the exclusion of everything else. You would have no friends. Online, this constraint does not apply. On X it is not uncommon to find people who have tweeted 100,000 times exclusively on the same topic. Hence 2 percent of nutters may generate 15 percent of content, leading everyone else to believe 15 percent of people are mad.

This means there may be an easy way to preserve free speech online while minimizing extremism. The Greeks not only promoted parrhesia — free speech — but also isēgoriā: “equal access to speech.” If you want to reduce online extremism, one solution is to let people say what they like but limit the frequency with which they can say it. A daily cap on everybody’s posts could work wonders.