In early August, Google engineer James Damore made headlines when his skeptical critique of Google’s diversity agenda was leaked to the public. While most of the tech media had a collective meltdown, Quillette published an intelligent response by four respected academics, defending some of the science underpinning Damore’s arguments. Unfortunately, it didn’t stay online for very long. The site was promptly hit by a denial of service attack which rendered it inaccessible for about a day. Whether the attack was the work of some left-leaning script kiddie offended by the support for Damore, or a false-flag operation designed to make us assume as much, we will likely never know. Whatever the motivation, the fact remains that person or persons unknown were able to unilaterally, albeit only temporarily, decide what legally-published content was allowed to appear online.

The normal solution to attacks such as this is to enlist the services of a company like CloudFlare, who can defend a site by concealing its true location within a much larger network. This was presumably what the administrators of the Daily Stormer, a notorious neo-Nazi community, were expecting when they signed up for CloudFlare’s protection. But after the Stormer published offensive remarks about a protester murdered at a rally in Charlottesville, the CEO of Cloudflare decided he no longer wanted their business, as did the domain registrar GoDaddy and several other hosting providers. The site was forced to move to a so-called “hidden service” within the anonymizing Tor network, at which point the developers of Tor jumped on the bandwagon with their own public denunciation of far-right politics. Amidst all the media cheering, only the Electronic Frontier Foundation thought to question the wisdom of asking the providers of internet infrastructure, who are supposed to act as content-neutral conduits, to take on the role of extrajudicial online thought police.

By all accounts, the Daily Stormer publishes repulsive content, including actual calls for violence against Jews. The internet is a better place without it. However, extremists are not the only ones being targeted in this way. Shortly after they fired James Damore for his wrongthink, Google also purged Gab—the anti-censorship alternative to Twitter which I have previously written about on Quillette—from the Android app store. The excuse given was that the app contained (you guessed it) “hate speech.” Although undoubtedly popular with Trump supporters, some of whom might cross the line into overt racism, Gab itself is just a platform open to anyone who wishes to exercise their First Amendment rights. It does not advocate any particular ideology, it does not endorse the opinions of its users, and its rules specifically forbid the promotion of violence or terrorism. The thing about witch hunts, though, is that it’s never enough to burn the witch: you also have to burn anyone suspected of enabling, sympathising with or defending witches. As far as the no-platforming paternalists of Mountain View are concerned, allowing someone the freedom to speak means that you agree with everything they say.

What is remarkable about this new Nazi hysteria is how selective it is. A few days after Charlottesville there were two Islamic terror attacks in Catalonia that killed 16 people. How many CloudFlare customers were booted off the platform as a result? How many Islamic State websites did GoDaddy shut down? How many blog posts did the Tor project write about the violence and hatred of radical Islamists? None, as far as we can tell. Either these companies believe that a tiny fringe group of deeply unpopular white supremacists is more of a threat than a global terror network that has killed thousands of people, or their actions are just publicity stunts designed to appease the latest Twittermob.

Censorship on mainstream social media is old news, but we are now seeing a worrying escalation of the phenomenon, as the providers of core infrastructure—web hosting, domain registrations, DDoS protection—are getting in on the act. Previously, if you didn’t like the way Twitter and Facebook run their platforms, the solution was simple: make your own site, where the rules are more relaxed and legally protected speech is not removed for political reasons. Now, even this option is being taken off the table. Your alternative site still depends on commercial organisations to host it, route traffic to it, list it in search results, and protect it from attacks. Once these companies abandon their role as content-neutral network operators and fall down the rabbit-hole of identity politics and virtue-signalling, the internet as we know it can no longer function. It ceases to be a free market where services compete on their own merits, and becomes just another walled garden: patrolled by a conformist monoculture of unaccountable Silicon Valley elites, who tolerate no skepticism or dissent, and from whom there is no escape.

In the early days of the Internet, it was hoped that its distributed nature would offer resilience against restrictions imposed by powerful entities such as governments. This robustness does still exist, but only at a fairly low level, where raw packets of data are routed between numerical addresses. The multitude of services that make the internet usable for ordinary people are very much centralised, and with the rise of the smartphone duopoly and their curated app stores, this centralisation has got a whole lot worse. If we can’t trust infrastructure providers to uphold the essential value of free speech upon which the internet—and arguably, Western civilisation itself—was founded, then we either need to replace them with somebody else who will, or find a technical solution that removes them from the equation altogether.