In XKCD comic 1357, “Free Speech”, Randall Munroe offers a characteristically concise and snappy summary of one of the canonical arguments about free expression: “The right to free speech means the government can’t arrest you for what you say. It doesn’t mean anyone else has to listen to your bullshit, or host you while you share it…. If you’re yelled at… or get banned from an internet community your free speech rights aren’t being violated.”
I have a lot of sympathy for this argument! After all, what’s the alternative? That the government should order other people to host your objectionable speech? Surely there’s a free speech interest in not being compelled to publish things you disagree with. “Free expression” includes the right not to speak, and the right not to listen.
But Munroe’s account of “free expression” is nevertheless incomplete. It’s true that the US First Amendment shields us from government interference in our expressive activities, but just because the government isn’t silencing you, it doesn’t follow that the question of free expression has been settled. Imagine two different restaurants: one prohibits any discussion of any subject the management deems “political” and the other has no such restriction. It’s easy to see that we’d say that you have more right to freely express yourself in the Anything Goes Bistro than in the No Politics at the Table Diner across the street.
Now, the house rules at the No Politics at the Table Diner have implications for free speech, but these are softened by the fact that you can always eat at the Anything Goes Bistro, and, of course, you can always talk politics when you’re not at a restaurant at all: on the public sidewalk (where the First Amendment shields you from bans on political talk), in your own home, or even in the No Politics Diner, assuming you can text covertly under the tablecloth when the management isn’t looking.
Depending on your town and its dining trends, the house rules at The No Politics Diner might matter more or less. If No Politics has the best food in town and everywhere else has a C rating from the health department, then the No Politics Diner’s rules matter a lot more than if No Politics is a greasy spoon that no one eats in if they can get a table elsewhere.
What happens if some deep-pocketed private-equity types hit on a strategy to turn The No Politics Diner into a citywide phenomenon? They merge The No Politics Diner with all the other restaurants in town, spending like drunken sailors. Once that’s accomplished, the NPD cartel goes after the remaining competition: any holdouts, and anyone who tries to open a rival is given the chance to sell out cheap, or be driven out of business. NPD has lots of ways to do this: for example, they’ll open a rival on the same block and sell food below cost to drive the refuseniks out of business (they’re not above sending spies to steal their recipes, either!). Even though some people resent NPD and want to talk politics, there’s not enough people willing to pay a premium for their dinner to keep the Anything Goes Bistro in business.
Now, NPD moves to cement its dominance: after food-poisoning scandals involving its own kitchens, it successfully lobbies for food-safety rules that are so expensive that only the most profitable, best-capitalized businesses can afford to comply with them, strangling the dreams of would-be restaurateurs in the cradle. The local farmers and butchers, now unhealthily dependent on NPD, are squeezed into preferential pricing arrangements and “most favored nation” deals with the restaurant that mean that upstart restaurants will struggle just to source ingredients.
NPD is becoming synonymous with dining itself: they take over some school cafeterias, and even where they aren’t able to infiltrate the schools, they are the hangout of choice for schoolkids on lunch and after the final bell.
Soon, they’re offering barber shops, cab services, and a constellation of services and operations that comprise the majority of the places in which people would normally converse with one another. All of these businesses hew to the corporate motto, the one that’s right there in the name: “No Politics.” Anyone who tries to discuss politics gets shown the door.
Now, there is no government censorship involved, and there are still those sidewalks, and you can always converse in your own home, or in a private club, but all the spaces that are “public” (in the sense of being open to the public, though not owned by the public) where politics were once discussed are now no longer open to political debate.
Without any government action, we have arrived at a city in which the free speech rights of the residents are perfectly intact, but the ability to exercise those rights has been severely constrained.
We don’t live in a world of No Politics Diners, but we do live in a world in which most of our discourse has migrated online, and in which only a handful of services dominate: “I’m old enough to remember when the internet wasn’t a group of five websites, each consisting of screenshots of text from the other four.” (Tom Eastman/@ tveastman, Twitter).
These services – even the ones that started with a supposedly uncompromising commitment to free speech – have all had “house rules” about what could be said there since the very beginning, and, over time, these policies on acceptable speech have sprawled into vast libraries that set out a speech regime in which everything that is not permitted is forbidden.
The complexity of the speech rules (oh, for the admirable compactness of “No Politics!”) and the dominance of the platforms that created them are intimately related.
These companies, after all, did not grow by merely attracting more users than anyone else, nor is that how they’ve maintained their dominance. Rather, the Big Tech platforms have pursued growth through classic monopoly tactics: buying their nascent competitors, merging with their biggest rivals, cornering entire vertical markets.
To see how this plays out in the real world, consider Facebook: it’s a company that has hostages, not users. In 2018, Facebook was rocked by the largest exodus of US customers in the company’s history: 15 million 13-34 year olds left Facebook. See, the market works!
Not so fast.
The vast majority of those users who left Facebook ended up on Instagram, which has been a Facebook division since 2012, when Zuckerberg and co paid $1 billion to acquire the company, explicitly to capture younger users who were abandoning Facebook.
If you know even a little about US antitrust law, you may be scratching your head right now. Isn’t this kind of thing, you know, illegal? It certainly is! The kinds of tactics used by Big Tech to attain and maintain dominance are obvious violations of US antitrust law, the kind of thing that would have triggered charges, fines, breakups, and worse, within living memory.
Take Google, a company that’s only ever made one and a half successful products (a great search-engine and a pretty good Hotmail clone), whose growth beyond those products was entirely driven through mergers and acquisitions. How could Google get away with behavior of the sort that put Microsoft through nine years’ worth of antitrust hell, just a decade before Google was founded?
The answer is that since the Reagan years, big business and its advocates in the US government and judiciary have been steadily “reforming” antitrust law on lines proposed by the disgraced economist Robert Bork (the guy who was denied a Supreme Court seat for his complicity in Nixon’s crimes). Bork argued that the only time the US should wield its antitrust authority was when there was clear evidence of “consumer harms” in the form of higher prices in the immediate aftermath of an anti-competitive action.
Bork’s doctrine is ridiculous, and the only thing more ridiculous is that it’s become mainstream in US political and economic thinking. If you’re wondering how the No Politics Diner could be allowed to buy all its rivals and use its market power to crush small farmers and any holdouts who wouldn’t sell, thank Robert Bork. So long as NPD isn’t raising prices, Bork’s antitrust has nothing to say about it. And if NPD is using its investors’ cash to subsidize meals and sell them below cost to corner the market, so much the better! (If you’ve wondered how Uber and Lyft can get away with losing 41 cents on every dollar they charge you, now you know).
The thing is, turning the internet into “five websites, each consisting of screenshots of text from the other four” is a pretty ugly business. This concentration of market power means that just a few executives wield tremendous power over our political, romantic, educational, employment, civic, family, and social lives. When they get it wrong, it makes things terrible for millions of people.
Of course, nobody’s perfect, least of all tech execs, who are completely ordinary mediocrities who are no more suited to managing the online lives of billions of people than you or I are.
This means that Big Tech is routinely rocked by horrific, mass-scale scandals, and when that happens, lawmakers and policy makers begin to shake their fists at the tech companies. Sometimes, they even pass laws about them: in the past year, the EU passed the Copyright Directive and the Terror Regulation, both of which require Big Tech platforms to install hundreds of millions of dollars’ (or euros) worth of filtering technology that is supposedly going to catch and block all the terrorist and copyright violating material posted by internet users (spoiler alert: they won’t work).
Even when governments don’t enact laws limiting speech on the platforms – the US, with its First Amendment, struggles to pass laws of this sort – they can still gather in committee, call the CEOs of the companies before them, berate them for the cameras, and insist that Something Must Be Done. This can lead to boycotts, pressure on advertisers and other suppliers, and bad employee morale, with the risk of departures for another Big Tech company (good luck finding one that behaves any better) or a startup (whose only meaningful path to success is absorption into a Big Tech company).
So the Big Tech platforms play Old Woman Who Swallowed the Fly. They make a policy banning a certain kind of conduct, which captures another, benign conduct, and so they make a policy that’s an exception to the first policy (think of Facebook trying to reconcile its “Real Names” policy with trans people who took new names when they transitioned). But now you’ve got two problems, because there’s new conflicts brewing with both the old rule and the new refinement rule, and that begets two more exceptions, which beget a fresh litter of problems.
Complexity is the enemy of security, and when you’re trying to devise a policy whose mandate is “Permit all conduct that is socially acceptable among our 2.5 billion users” and also “Do not permit any conduct that is socially repugnant among our 2.5 billion users” and also “Do not permit any conduct that violates terrorism, harassment, hate speech, blasphemy, lesse majeste, libel or other speech regulations in any of the 150 countries where we operate,” you are going to produce an incoherent tangle of rules akin to what you’d get if you put every etiquette manual and every law book ever published into a blender and frapped the lot.
What’s more, the resulting slurry will be about as structurally sound when it comes to preventing bad people from doing bad things on your service.
That’s because a system with a lot of loopholes and exceptions will always yield up exploitable vulnerabilities to people who systematically probe it. The lines between good conduct and bad conduct are necessarily arbitrary: if you create a rule against “harassment” then you have to create a threshold beyond which conduct crosses over from merely “unpleasant” and into “harassing.” Given enough time, an adversary can compile a list of words, phrases and activities that are almost harassment, and deploy those.
The difference between “harassment” and “very nearly almost but not quite harassment” is so arbitrary that the subjects of this v-n-a-b-n-q harassment will probably not register that there’s any difference. Meanwhile, if you replace a system of rules with the judgment of moderators – “If you think it’s harassment, shut it down” – you will both overenforce the rules (for example, banning or silencing people who are responding to hateful, racist, homophobic, or misogynist goading), and underenforce the rules (ignoring harassment because it “just doesn’t seem like a big deal to me”).
At small scale, these contradictions aren’t a big deal – but once you’re dealing with thousands of moderators overseeing billions of users speaking dozens of languages all over the world, a bad call can result in de facto bans on entire fandoms, genres, political ideologies – or it can lead to literal genocide.
As bad as that seems, it gets worse. Not only is complexity a gift to abusers and anyone acting in bad faith – it’s also a curse for anyone trying to conduct legitimate discourse.
A professional troll has all the time in the world to become a Facebook Lawyer and learn the ins and outs of the the ever-shifting Facebook rulebook. But non-trolls – members of political oppositions, sexual or other minorities, people in stressful situations such as police shootings or the wrong end of terrorist attacks – can’t afford to go to Facebook Law School. They’re too busy living (or fighting for) their lives. That means that the rules intended to catch bad conduct will only stop amateur trolls – and innocent people who are the by-catch of speech policies, dolphins in the content-moderation tuna-net.
This is the free speech version of Gresham’s Law: “bad money drives out good” (people spend counterfeit money as quickly as they can so they can get rid of it, so that eventually all the money in circulation is counterfeit). The more complicated speech-policing rules we create to prevent bad speech, the more good speech we’ll catch by accident, and the more really determined bad speech we’ll let through.
The rise and rise of content moderation rulebooks has only fed the troll-industrial complex. Now, professional trolls (“reputation management” or “marketing” firms) know exactly how to skate right up to the cliff’s edge of Moderation Canyon, and they also know how to goad the people they’re paid to argue with into going right over the edge.
The US government has made very few laws regarding speech online, and yet we live in an environment that has been very toxic to speech itself. For years, trans activists, indigenous activists, anti-pipeline activists, #BlackLivesMatter activists, and others have discovered that once they’re banished from a handful of dominant services, they are effectively out of the conversation. They can set up a private message board or mailing list on their own server, but without access to the prisoners of Big Tech’s walled gardens, they are marginalized, left with dwindling numbers and few ways to get their messages out.
Lately, a very different kind of political actor has discovered that market concentration has grave consequences for speech: the political right, who, ironically, championed the dismantling of antitrust law and militated against any state-provided online spaces (which would be subject to First Amendment limitations on moderation of political speech), are now waking up to find that their heroes are being “deplatformed” – kicked off the handful of dominant services and thus disappearing from the discourse.
It’s tempting to simply laugh off these “free market” fetishists as they get their comeuppance when Alex Jones and the Daily Stormer get kicked off the internet, but that is to miss the wider point: we are now in a speech environment where power is so concentrated that the whims of a half-dozen tech execs determine – for all intents and purposes – who may speak and what they may say. If you think that power will only be wielded against Alex Jones, there’s a bunch of trans activists, indigenous activists, anti-pipeline activists, #BlackLivesMatter activists, and others who’d like to have a word with you.
What’s more, this situation is a form of government regulation of speech – even if it doesn’t violate the First Amendment. When the government declines to enforce antitrust laws so the market for speech forums is cornered by a handful of companies, when it creates compliance rules that only these companies can afford, when it fails to build publicly owned alternatives bound by the First Amendment, it is making speech policy. Failing to use your legal powers to prevent Big Tech from gaining a monopoly on speech is a form of action. It’s a policy. It’s a regulation of speech.
When the city council let the No Politics Diner take over your town, even though it’s in their power to stop the takeover, it has the same effect as making an ordinance that says, “No one is permitted to talk politics within a dining establishment” – but without opening the town up to a First Amendment lawsuit.
The world’s governments probably didn’t set out to create a shadow form of speech regulation. Instead, they swallowed a fly (“let’s stop enforcing antitrust law”) and then they swallowed a spider to catch it (“let’s shame and pressure the giant tech companies into policing their users’ speech”) and then a bird to catch the spider (“Automated filters will solve this!”) and now, in the EU, they’re considering swallowing a cat to catch the bird (“What about a tribunal to oversee the filters?”).
It’s time to stop trying to fix the platforms and time to start working on fixing the internet. Even if the “group of five websites, each consisting of screenshots of text from the other four” is made more civil and less toxic (it won’t!), we should still demand a better internet. Relying on the noblesse oblige of the perpetual Lords of the internet to manage the digital nervous system of the 21st century would be a bad bet even if these guys weren’t completely ordinary mediocrities who are no more suited to managing the online lives of billions of people than you or I are.
A restored internet is one that values pluralism (power diffused into many hands) and self-determination (you get choose which tech you use and how you use it). Achieving a pluralistic internet of technological self-determination will be a long process, but the tools to attain it are well understood:
* Competition law. Ban “mergers to monopoly,” prohibiting dominant firms from merging with their rivals. Ban acquisition of nascent competitors. Prohibit market-cornering (companies can be platforms or they can sell services on those platforms, but not both).
* Pro-competitive rules. Ban binding arbitration in terms of service, so that internet users don’t lose the rights Congress gave them just by visiting a website. Ban non-compete agreements so tech workers can leave their employers when those employers do objectionable things. Require transparency in algorithmic decision-making, including transparency about training data, model tuning, and any special cases the company has added. Reinstate Net Neutrality.
* Pro-user rules. Make it legal to jailbreak the locks on your own devices so you can choose who repairs it, how it’s configured, which apps it runs and which data it shares and with whom. Make it legal to violate terms of service so that you can extract your data, or use a tool that interacts with a service on your behalf (say, a bot that logs into Facebook and fetches your waiting messages and puts them into the inbox on a rival service, from which you can reply to them).
* Undo the damage. Break up the Big Tech companies (and other big companies). Close tax-laws that allow profit-shifting.
* Create a public sphere. Fund municipal fiber networks, nationwide broadband backbones, and public digital spaces that are owned by the public (while you’re at it, ban the state from conducting warrantless online surveillance, including on these services).
Randall Munroe wasn’t wrong when he wrote, “The right to free speech means the government can’t arrest you for what you say. It doesn’t mean anyone else has to listen to your bullshit, or host you while you share it…. If you’re yelled at… or get banned from an internet community your free speech rights aren’t being violated.”
But he was incomplete.
When the state allows the online world to become the near-exclusive domain of a small coterie of tech execs, with the power to decide on matters of speech – to say nothing of all the other ways in which our rights are impacted by the policies on their platforms, everything from employment to education to romance to (obviously) privacy – for all the rest of us, they are making policy.
Because inaction in the face of danger is a form of action.
Cory Doctorow is the author of Walkaway, Little Brother, and Information Doesn’t Want to Be Free (among many others); he is the co-owner of Boing Boing, a special consultant to the Electronic Frontier Foundation, a visiting professor of Computer Science at the Open University and an MIT Media Lab Research Affiliate.
This article and more like it in the January 2020 issue of Locus.
While you are here, please take a moment to support Locus with a one-time or recurring donation. We rely on reader donations to keep the magazine and site going, and would like to keep the site paywall free, but WE NEED YOUR FINANCIAL SUPPORT to continue quality coverage of the science fiction and fantasy field.