(This is the final post in a five-part series on regulating online content moderation.)
In Part III, I showed how it is possible for private actors to remove an unpopular viewpoint from the internet by preventing websites that express that viewpoint from operating, a phenomenon I call "viewpoint foreclosure." In Part IV, I explained why every lawful website should have the right to exist—that is, to stay online. In Part II, I argued that regulating core intermediaries—the entities that administer the internet's core resources and, thus, the entities capable of effecting viewpoint foreclosure—would not run afoul of any First Amendment right to editorial discretion.
In this Part, I explain how the state can prevent viewpoint foreclosure by guaranteeing certain basic internet rights. To identify these rights—the bare minimum case for the state to intervene in private content moderation—we simply need to determine what a person requires of others to stand up a public website.
Connectivity. In the first place, for a person to operate a public website, she needs to connect to the internet. For residential subscribers—those who use their internet connection merely to consume websites and other applications—a standard internet access service will suffice. But for those who wish to host their own websites, a residential subscription will not do. They require commercial internet service with the requisite stability and bandwidth to make a self-hosted website available to the world. Although it might seem that net neutrality rules guarantee a right of connectivity, that is not so. The FCC's 2015 Open Internet Order (before it was repealed) pertained only to mass market broadband internet access, not to the kinds of commercial internet service public websites need to stay online. Moreover, the Open Internet Rules were concerned with website blocking when a subscriber is already able to enjoy internet access but his ISP prevents him from accessing certain websites or services that compete with the ISP. They did not concern themselves with ISPs who might refuse internet access altogether. To ensure enjoyment of that resource, we need a right of connectivity.
Addressability. Internet access is important, certainly, but unless you have use of a static IP address, your website will not be long for this world. IP addresses uniquely identify resources, including websites, on the public internet. Without a static IP address, any site reachable today may not be reachable tomorrow.
Just five regional internet registries (RIRs)—each a private organization—administer the IP address space, and each RIR can revoke any in-use block of addresses if the holder violates its terms. Thus far, RIRs have largely steered clear of the culture wars that rage within the application layer of the internet. But the decision by LACNIC, one of those RIRs, to revoke IP addresses that Parler relied on to escape deep deplatforming is concerning. LACNIC claimed to be enforcing a neutral (non-content-based) policy that Parler's new hosting provider had allegedly violated. But as I explain in my article, The Five Internet Rights, there are reasons to question whether LACNIC was enforcing its policies equally (or accurately). That fact, coupled with the broader ideological drive to deplatform Parler, raises the possibility that we may have seen the first instance of IP-based deplatforming.
Even if that suspicion is misplaced, it doesn't change the fact that five entities alone control the world's IP addresses, each of which is as free under current law to revoke IP addresses under its policies as Facebook is free to delete user posts that violate its standards. Given the centrality of IP addresses to online speech, a complete view of expressive access to the internet must include a right to maintain one or more static IP addresses for public websites. It must include a right of addressability.
Nameability. Nameability refers to the right of a website operator to maintain a domain name and, when users query that domain name, to have those queries answered (resolved) by returning the IP address at which the website is hosted.
As with the IP address system, the domain name system (DNS) has historically operated in a content-neutral manner. But that neutrality is waning. Domain registrars have increasingly waded into the content moderation game by inserting "morality clauses" into their registration agreements. For example, various registrars have prohibited registrants from associating domain names with websites that host "profane," "vulgar[]," "embarrass[ing], "derogatory, "racist, "homophobic," "blasphemous," or other "morally objectionable" content. Thus, dailystormer.com, gab.com, ar15.com, and other websites have seen their domain names suspended because their registrars disliked their viewpoints. Guaranteeing expressive access to the internet should therefore include a right of nameability that prevents DNS intermediaries from taking adverse action against domain names associated with lawful websites merely because they disagree with the viewpoints expressed on those websites.
Routability. Routability refers to the right of a website operator to have traffic to and from her website faithfully routed between intervening networks.
The term internet—short for "inter-network"—concisely captures the fact that the internet operates as a network of networks (in technical parlance, "autonomous systems"). Internet communication, therefore, is fundamentally a matter of "hopping" across networks, where each intervening network represents an additional hop between source and destination. Each network learns where to route internet traffic by receiving information from border routers in neighboring networks that "announce" which IP addresses they own and which other networks they can reach.
This ability to dynamically route traffic via different combinations of independently operated networks was one of the great innovations of the internet. But it also provides opportunities for viewpoint foreclosure. Network operators or others can attempt to take down unpopular websites by falsely announcing their IP addresses, a technique known as "BGP hijacking." Or, less dramatically, network operators could simply refuse to route traffic to or from an unpopular website by declining to announce the website's addresses or network number to neighboring networks. For example, after Cloud Innovation, an English colo provider, made itself a pariah in the African community by suing AfriNIC, certain African ISPs publicly discussed ceasing to route packets to IP addresses belonging to the company as a form of ideological retribution. Any regulatory scheme premised on preventing viewpoint foreclosure should therefore take account of a right to routability.
Accessibility. Finally, accessibility refers to the right of a website operator not to have users blocked from accessing her website.
Some ISPs already block access to illegal or infringing websites. Although the federal Stop Online Piracy Act (SOPA) and PROTECT IP Act (PIPA) bills, which would have required ISPs to block certain infringing sites, met an ignominious end after a well-publicized online revolt, copyright holders are nonetheless securing similar injunctions from courts.
But ISPs have just as much power to block lawful websites as they do unlawful websites. Such actions would no doubt violate net neutrality rules were they still in effect. Each of the FCC's three attempts at net neutrality included some form of a no-blocking rule that prevented ISPs from blocking subscribers' access to lawful applications and websites. And California's Internet Consumer Protection and Net Neutrality Act, which the state enacted in 2018 after the demise of federal net neutrality rules, currently prohibits such behavior. But net neutrality has historically targeted economic discrimination, not moral or ideological discrimination. It aimed to prevent ISPs from leveraging their power over their own networks to block subscribers from accessing websites either because the ISP offered a competing product or because the ISP wished to extract a toll from website operators to reach its subscribers.
Of course, taken literally, net neutrality rules would prevent ISPs from blocking any lawful website, regardless of the reason (economic or moral). But it remains to be seen whether net neutrality advocates, who tend to populate the political left, will continue to support such broad rules if those on the political right begin attempting to use them to protect far-right websites from left-led deplatforming campaigns.
[* * *]
In sum, if it is indeed possible to boot unpopular users, groups, or viewpoints from the internet altogether; if lawful (even offensive) websites should have a basic right to stay online; and if the First Amendment permits the state to intervene in attempts at "content moderation" by the entities that administer the internet's core resources—all of which I believe to be the case—then, fortunately, there is a workable solution. The architecture of the internet reveals five distinct choke points that could be used by private parties to excise disfavored speech from the internet. The state could protect lawful speech from that excision by enshrining five basic internet rights—the rights of connectivity, addressability, nameability, routability, and accessibility—a non-discrimination regime that would be clear and administrable. It could also be bipartisan, as it marries the economic discrimination concerns of the left that animate net neutrality with the ideological discrimination concerns of the right that animate social media non-discrimination laws like those in Texas and Florida. And because it focuses only on the core infrastructure of the internet—and the core issue of whether users should enjoy a basic right to speak on their own websites—its modesty is perhaps its most attractive feature.
Does that mean the state should never intervene in content moderation in higher layers of the internet stack, such as cloud computing or social media? I don't know. I remain unconvinced that either side of that debate has a slam dunk take on the issue. But if we start with the simple premise that all users should be able to speak not on individual websites but on the internet, then I think that that premise of basic viewpoint access could provide a platform from which to tackle those and many other thorny questions of content moderation.
