At the city water treatment plant, I learned that someone comes in each and every day of the year to take bacterial plates out of an incubator and verify that there are no pathogens in the drinking water. Then they prepare another set to be read the next day. And the next, and the next. Water treatment plants are a physical instantiation of the idea that politics are the structures we create when we are in a sustained relationship with other people. They’re more than just the technological systems.¹

And it's not just water treatment that requires such constant care. It's typical for infrastructure to require not just regulation from a distance but active, hands-on governance. There are good reasons for that.

First, infrastructure is always shared means, which means that it requires coordination, and second those means are shared to many ends — ideally very open-ended, unpredictable ends so that society may express its greatest creativity — which means that you cannot predict how it will get used and not all uses will be good.² Because of this, you can't just regulate and forget: you need an institution to remain involved in the continued operation of the infrastructure system.

Most major, typical infrastructure systems emerged gradually over time. And since becoming widespread, they have evolved and innovated, but usually within a relatively similar set of means and ends, at least in terms of what they enable directly. Thanks to that gradual emergence and relative stability over time, the governance of those systems has had the occasion to become reasonably established and reliable.

But the digital world has been different. In fact, this may be the biggest but least recognised shock delivered by the internet. The flexibility of computers, the connectivity of networks, and the end-to-end principle of internet architecture that means new services can be invented by users rather than by infrastructure operators have together made it incredibly easy to create novel types of services that precisely offer "shared means to many ends." And many of these services have become critical infrastructure over a highly compressed timeline.

When I started almost 30 years ago, you were a "webmaster" or a "sysadmin" or both, and while you had to rely on an ISP, on DNS, and on some internet functions, it wouldn't be rare to run your own everything — email, web server, DNS resolver — and to build any web site entirely from scratch, relying on nothing other than simple hosting software and basic browsers. Today, the list of infrastructural services you need to rely on to build a successful, professional, business on the web is longer than most people, even professionals, realise. As a user, the story is just as stunning: you used to do everything with a desktop computer, an OS, an internet connection, and some relatively basic software like a simple browser. In contrast, daily life now relies on a plethora of services, many of them evidently infrastructural. Digital infrastructure, the full collection of shared means to many ends in the digital sphere, is incredibly vast.

And that's the shock. This massive and sudden influx of infrastructure has overwhelmed our governance capacities. We simply do not have the institutional capacity to cope with that much more infrastructure coming into existence all at once. We have not been able, collectively, to keep up and to stand up for social, commerce, browsers, search, advertising, chat, operating systems, app stores, mapping, and more the kind of institutional bodies that we rely on so strongly that we hardly ever think of them but that work day in, day out to govern water, transportation, energy, sewage, libraries, the legal system, and everything else that keeps civilisation ticking. Ungoverned infrastructure will almost invariably collapse from overuse or destructive use, so‑called internet governance bodies have restricted themselves to governing but a tiny fraction of the digital sphere, and most states have at best attempted regulation from a distance — which given the kind of issue at hand has often proven predictably ineffectual.

This has created a power vacuum and power abhors a vacuum. Tech companies have rushed into that vacuum, and the sheer volume of infrastructure they now control explains why Big Tech companies are now, for all intents and purposes, supranational governance bodies that happen to also own some computers.³

Which leads us to a key question: if they provide governance, why would we want to take it over? For several good reasons.

Because shared means are public means (even if privately operated). There is a key difference between a system that many people happen to use — say, we all have a microwave over — and a system that we use in a shared manner. Shared means are already collective means, and are thus public in nature simply by the way in which they are deployed. The use of public power for private ends is elite capture: it is detrimental to most and we should correct it when it arises.⁴

Because democracy produces superior outcomes to autocracy. Infrastructure is key to complex society. It's how we shine, it's how we sustain our own complexity, it's how we solve ever more challenging problems. There is plenty of evidence to support the idea that this works better in more egalitarian systems (as covered in greater detail in Stewardship of Ourselves). When we keep infrastructure under captured, autocratic governance, we are limiting what we can do as a society, we are foreclosing our future. We cannot afford to do that now.

Because we need infrastructure neutrality and infrastructure must not be optimised for a specific usage. As I said at the beginning, the "many ends" of infrastructure ought to be unpredictable if we want to make happy accidents possible⁵, if we want to make it possible to invent open-ended futures. This breeds uncertainty, but "[u]ncertainty can be understood from two angles. It can be seen as risk to be eliminated, ignorance to be optimised away, information costs to be reduced — but it can also be understood as flexibility, freedom, opportunities, options, path independency."² When a single entity can govern infrastructure autocratically, there are many temptations. To extract rent and foreclose competition of course, but also to optimise and normalise infrastructure for a specific profit structure. This destroys the value of infrastructure for everyone else.

Because all peoples have a right of self-determination. That's how Article 1 of the UN's International Covenant on Civil and Political Rights and is also covered in Article 1 of the Charter of the United Nations. Digital infrastructure is in scope here. "While self-determination is usually discussed in its external dimension, i.e. territorial and political independence from external actors, it is essential to stress (…) the internal dimension of self-determination, i.e. the right to freely determine and pursue one’s economic, social and cultural development, including independently choosing, developing and adopting digital technologies."⁶ Having a small number of powerful corporations impose their will on free peoples isn't new — we've certainly seen our share of it in agriculture, oil, or minerals extraction — and tech corporations are just the latest generation of the same, but they call for a new playbook.

And, frankly, because it's hard to imagine someone doing a worse job. It's simply good governance that leaders who do bad work don't keep their job, instead of failing upwards as we've allowed tech companies to do.

Owning governance gives them the ability to command higher profits and to avoid meaningful public accountability — it's predictable that they will resist the transfer. I'm seeing this in W3C where an entrenched clique is resisting so much as establishing a public venue to discuss better governance or allowing the wider community a say in what the Consortium does. They will speak highly of the W3C's public interest mission — but only so long as they can control how that mission is deployed, who gets to be the subject of governance, and can keep the plebs out where they can't so readily point out the hypocrisy. We shouldn't expect change to come unforced.

This is where states have a role to play. We need to create an institution factory that can establish governance for all significant components of transnational digital infrastructure — with a properly digital focus, not based in communications. Digital sovereignty is typically understood as control over digital infrastructure, but due to the transnational nature of the digital sphere, we have to envision it as a form of cooperative sovereignty⁷ in which not all actors are states but rather form a public/open partnership both with one another and with a multiplicity of digital governance bodies.

In infrastructure, no dictatorship is benevolent. It's time to get over the shock and make it work for us, democratically.