The Missing Middle


The promise of social technology is that we can discover more friends; specifically, people more similar to us than we previously had access to. The means by which we do so, the methods of collaboration, the quality of connected experiences; these factors matter, but they are marginal considerations compared to the basic offering of an expanded choice of connections.

The wilder-eyed, utopian end of early computing drew greatly from this promise; such unprecedented connection could bring about world peace, intellectual serendipity, higher consciousness. We didn’t get world peace, but we did get a genuine revolution in collaborative information-sharing and rapid communication. That ability to refactor our social preferences gave birth to an uncountable universe of online communities and subcultures that would be unimaginable to anyone who didn’t see it happen.

We didn’t get a million friends either. Dunbar’s number, which posits an upper bound of ~150 trusted relationships per person, put the lie to dreams of a ‘global community.’ Better social tools give us more choice in who to fill those slots with, but not the ability to expand the personal spheres where we actually spend most of our time and forge most of our familiarity and trust with others.

We know this, and the social networks know it. To combat a slump in user activity, Facebook jettisoned promises of boundless discovery and started promoting Groups. The success of Discord and Slack happened precisely because they promised tools for gated communities, not open commons.

One of those early promises came true in spades, though: while we still live in Dunbar communities, we’ve sped up the sorting process. We can find people like us at astonishing speed, and the diversity of options for what ‘people like us’ means has expanded exponentially. The ‘Big Sort’ of modern society is rapid, unprecedented, and utterly reliant on social technology. The problem is that for all of our freedom to find each other and make mutual agreements, the underlying technologies used for all of this sorting and trusting don’t actually reflect the diversity of expectations, norms, or approaches we might want to try. One major reason for this status quo is the underlying structure of the client-server Internet. Unlike in the physical world, where cities, religious groups, and professional associations both own real assets and collectively determine rules and norms, there is no median level of civic compromise on the Internet; just massive technical providers and platform companies, and dispossessed Dunbar communities and individuals unable to determine such rules and norms for themselves.

Both internal norms (‘how do we act’) and external perceptions (‘who are we?’) get yanked around by the dictates of platform owners, routing infrastructure providers, and the societal consensus that governs both, at levels far above our Dunbar phyles. Privacy is contradicted by the need to surveil internal activity in order to target ads. The current model of social connection technology sets the bar for self-determination so high as to prevent it from adapting to external circumstances. The technical realities of the situation are due to client-server technology and its institutional history. The social aspect is a holdover from the 20th-century media paradigm of mass consensus cultures.

The ‘long 20th century’ of mass institutions and broad consensus realities is over. Contrary to impressions, it was not the inevitable or final result of moral and civil progress. Rather, it arose from pragmatic external incentives of mass advertising and great-power ideological rivalries.

The assumption that we must inevitably play by the same rules of discourse, given the autonomy granted by wealth, technology, and knowledge, is rapidly degrading. This status quo is a leftover from a time when mass education prepared workers for mass jobs, where mass communication hammered the same messages over the factory loudspeaker and advertised the same small number of products over the few television channels available. Yet we still assume that the shape and function of the tools and spaces for dialogue we inherited from that era are immutable and inevitable; this outdated belief cuts us off from new methods of choice and collaboration.

Without the external political and economic pressures which created our mass culture, inclusion in or acceptance by it becomes just another sinecure for sale, the outcome of a zero-sum auction to benefit from and utilize for further gain, the power accumulated by the legacy institutions safeguarding it. The ‘objective’ model has become a straightjacket on human ingenuity and community self-determination. The fact that it is ‘supposed to exist’ means that it is the ultimate prize. Inclusion in the mainstream Discourse grants scalable exploitation of any idea and prestige to its proponents. But it is a booby prize. No serious and nuanced idea, no dedicated community can withstand its flattening effects or the ways that it commercializes and cargo-cults its subject into an artificial shape. The resulting conversations are lackluster in form as well as subject.

The stiffer the Discourse is; the less input it deems useful, the more fragile it becomes. The inside becomes a stultifying garden party of outdated, anodyne official narratives; the outside a stew of paranoid, enraged rump factions shut out of any path to respectability or good-faith engagement with their perspective.

The same dynamic wreaks havoc inside Dunbar groups as well; when there are fewer gradations of acceptable difference, internal schisms take on a stark tinge of zero-sum loyalty versus betrayal, rather than consensual drift. Savvy users, knowing the stakes, migrate quickly, leaving behind wastelands of the obstinate and clueless, who are in turn preyed upon by grifters and demagogues. Only when external events force a ‘tectonic shift’ in official understanding do these standards change. The problem is that the usual result of a ‘tectonic shift’ is a sudden, destructive earthquake. If networked culture is going to survive dramatic events out in the real world, we’re going to need a middle structural layer that allows norms and agreements to shift and negotiate based on observed reality and the wishes of ordinary participants.

This omission of a ‘middle layer’ of technical governance directly informs our vision for the Urbit network. In an Urbit-based world, we envision stars will be consensus hubs. There are 256 Urbit galaxies, each able to spawn 255 stars for a total of 65,280 potential stars. Each star, in turn, can spawn 65,535 planets—far above the Dunbar number of an intimate community, but also far below the ratios of authoritarian control under which current platform regimes operate. Stars and galaxies perform the essential routing and peer discovery functions of the Urbit network; planets represent distinct individual identities.

Although every star is spawned by a particular galaxy, it can move to any other galaxy which agrees to host it. Similarly, planets may migrate between stars. Many imagine the extreme ramifications of such a system, projecting a Darwinian competition of starkly diverse political models encoded in governance. When we think about the future of Urbit, though, we don’t see a chaotic universe of iconoclastic outposts. We don’t doubt there will be diversity among stars (driven by the wishes of their owners and resident planet users/customers), but the real value is in the antifragility allowed by far more subtle gradations of difference in rules and structures decided upon by stars. Each star will have much, but not all, in common with its immediate neighbors, across a constellation which contains a great overall diversity of norms. Changes in policy or alignment will resemble ripples across a pond more than a buildup of tectonic force against a brittle fault.

Stars are Schelling points for such gradations precisely because they are discrete cogs in the overall technical infrastructure of Urbit. Technical responsibility and any bounds of ownership will always mark such points. Their designation in the case of Urbit or any other system is necessarily somewhat arbitrary. Our vision is orders of magnitude less centralized than the status quo, but not devolved so far towards the individual that community formation demands massive amounts of individual technical knowledge and maintenance tasks. There will always be a need for routing infrastructure, ergo power (the power to set community norms and external relations) will always accrue to the points of control over that infrastructure.

Urbit simplifies much of this infrastructure but, just as importantly, it makes those technical locations, and the conditions of ownership and control, explicit. It allows for them to be owned and controlled by anyone who purchases them, rather than controlled by network-level overlords and unelected government regulators. Stars are somewhere between ISPs and consumer hosting platforms (like Geocities an eon ago, or Wordpress today); independent enough to set their own policies, but generally subject to the realities of running a profitable business. The decisions made by service providers inevitably have a great influence on the norms of any network, and Urbit is no different. The advantage of the Urbit model is that there are far more stars than ISPs, they are easier to run, they are fungible properties with clear ownership, and they are not burdened with technical and regulatory debt which ties them to the status quo. These qualities allow them to be purchased and maintained by a far wider range of people than might be able to control an ISP today; the massive decentralization of routing infrastructure also seems a likely antidote to the current state of regulatory overreach and centralized snooping which ISPs are subject to.

The Internet gave us an unprecedented flowering of diverse cultures; it gave communities the option to pursue a vast number of paths and goals. And we should have the ability, granted by wealth, free time, and communications technology, to arrange the rules of our collaboration as we wish. The structural design of the Urbit network is an effort to provide that power, encoded in a digital asset with private keys that can be held both by individuals and (via multisig arrangements) by groups. The lower price and complexity of stars compared to current ISPs also makes a lot more room for innovation in business models; star owners can make their own judgments about how to prioritize commercial revenue versus cooperative ownership, anonymity versus identity, short-term versus long-term contracts.

We see stars as an opportunity to incubate community hubs, to plant a flag in the wilderness, declare values and alignments, and see who shows up. This process can be varied; some stars might advertise widely, others might quietly and slowly invite other communities on board. That’s the point. There’s a nice juxtaposition between first mover-effect and overall availability here; early adopters will have an opportunity to shape discourse, but the supply is not so constrained that they’ll capture all of it. Early adopters will likely have an immense and permanent value of voice, infrastructural ownership, and ongoing revenue over their shorter-term speculative one.

The diversity and non-discoverability of Urbit are already showing themselves: we hear whisperings of insular communities that we not only don’t know about but by our express intentions can’t know about unless they make themselves known.

Regardless of whether we or anyone else know about them, star owners already possess legible chunks of the ‘missing middle’’—large enough to affect the map of the Urbit universe, small enough to allow a wide array of voices—in a way that the ownership models of the current Internet can’t come anywhere close to providing. Not every neighborhood poker club needs an Urbit star. But the institution-founders, tastemakers, and social connectors of today and tomorrow, those who want to build opera houses on the digital prairie, should consider the advantages of building on a star.