We take the Internet for granted, not realizing that such a global, decentralized system is a rare thing. Protocols, rightly, get credit, but they alone are insufficient. TCP/IP did not create the Internet. The Internet is not just a set of protocols, but rather a real thing. People and organizations created the Internet by hooking real hardware and communication lines together. To understand the importance of this, we need to understand what's necessary to create social systems like the Internet.

Social systems that are enduring, scalable, and generative require coherence among participants. Coherence allows us to manage complexity. Coherence is necessary for any group of people to cooperate. The coherence necessary to create the Internet came in part from standards, but more from the actions of people who created organizations, established those standards, ran services, and set up exchange points.

Coherence enables a group of people to operate with one mind about some set of ideas, processes, and outcomes. We only know of a few ways of creating coherence in social systems: tribes, institutions, markets, and networks¹. Startups, for example, work as tribes. When there's only a small set of people, a strong leader can clearly communicate ideas and put incentives—and disincentives—in place. Coherence is the result. As companies grow, they become institutions that rely on rules and bureaucracy to establish coherence. While a strong leader is important, institutions are more about the organization than the personalities involved. Tribes and institutions are centralized--someone or some organization is making it all happen. More to the point, institutions rely on hierarchy to achieve coherence.

Markets are decentralized—specifically they are heterarchical rather than hierarchical. A set of rules, perhaps evolved over time through countless interactions, govern interactions and market participants are incented by market forces driven by economic opportunity to abide by the rules. Competition among private interests (hopefully behaving fairly and freely) allows multiple players with their own agendas to process complex transactions around a diverse set of interests. Markets don't displace institutions completely but do help institutions process transactional exchanges.

Networks are also decentralized. The rules of interaction are set in protocol. But, as I said earlier, protocol alone is not enough. Defining a protocol doesn't string cable or set up routers. There's something more to it. One form of organization doesn't usually supplant the previous, but augments it. The Internet is the result of a mix of institutional, market-driven, and network-enabled forces. The Internet has endured and functions because these forces, whether by design or luck, are sufficient to create the coherence necessary to turn the idea of a global, public decentralized communications system into a real network that routes packets from place to place.

My interest in coherence is driven by efforts to create a global, public, decentralized identity system that we call Sovrin. Sovrin is an identity network that, like the Internet, achieves coherence through a combination of institutional, market-driven, and network-enabled forces. I think of Sovrin as the Internet for Identity. The Sovrin network is not owned by anyone, but, like the Internet, is created through the cooperation of many people and organizations. Sovrin is seeking to create a global, decentralized, identity network that is open to all. That's a little bit different thing that creating a standard or a protocol.

When I say Sovrin is "public" I mean that we intend for it to be a public good that anyone can use so long as they adhere to the protocols, just like the Internet. Public spaces require coherence. Coherence in Sovrin springs from the ledger, the protocols, the trust framework, standards, and market incentives. These gives Sovrin greater coherence than a protocol alone can achieve.

• The ledger creates a decentralized service where various participants can exchange important information without the need for a central authority. But like the Internet, the ledger requires multiple organizations to commit real resources. The Sovrin ledger is a significant piece of the Sovrin network. The ledger provides a global registry for identifiers. This has several significant advantages over DNS as a registry. DNS-based identifiers aren't permanently owned. This can present real security concerns. With a ledger, identifiers can be persistent and non-reusable. Second, DNS-based solutions require that every site that wants to run registry functions has to install and maintain software for that purpose. This incents centralization of the directories. With a ledger-based solution, a global network does this without special effort on the part of participants.
• The trust framework is an agreement that is administered by the Sovrin Foundation. The trust framework makes shared goals and principles explicit so that the various participants in the system can work together. The trust framework is the basis for shared action and processes.
• Standards like those being developed for Decentralized Identifier (DID) and the verifiable credentials are important for interoperability, but more so specifications and standards on how to communicate with the ledger and the peer-to-peer communication of agents are critical to portability and the rise of a functional ecosystem of software that isn't controlled by a single vendor.
• Market incentives have traditionally been difficult to incorporate into a networked ecosystem, but the rise of tokens is a recent development that promises to change this. Token-based ledgers come with incentives to participate and thus are self-sustaining in ways that DNS-based identifier registries can't be. This also alleviates power imbalances where large companies have an advantage relative to individuals. As Chris Dixon says "Token networks ...[align] network participants to work together toward a common goal—the growth of the network and the appreciation of the token." In short, tokens provide a means for establishing coherence.

Someone asked me recently how Sovrin differed from an identity protocol like OpenID Connect. The factors that give rise to coherence are the primary difference between a protocol like OpenID Connect and something like Sovrin. Simply put, Sovrin is a network and OpenID Connect is not. This is where I think we've largely gone wrong in building decentralized systems in the past. We've shied away from putting a stake in the ground and building a thing. Instead we've defined protocols and allowed the network effects to accrue to centralized companies who can establish the coherence necessary to create a real thing. My strong belief is that real networks are necessary for enduring decentralized systems.

Endnotes

1. To explore this more, see this John Robb commentary on David Ronfeldt's Rand Corporation paper "Tribes, Institutions, Markets, Networks" (PDF).