Distributed Network Organizations

I'm going to start here with an example.

The most widely known distributed network is the Internet. There are many who know far better than I the principles on which its protocol, TCP/IP, is based. Instead, I'd like to highlight some features of the Internet that I believe are relevant to any other organizational design effort:

• A basic agreement (protocol) under which every participant operates, and which is readily identifiable anytime two nodes are considering connecting.
• Terms under which new parts are recognized and added to the network are set globally, but are executed and managed locally.
• Potential for subnetworks to form, with non-trivial extensions of the basic agreement.
• Potential for fully private subnetworks.
• Public addressability and identifiability.

There are also some deficiencies of the Internet that I believe are also relevant:

• The human decision-making/governance of the defining agreements are not themselves structured as distributed networks. (They tend to be centralized structures.) They are slow, overly inclined toward uniformity, and subject to outside political pressures.
• There is no agreed-upon capitalization and operating expense structure that matches its distributed nature. (It has a decentralized structure.)
• The public addressing, identity and privacy structures are crude. For example, DNS is centralized, identities are relatively easy to spoof, and privacy is left to the good intentions of site owners, whose interests can be very different from those who visit the sites.

None of these issues are impossible to solve within or on top of TCP/IP. They just haven't been. Or perhaps most accurately, because of these deficiencies, solutions have been unnecessarily slow in evolving. Distributed systems tend good at evolving; centralized and decentralized systems tend to change either more slowly or more haphazardly.