A design for decentralized organisations: part 1

The world of crypto is rich in attempts at decentralized project-governance — many of them functioning merely to signal that an ICO project has not thought through the design problems it faces. Yet there are also many serious-minded governance implementations: from the extreme legal positivism of the ill-fated DAO, to the direct democracy of masternode-approved superblock funding systems, the algorithmic meritocracy of Colony, Bitnation’s choose-your-own-nation holocracy, and whatever DAOstack might become, the design space is clearly wide open. However, I have yet to see an attempt to discuss decentralized governance from first principles. What follows is a first stab at grasping the fundamentals of decentralized organisational design.

Some paradigm-defining virtues

I will assume that an organisation is partially modellable as processes of moving from state to state in a planned manner, that is, a state machine. Deterministic state transitions are desirable if the organisation is to form and achieve goals (that is, to be purposefully coordinated). Performant processes of state transition are desirable if the organisation is to function efficiently. In a decentralized organisation, state transitions must occur at the edges of a network to avoid requiring central coordination. In such a scenario, when a state-change has determinately occurred (whether by consensus or by a datum’s inherent provability) the system has reached finality with regard to that state.

I’ll now informally integrate these virtues into the familiar “scalability trilemma” of distributed systems design, that is, the apparent intrinsic conflict between performance, security, and decentralization. More concretely, I will focus on the goals of minimising intercommunication load, reaching finality quickly, minimising the need to trust others to behave honestly (including the maximising of transparency in inherently trust-laden processes), and preserving the sovereignty of actors in the system.

The material in this post series is organised into four principles, which separate the series into its parts. At several points, the discussion supports “conclusions” which are concerned with the practical goal of implementing this design. The intention, provided this series is met with a favourable response, is to convert these conclusions into a concrete systems design for the Blocknet.

Principle 1: don’t design

In a world of equals, the Grand Architect is first to be ostracised. After all, who could purport to solve the world’s problems without some vested interest? As I understand it, the general statement of this problem is that, for any group whose members cannot trust one another unequivocally (that is, for pretty much every human group), anyone aiming to solve its problems should rightly be contended with unless his/her truth-claims are verifiable independently of the claimed neutrality or honesty of the messenger.

In other words, decentralized organisations should not merely be designed, but their correctness should be verifiable in some publicly demonstrable way so that designs can be adopted at the edges of the group. Either this, or people will do exactly whatever they want to do, and the only systems that emerge will be ad hoc and will not be able to reap the benefits of good design.

Now, organisational processes are orders of magnitude more vague and organic than cryptosystems. As such, they hardly admit of formal proofs of correctness, and so it is unsurprising that almost any social design is met with suspicion, and that rival factions are the norm in an organization. Given this, a designer would commit a fundamental error if (s)he were to present a design as if it were the right, good, and true solution for an organisation. No-one is in a position where their peers can unequivocally verify one’s qualification to make that claim.

Since this article explicitly aims to identify (some) first principles and then to construct an organisational design, it’s a good time for me to note that anything “designed” here is intended as a suggestion for further discussion. (Moreover, it is only a flatus vocis.) That said, I believe that this series unearths the fundamentals, or first principles, of decentralized organisational design in a way that can be followed by any interested reader, and its conclusions arrived at independently. The test of this belief, of course, will be in the ensuing discussion.

Thus ends the introduction to this series. In part 2, I will discuss scaling communication, which extends to decision-making at scale, and truth-finding at scale, and includes concrete means by which these can be achieved today. In part 3, I discuss transparency and its complex relationship with decentralization and scalability, recommending the use of consensus-free self sovereign action wherever possible. In part 4, now equipped with the materials from the prior three parts, I tackle consensus directly and derive the fundamentals of decentralized organisations. Since my primary goal is to allow the reader to arrive independently at my conclusions, it includes considerable discussion on the method of derivation, and why it is a rigorous one.

Continued in part 2.