On the meaning of decentralisation.

It was a warm day in London when my friend and I were discussing novel protocol architectures being proposed today. He sipped from his coffee before asking the question of how many consensus nodes should be functioning to achieve ‘true decentralisation’ for a blockchain platform.

Admittedly, there was a pause. I thought the question of the right number wasn’t so straight forward to answer, particularly objectively. No matter how much I may argue one position or the other, the definition of decentralisation I am conceptualising for my argument may be wholly different to the person I am talking to.

This is just one of many reasons as to why the meaning of ‘decentralisation’ is complex and deserves deeper exploration. There are clearly some nuances at play.

This essay attempts to explore some of these nuances. I’ll assume the reader has a practical understanding of blockchain functionality and public blockchain frameworks.

Vitalik Buterin wrote a concise piece about the meaning of different types decentralisation, which I will discuss and expand upon for this section.

In order to conceptualise software decentralisation, we have to accept that the word has many axes. Generally speaking, 3 axes can be identified — these axes are closely related but can operate independent of each other.

Architectural decentralisation

One of the virtues of implementing a blockchain is that it has no single point of failure. This benefit is intrinsic to designing a distributed network as opposed to a database that does not rely on one physical computer.

From this perspective, blockchains are architecturally (and intrinsically) decentralised based upon their distributed infrastructure design.

In answer to my friend’s query above, any blockchain platform is some sense decentralised without having to consider consensus nodes at all. Clearly, this may be unsatisfying for many people, as there is no mention of network maintenance, consensus nodes or even governance mechanisms.

Why are certain facets of public blockchains, such as consensus nodes, common in topics around decentralisation?

Political decentralisation

Political decentralisation can be thought of in the following way:

What is the number of people, organisations, institutions, or parties that control the computers that make up the very system in question?

Public blockchains can also be decentralised in this way –they do not aim to be controlled by a single entity. The salience of consensus nodes or block producers in decentralisation debates are widespread. After all:

There has to be a consideration to the number of people controlling the network and how governance is to be achieved — for example, how many different entities, organisation, or individuals are controlling the consensus nodes? ‘Controlled by one single entity’ itself needs to be broken down — we could be talking about governing organisations, not just consensus nodes/maintenance layer.

What if there were thousands of active consensus nodes for a platform but this was controlled by just one or two foundations? Do people generally consider this to not constitute decentralisation? Moreover, what roles do foundations have for public blockchains? Foundations may arguably have decentralised decision making processes in place but to what extent do they represent centralised power?

Now, in response to the above question again, if we conclude that no single entity controls the consensus nodes then the platform has reach political decentralisation to some degree. We will come back to this point later on.

Logical decentralisation

Logical decentralisation is the idea of having one ‘body’ of a thing- a single structure. Theoretically speaking, a public blockchain can be run across several computers, maintained by millions of consensus nodes and can still be centralised on the logical axis.

The complex infrastructure effectively presents itself as a monolithic structure — the public blockchain is run as ‘one’ blockchain. The centralised structure is composed of its coded structure and run within its technological design parameters.

If we return to the query again, it now doesn’t matter how many consensus nodes we might articulate as being the ‘correct number’. If we consider logical decentralisation, a public blockchain platform can never achieve decentralisation on these grounds.

(As a side note, it is important to consider the interoperability of public blockchains in light of logical decentralisation. If we have a ‘meta-network’ of public blockchains, have we reduced its logical centralisation? Is the answer binary, and can we never reach meaningful logical decentralisation by combining the network structures of all interoperable public blockchains?)

As discussed before, if there are numerous organisations or individuals (and thus consensus nodes) then by definition we have moved away from centralisation on the political axis.

However, there remains a further problem. We might conclude that decentralisation has not been meaningfully reached even if multiple entities maintain the network as consensus nodes (contrary to what the political definition allows).

The nuance is now not of type but of degree: “what is politically decentralised enough”.

Debates on decentralisation in blockchains often reflect this very issue and generally stems from this political axis. To highlight this point, I will use EOS.IO, an open source blockchain network that is governed by 21 block producers that ‘produce’ the next block. By its definition of being maintained by no single entity it achieves political decentralisation. Why can this be an unsatisfactory conclusion for some and not others?

Achieving meaningful decentralisation is not an exact science. When designing blockchain architecture, we ask how many block producers would be appropriate, how responsibilities should get assigned, what the criteria is for elections, how often they are voted on, can they be voted out?

There are numerous factors to consider. The point is that because there are intricate and interdependent design aspects of public blockchains at play, this encourages ever increasing disparate perspectives and opinions on how a chain can be a meaningfully decentralised (or centralised) public blockchain.

Do people judge the meaningfulness of decentralisation based on architectural prioritisation or vice versa? Does person A believe that having 109 block producers is a satisfactory number — prioritising scalability and preferring it to a version of the same platform which has 1,405 block producers? Conversely, what about person B who believes 1,405 block producers is more meaningful in achieving decentralisation (and thus better) even with the potential cost of reducing a platform’s scalability?

Our architectural prioritisation and our perception of meaningful decentralisation are connected intimately.

As well as people differing between themselves, our own internal judgement meaningful decentralisation is non-static. Judgements of decentralisation appropriateness often require case by case analysis of the blockchains in question.

This naturally leads to intra-individual differences. If I analyse the decentralisation of a Proof-of-Work (PoW) blockchain I might potentially put a greater emphasis on the number of mining nodes that operate within the network maintenance layer. However, I might also place less emphasis on the number of block producers in a Proof-of-Stake (PoS) blockchain if I am satisfied with how these block producers are voted in and the criteria needed to become one in the first place etc. I could perform ‘analyses’ and conclude that decentralisation has been meaningfully met for both cases — albeit through very different ways.

This essay should not insinuate that decentralisation is too complex a term to tackle. Quite the opposite. By starting to explore the ways in which we ascribe meaning to decentralisation in blockchain, we provide the groundwork to further necessary debate in this unique and evolving landscape.

Thank you Francesca for providing feedback for this essay.

Related posts:

To start answering that question means knowing all the social media accounts used to reach audiences, because you can’t make more effective what you don’t know about. Our research suggests that many…

Beira is definitely not what you would call a “touristic” attraction. The old, rusty buildings, are the only reminiscence of an ancient splendour that is nowhere to be seen today. Once upon a time…

Opening up the miniblinds on the north side here, I thought “none of the dark came in.” What a quirky-ass thought. You can’t let in more dark. And how grand in the dark it is to hear the crickets…