Decentralization is Not Binary

With fast-growing interest in crypto and the speed at which new protocols are coming out, the conversation around what constitutes ‘centralization’ vs. ‘decentralization’ is, as expected, getting louder.

The simplicity and lack of nuance with which we talk about centralization & decentralization has always made me uneasy. In both the crypto world and in the real world, we talk about centralization and decentralization as if they’re two distinct, mutually exclusive states.

“That system isn’t actually decentralized.”
“It will be ages before protocols reach true decentralization.”

We hear statements like these all the time — and, in the crypto community, centralization is viewed as a dirty word: an antiquated state of being that exists to serve the interests of those at the top of the hierarchy.

It’s time we shift the nomenclature around centralization and recognize it for what it is: a spectrum, and one that traverses multiple dimensions. We don’t do ourselves any favours by blindly talking about a ‘decentralized future’ as utopia, without considering the depth of its implications.

Centralization as a Spectrum

We often talk about centralization vs. decentralization as if they are two distinct states:

This is a massively over-simplified perspective that leaves out the missing middle. In reality, centralization is a spectrum:

To illustrate my point, I’ll share a basic analogy that keeps popping into my head whenever I think about centralization: a company’s organization chart.

Let’s look at the two extremes.

  • If we think about a completely decentralized reporting structure, it looks like a flat organization with no reporting lines.
  • If we think about a completely centralized reporting structure, it looks like a straight line from the CEO down.

If you’ve worked in any organization ever, your immediate reaction is likely that both of these are pretty terrible org structures, in most cases. Even companies that claim to have ‘flat structures’ nearly always have much, much more centralization than what’s pictured on the left-hand side.

The reality is that most companies have org charts that fall somewhere in between — some level of centralization, with everyone accountable to the CEO who keeps everyone on track and makes sure the show goes on; but also some level of decentralization, whereby teams are split out across the organization, generally based on shared capabilities or scope.

The org structure analogy offers a framework for thinking about centralization of systems, more broadly:

As made obvious when visualizing it, centralization is certainly not binary.

And, full decentralization and full centralization are not necessarily structures to aspire to. In many cases, there’s good reason to sit somewhere in the middle.

Moving beyond just company org structures, this spectrum also holds true. Writing this piece was inspired partially by David Phelpsrecent podcast with James Wang, in which they discuss the level of centralization of crypto protocols.

A nice real-world parallel that James suggested is the autocracy → democracy spectrum. Again, these are not two mutually exclusive states of being; they are part of a much broader set of options.

The Dimensionality of Centralization

Now that we’ve established a spectrum, it’s worth calling out that the centralization spectrum can apply across many dimensions — a system might be more decentralized in some ways, and less so in others.

There’s an excellent piece by Will Murphy proposing four levels of centralization: technical, governance, business, and state.

Whether or not one agrees with the framework of these four dimensions specifically, it’s plain to see that, in addition to thinking about decentralization as a spectrum, it’s also worth considering the dimensions of decentralization being discussed — while a given system may be decentralized on one of these dimensions, it may look very different on the others.

An illustrative, hypothetical example (of no specific system in particular):

This example explores centralization across four dimensions, and the levels may vary for very good reasons.

In the crypto world, Uniswap has come under fire for being ‘not truly decentralized’. There are also many thought pieces, like this one, comparing the relative decentralization of different networks. In the podcast I mentioned earlier, David coined a new term I love: DAO-ish — bringing refreshing nuance to the term ‘Decentralized Autonomous Organization’, because ‘Decentralized-ish’ is, in fact, a viable state of being, whether or not you agree with it in practice.

All in all, the discussion & debate around centralization is important, and we should be careful not to over-simplify. For any system we’re analyzing, we ought to be super clear on areas of nuance and what dimensions we’re discussing.

My hope is that we might come up with some common frameworks to serve as a starting point in bringing clarity and specificity to these important conversations — this is just a start, and I’d love to see more of them.