Disclaimer: I wrote this blogpost as I was getting into the topic of decentralisation. I did become more understanding over time. A follow-up can be found here.

A word that you keep on running into when getting into the web3 space is 'decentralisation'. Web3 educators, like Patrick Collins, emphasise decentralisation as the unique property of blockchains that enable the emergence of ‘Defi’, or decentralised finance, as alternative to the traditional banking system. Aragon and DecentDAO, organisation that provide solutions for on-chain governance, offer apps to help founders ‘progressively decentralise’ their on-chain organisation. Arbitrum, one of the largest layer-2 chains, is actively working toward decentralisation with governance moving from its foundation to its Arbitrum DAO. Decentralisation emerges as the defining feature of blockchain applications: they are designated as dApps and DAOs, or decentralised Applications and Decentralised Autonomous Organisations.

I've been around the web3 space for a little bit now and…  I have questions.

First, not to add to an already confusing situation, let me clarify what I see as centralisation and decentralisation. I take centralisation to mean, simply, the process through which communities become more centralised. It implies an increasing division between a centre and periphery, expressed in some kind of inequality. A centralised community is one where a small elite holds all political or economic power; where one institution has so much power, that it leaves its members without means to challenge its decisions. Decentralisation, then, is the opposite to the above. It is the process of decreasing division between a centre and periphery. It implies decreasing economic, social, political or institutional inequality.

The first reason for the confusion is that blockchains are highly centralised entities. There is a classic blog-post by Balaji S. Srinivasan, published in 2017, that lays out two different approaches to measure centralisation on bitcoin and  Ethereum. It shows that Ethereum is highly centralised at every subsystem (mining, clients, developers, exchanges, nodes and ownership) and with each measure the authors used. Of course a lot has changed in the last seven years. Not least Ethereum's move to a Proof-of-Stake method to secure consensus on its distributed ledger. Although figures are improving, the market share of the two largest staking pools - Lido and Coinbase - is still dangerously close to fifty percent. Even after it’s move to proof-of-stake, centralisation remains an issue.

But the likes of Patrick Collins, Arbitrum and Aragon are not talking about the state of blockchains as such. Rather, they are referring to the communities and decentralised autonomous organisations that blockchains facilitate and enable. Fair enough. So I went down the rabbit hole of web3 communities.

I still ended up confused.

Do blockchains really enable decentralised organisations and communities? Decentralisation is possible on a distributed ledger, but it is also possible without a distributed ledger. The decentralised uprisings that rocked the Arab world in 2011 are an example, as are many, many, other decentralised social movements.  The distributed properties of blockchains are not a necessary condition for decentralisation. They do not, as such, enable decentralisation. It is also possible without a distributed chain. Blockchains bring nothing new here.

Ok, then blockchains surely facilitate decentralisation. Right? If this were the case, blockchain communities would end up, on average, less centralised then their non-blockchain counterparts. There is some data that can be used to assess if this is indeed the case. Increased centralisation will reflect in unequal wealth distribution. Let us take the Arbitrum chain mentioned above, and map out wealth distribution in its community. The following is a chart of the 200 top owners of the $ARB token [1]. It excludes the two largest accounts, as they relate to the Arbitrum foundation itself.

It just looks like wealth is very much centralised around a few addresses. To drive home this point: The smallest account in this case (number 200) still holds almost five million ARB tokens, worth almost six million dollars at the time of writing (end of April 2024).

It is quite a damning first assessment, but let’s be a bit more systematic. The standard way to quantify inequality is through the Gini-coefficient. I’ll let others explain how to calculate it, for now it suffices to state that a Gini of zero is a community that is completely equal; and a community with a Gini of one is completely unequal. With wealth inequality IRL, Brunei has the highest Gini with 0.889; Slovakia the lowest with 0.55. Let’s take five of the most active DAOs from tally.xyz, a website for on-chain voting. These DAOs are Arbitrum, Uniswap, GitCoin, Nouns DAO and Lil Nouns [2]. Doing a bit of back-of-your-napkin-maths [3], their Gini-coefficients are as of writing (April 25, 2024) [4]:

None can be called equal, but the larger DAOs are just insanely unequal with wealth centralised around a tiny economic elite. Worse, in this case economic and political power are directly related: you vote with your tokens. Whereas in democracies one person has one vote, in DAOs that use coin voting, as is the case with these DAOs, it is more like one dollar is one vote. Economic centralisation directly translates to political centralisation. Attempts to mitigate the power of economic elites, for instance through advanced voting mechanisms such as quadratic-voting, reduce the effect of economic power on voting power, but do not mitigate it. It leaves most members of these communities powerless.

This has some serious repercussions for the extent that community members engage in these communities. Most proposals in the Arbitrum, Uniswap and Gitcoin DAOs passed with huge majorities. Majorities that are reminiscent of parliaments in autocratic countries that are controlled by a single party that rubber-stamps any proposal brought in by the executive. To an outsider this is a clear sign of highly centralised political power. Another sign is voter apathy. And sure enough, on Arbitrum most proposals had a voter participation of less than three percent. At Uniswap the highest voter participation almost reached ten percent, with most far below this. These numbers would bring any established democracy directly into a tailspin. Among DAOs, it is actually not all that bad. At Gitcoin DAO around fifty addresses voted. Out of 18.680 registered ones. It is a voter participation of less than 0.3 percent.

So, after all this I am still confused.

All this talk about decentralisation implies that it is very important to many in the web3 space. Meanwhile, in practice, it seems that blockchains do not enable decentralisation, did not facilitate decentralisation and are not decentralised in their day-to-day governance. So why, after a decade, has there not been an effective solution to this problem?

seven cedars

• [1] Data downloaded from Arbiscan, on April 24, 2024. Contract address: 0x912CE59144191C1204E64559FE8253a0e49E6548

• [2] Note that this list changes from day to day. This list was create at the end of April 2024. I excluded OpenDollar as it had only one proposal at the time. These organisations are ranked by number of votes. DAOs with no new proposals in 2024 are excluded.

• [4] Data from etherscan.io and Arbiscan; based on addresses provided at tally.xyz.

• [5] The largest five addresses were skipped to excluded addresses that belong to the DAO. Additionally, data downloaded from Arbiscan was incomplete. The first 500 addresses were reinserted; many addresses after these first 500 are missing. This Gini coefficient is an approximation.

• [6] The largest five addresses were skipped to excluded addresses that belong to the DAO.

• [7]

  The largest five addresses were skipped to excluded addresses that belong to the DAO.