Who controls decentralization?

Anyone who delves into current issues in the crypto or Web3 space will inevitably come across a debate about decentralization. It is a perennial topic and seems to be the unique selling point (USP) of this new technology. Why is it the subject of so much discussion, and why is it so desirable? Is there a methodology for evaluating decentralization? I would like to explore these questions and present a possible tool for evaluating such claims.

What does "decentralized" or "decentralization" mean? 

As is often noted in these discussions, decentralization is not a binary attribute but rather a spectrum. This means that there is no threshold at which it can be said that a system is decentralized or not. Each system has to be evaluated separately based on its nature and functionality. For example, the Bitcoin network operates differently from the Ethereum network and must therefore be assessed differently. But before we get technical, let's briefly discuss why this attribute is so desirable.

In my opinion, a decentralized system is one in which no part of a subsystem has such a high degree of influence that its malfunction or misuse would significantly affect the system's ability to perform its core functions. And that is the value proposition of such a system: it is resilient and reliable. No one should own the system or be able to arbitrarily change the rules unless the majority of participants decide to do so. This keeps the infrastructure neutral and makes it essentially public infrastructure.

How do you measure the decentralization of a network? 

This is currently very difficult to measure because there is no precise definition and, as far as I know, no universal formula for measuring it. One approach is the so-called Nakamoto coefficient (NC). This coefficient is supposed to indicate how many entities have to work together to affect the functionality of a network. Put simply, the blockchain is broken down into its components, or subsystems. The NC indicates how many entities must come together to take over a subsystem in order to disrupt the functioning of a network. The larger the NC, the more decentralized and resilient a network is. The intellectual father of the NC, Balaji Srinivasin, has proposed the categorization of six subsystems of a blockchain: mining/valuators, clients, developers, exchanges, nodes, and owners.

Let's compare the Bitcoin and Ethereum networks within the mining/validators subsystem. In the Bitcoin network, an entity needs 51% of the hash rate to take control of or disrupt this subsystem. This network is secured by computing power ("proof of work"), the so-called hash rate. The higher the hash rate, the more secure the Bitcoin blockchain. Attackers could dictate the history of the blockchain by excluding certain transactions from new blocks or reversing other transactions. 

Ethereum's consensus mechanism works differently and, therefore, needs to be evaluated differently. It is not about computing power but about how much Ether is staked. Ethereum is secured by the so-called "proof of stake" consensus mechanism. It does not require computing power to secure the network, but rather economic security in the form of Ether to be deposited. A validator must deposit a certain amount of Ether to secure the network and receives a small return on the deposited assets. In Ethereum, it is enough for an attacker to control 33% of the staked Ether to disrupt this subsystem and prevent blocks from being finalized. Now one would need to know how much Ether is currently staked and how high the Bitcoin hash rate is to make an accurate comparison and multiply by the respective ratios to identify the NC for each subsystem. This has to be done for each subsystem, and the lowest value ultimately determines the NC.

The reason I've delved into this topic is relatively simple: Most of you have heard some experts say that some networks are not nearly as decentralized as they claim to be. My aim was to get a better understanding of these statements, arguments, and ways of thinking in order to evaluate this technology. This topic is also food for thought. How much influence do the core developers of each protocol have? How decentralized would Bitcoin still be if Michael Saylor kept buying more Bitcoin? Are there software components within the ecosystems that have a supermajority (such as Wallet Connect or the Geth consensus client)? What would happen if a critical bug were found in that software?

The topic of "decentralization" is very abstract and difficult to grasp, especially since there is no proper definition for it and it is interpreted differently by everyone. In my opinion, the NC provides a coherent approach that has helped me understand the subject better. It also provides a methodology to quickly evaluate a network. The exact calculation of the NC is more an art than a science because it depends on where exactly we draw the boundaries of the observation. The larger these boundaries are, the more complicated the observation becomes. It is also worth noting that some information is private, such as the fact that Node Operator I and Node Operator II may be the same entity, etc.

Overall, the discussion of decentralization in the crypto and Web3 spaces illustrates the complexity and multifaceted nature of the concept. Decentralization is more than a binary attribute; it is a spectrum that depends on various factors and needs to be considered in context. Nevertheless, assessing decentralization remains a challenging task, involving not only technical but also economic and social aspects. Despite the challenges and the abstract nature of the concept, studying decentralization is fundamental to the advancement of blockchain technology and its applications. It encourages critical thinking about power dynamics and potential risks, thus contributing to the creation of more robust and transparent systems.

I do not believe that we will ever be able to create a fully decentralized system, but just because something is a utopia does not mean that we should not strive for it.