Deconstructing Decentralization

#Decentralization is an important concept in today’s internet - both for users and for business. The benefits to users have been stated many times - enhanced control, ownership, responsibility. However, the benefits are not as often defined or discussed from a business standpoint. If done right, decentralization can help reduce organizational risk and, when applicable, can be useful in connecting with your desired audience or consumer base.

The term decentralization is used a lot in #Web3 , but what does that mean, really? Consider a data center - large data centers handle incredible amounts of sensitive data spread across many purposes such as financial software, personal identity information, government information, cloud service infrastructure - the list goes on. If the data contained in the data center is not replicated somewhere else (i.e., the data storage is centralized) a catastrophic event at this one location could interrupt many services or destroy loads of unrecoverable information. As an additional layer - for the purposes of an extreme example - say only one person can access this data center to perform critical maintenance or upgrades. What happens if this person becomes unavailable for some reason?

In the context of Web3, some claim decentralization relates only to the distribution of nodes in a blockchain network, but decentralization is more than the hardware that enables a blockchain to function. People, processes, points of governance, and aspects of the technology stack are? important factors when evaluating decentralization within an overall Web3 system. Moreover, each of these factors can exist on a spectrum from centralized to decentralized, and they can be modified as needed depending on the use case. To offer a holistic view of decentralization, we argue that the level decentralization is customizable and your intended audience should be a major factor in that decision. We aim to define decentralization across various layers of a system, gauge the vision of “full decentralization” against the reality of a spectrum of decentralization, and address why decentralization matters.

Your audience is likely to have the greatest influence on your level of decentralization

Decentralization is a malleable concept! If you are in the fortunate position of being able to design a new system (system here is a flexible term - could be a network, an application, a protocol, an organization) you can intentionally choose how decentralized you want your organization or product to be. Later, we discuss various trade offs and the various pieces of a system that can vary in level of decentralization. But from the onset, the most important factor businesses should consider is their audience - or their intended addressable market.

Decentralization does not always need to be the North Star. For example, if your audience is a community that is expecting to have access to customer support or recourse for lost data, your system design is likely better suited toward the more centralized end of the spectrum where you will have more authority in accessing or recovering user information. If your target market is a community of artists or developers who are looking for new ways to monetize their work, you may lean toward a more decentralized model as they will likely value ownership and lower risk of censorship.

Even within a specific product type - wallet software for instance - the target audience should still dictate decentralization trade-off decisions. A wallet designed for users who wish to achieve sovereignty over their funds vs a wallet that is designed to reach the average user should be designed differently. The sovereign individuals are not going to want you to be able to access their wallet information, but the average user will likely desire recoverability and guarantees against lost funds.

Decentralization varies across layers of a use case

Stated plainly - for the purposes of a Web3 system - we define decentralization as the distribution of trust, power and/or control. If power is more decentralized, users generally have more control over the strategic decision making that shapes the direction of the Web3 system. As mentioned previously, this distribution of power may vary across a variety of dimensions, or “layers”. Here, we define the Web3 system as composed of three main layers: the Technology Layer, the Business Process and Governance Layer, and the People Layer. Within the technology layer there exists a more nuanced separation between technical components such as the ownership distribution of network nodes, the types of activities taking place on the blockchain network, or the distribution of the use of specific applications that are being built and utilized on top of the blockchain network. Each layer will carry its own level of decentralization. A system could, for example, have a well-distributed set of node controllers and operators, but perhaps a large majority of activity is taking place on one application. The distribution of node controllers is not going to save the system from an attack on or failure of that application.?

The pursuit of decentralization can come at a cost

It is likely that the level of decentralization at each layer will be unequal. Different technologies, use cases, industries, and organizations will carry different degrees of decentralization, with some organizations endeavoring to achieve 100% decentralization across all aspects of the business. But how feasible is that? 100% decentralization would require, among other things, many individuals running and maintaining their own hardware, many protocol developers with differentiated opinions, and a vast ecosystem of applications built and run by discrete teams. These are just simple first order requirements - for instance, second order requirements such as funding, liquidity, and influence would also need to be varied and distributed across the protocol and applications as well.

Decentralization occurs on a spectrum

Given that the feasibility of full decentralization is relatively low, the question becomes how much decentralization can, or should, you achieve? It is not a foregone conclusion that the amount of decentralization your system carries is static - decentralization varies over time and exists on a spectrum. Throughout the process of designing a system, trade-offs will present themselves when choosing between more decentralization and less decentralization. For an ecosystem at large, one can attempt to gauge decentralization by examining single points of failure - if one developer quits, does the ecosystem falter? If one application gets exploited, is the entire ecosystem at risk??

Decisions that increase decentralization often increase friction for users who are unlikely to want to carry the additional risk or responsibility that comes with running their own hardware. This presents an important consideration: users may desire decentralization as a core feature of the system over their own experience - at what point does this tradeoff no longer make sense?

For any given use case, the sum of its components will serve to define its overall decentralization. When designing a system, an important first step is to consider where you want to live on this spectrum.?

Decentralization is important for both individuals and organizations

One of Web3’s major promises is that control and ownership will continue to shift towards users. When control and ownership is distributed, so too is risk. This is particularly notable when considering recent market events. When organizations and ecosystems create silos - silos of information, of money, or of developers - the risks are magnified. As a tangible example, if the ecosystem depends on only a handful of financial institutions, which could be interconnected themselves, any instability at one could cause damaging effects across the ecosystem at large. If these financial institutions are greater in number and more distributed in use by the ecosystem, it is a much more acceptable loss for one of them to fail.?

Decentralization matters. Yes, for users, who crave more control and more ownership, who want the ability to interact with their money, their data, and their networks as they choose. But importantly, and maybe less obviously, it matters too for organizations, who should look to isolate and distribute their risks and their potential points of failure. As organizations better distribute their risks, this can have a rollup effect, potentially reducing risk across entire sectors or industries. Continued evolution of Web3-focused technology, such as zero-knowledge proofs, enables decentralization of trust. Decentralized ecosystems should strive to refuse reliance on a few single entities - centralized risk and authority is largely how we’ve always done things - maybe now it's time to try something different.

As always big thanks to contributors Evan McGowan , Gabriel Blum and Alex Ferraro