Quantifying Blockchain Network Centralization
“Decentralization” doesn’t mean anything. It should mean something, given its popularity in discourse surrounding blockchain technology. Many projects aim for it as a target, but none can even provide a location for the firing range. We identified that as a problem in mid-2019 and thought, somewhat ambitiously, we should try to find a solution.
Today, as we approach the end of 2020, we don’t claim to have an answer. We have to acknowledge no simple method exists to determine when a network is “decentralized.” However, we have concluded the discussion around “decentralization” might benefit from anchors in the form of identifiable and quantifiable metrics. Our “Open Standards” project represents many hours of researching, critiquing, debating, organizing, and drafting to arrive at a set of data points we believe can be useful to push forward the conversation around the wishy-washy term.
Direct link to PDF: Ketsal Open Standards – Measures of BCentralizationlockchain Network
Our process has been deceptively simple. First, we chose to speak in terms of “centralization,” as that term is pinned down more easily than “decentralization” and has a logical connection to the term in that something “decentralized” is by necessity “not centralized.” Second, we have parsed “centralization” into three recognizable layers: computational, economic, and political, as these three layers provide a rather comprehensive landscape of the types of decentralization often talked about. Third, we produced objective data points we believed most relevant to each of those layers, then debated which ones were actually useful, which were duplicative, and which were not actually meaningful to “centralization.” Finally, we drafted the Open Standards document with the list of included data points, the list of excluded data points (capturing our rationale for their exclusion), and applied the included data points to the Bitcoin network in its current state.
Decentralized discourse discussing decentralization.
We have observed the pitfalls and difficulty – on Twitter and elsewhere – in quantifying “decentralization.” We understand the reputational exposure involved in submitting any work product to a highly opinionated space. We have thus taken care and time in our year and a half of research, debate, and drafting, and have sought and received feedback from industry experts throughout the process. (Endless thanks to Lane, Diana, and Inca Digital!) This is a longwinded way of saying we took care to hope you all find use in this, we have listened to and responded to critiques on it, and we by no means believe it is perfect.
We want this industry to mature, to challenge itself, and to grow as a result. As you read through, note that we had two additional goals beyond simply providing discussion points around network centralization:
To provide investors, exchanges, regulators, and others with a mechanism to hold projects accountable for their marketing language. If a project says it is decentralized, a counterparty should ask them to prove it by referring to hard data points, with a focus on those that are difficult to fabricate.
To provide developers with a mechanism to set goals for themselves. If a project wants to say it is decentralized, or to achieve decentralization in the future, it may use these data points as waypoints to guide its development.
In addition, if one wants to consider the question of “network maturity,” “sufficient decentralization,” risk stemming from consolidation of economic power, security risk stemming from computational centralization, or a range of other risks that arise when a small group of persons, computers, or network addresses control key features of a blockchain network, we believe these data points can prove useful.
We welcome feedback, iteration, debate, emotion, everything, really. So long as this pushes us closer to a future where the term “decentralization” carries comprehensible meaning, we’re happy, and thankful for the support we’ve received along the way.
