This article is part of a limited series of overviews on DAO fundamentals, a collaboration between the DAO Research Collective and The Defiant. Visit www.daocollective.xyz for more information and continue following The Defiant for future pieces on DAO governance, treasury management, community, employment and other topics.
Decentralization is a concept that historically describes local governance structures where the responsibilities of planning and decision-making are not made by a centralized authority, but rather distributed throughout its membership.
Although the terms “centralization” and “decentralization” were not coined until the 19th century in relation to significant political upheaval across Europe, the concepts have existed since the very inception of society. In spite of the obvious benefits of decentralization regarding inclusivity, representation and personal freedom, human history has primarily been a study in centralized authority because of the benefits it provides in efficiency in making decisions and its ability to be effective over large geographic areas.
However, recent technological advancements have allowed for decentralized principles to be utilized more effectively and many of the historical limitations are no longer applicable as robust mechanisms of governance utilizing decentralization are now available.
A Mechanism and a Choice
In particular, blockchains have emerged as an opportunity for the development of new systems that more effectively embrace the benefits of decentralization as both a mechanism for member representative governance models and as a choice for users to access and develop application functionality within a decentralized economy.
While the historical definition of the term is still useful as a comparative measure to centralization, the term “decentralization” itself has become synonymous with certain features of blockchain technology and a point of analysis in determining their technical, legal, economic and political function. Although there is still no unified singular definition of decentralization, the utilization of the term within blockchain generally carries certain characteristics that trace back the original Bitcoin whitepaper.
While the term decentralization is not explicitly found within the whitepaper, the ability to construct organizations and processes with “no central authority” absent a “trusted third party” for transaction or otherwise interacting with others has established a stand alone concept utilized within the industry. Developments in the application of decentralized technology, regulatory actions, legal analysis, economic potential and political theory have provided additional context for what decentralization means as it pertains to public blockchains and the applications built on top of these blockchains.
The goal of this paper is to help clarify why people in the broader crypto community organize around the concept of decentralization. Although a unified comprehensive definition would undoubtedly provide more clarity than the current contextual definition, the reality is that decentralization within the blockchain is an evolving concept and at this time, its utilization must be examined situationally to determine how the underlying processes fit within its current application.
In this summary paper, we will explain the different components of decentralization, the benefits of decentralized systems, examples of how different projects have approached the process of decentralizing, and good-faith critiques of decentralization. This paper will act as a TL;DR summarizing the concept of decentralization within its current usage, while providing linked resources throughout for those interested in digging deeper into specific areas.
This piece is not an attempt to reach an all-encompassing definition of the term “decentralization” or an objective measure of project decentralization. Vitalik Buterin, co-founder of Ethereum, wrote a piece entitled The Meaning of Decentralizationthat highlights the difficulty of precisely defining the term.
Attempts at measuring decentralization include: Miles Jennings’ paper about the principles and models of decentralization, Balaji Srinivasan’s post about the Nakamoto Coefficient; a Beijing Jiaotong University paper using various metrics to gauge decentralization; and Ketsal’s post describing open standards for measuring blockchain decentralization. Given the evolving nature of the space, we will publish updated versions of this document as the topic continues to evolve.
Decentralization Standards for Layer 1 Blockchains
The core value proposition of many blockchains, including Ethereum, is to act as a trustless infrastructure where developers can build immutable, decentralized applications. While other blockchains are working towards progressive decentralization, Ethereum’s first-mover advantage and wide adoption as the first smart contract platform — i.e, blockchains that natively enable smart contracts, allowing for a variety of composable applications to be built on top of the blockchain — makes it a natural benchmark for Layer 1 blockchain decentralization.
Coinbase engineer Yuga Cohler went so far as to say that Ethereum’s upcoming transition to a Proof of Stake consensus mechanism will, if successful, “prove the viability of decentralization as a social organizing principle.”
Applications built on top of a Layer 1 blockchain inherit some of the decentralized attributes of the base layer, but just being built on top of a decentralized layer does not mean that applications are necessarily decentralized themselves.
While the application inherits the immutability and censorship-resistance of the underlying base layer by default, each application makes design trade-offs that impact how decentralized the individual application is. In other words, the decentralized infrastructure layer provides a base where decentralized and centralized applications can work in conjunction, with each application and corresponding community making decisions as to what trade-offs to make to achieve their desired state of decentralization.
Ethereum’s level of decentralization is not without critique. Liquid staking derivative centralization and majority client risks have frequently been discussed as potential centralization challenges for the Ethereum blockchain, and both of these critiques revolve around unintended centralized points of failure that could potentially arise in the future of the network.
‘Simpler is Better’
The specifics of these two concerns are outside the scope of this piece, but have been discussed at length elsewhere. For more information on liquid staking centralization, refer to this post on the risks of liquid staking derivatives by Ethereum researcher Danny Ryan and research from decentralized staking provider Lido. For more information on majority client risks, refer to this post by Ethereum researcher Dankrad Fiest and the Ethereum.org section on client diversity.
The Ethereum blockchain can be considered a “sufficiently decentralized” infrastructure to serve as the foundation for a discussion of decentralization. The Bitcoin blockchain would certainly qualify as “sufficiently decentralized” as well, but by design, Bitcoin has less functionality embedded into its protocol compared to Ethereum’s more flexible smart contract platform.
The Bitcoin community largely prescribes to an ethos of “simpler is better”, where Bitcoin itself is an expression of decentralization, since simplicity creates less vulnerability than more complex protocols. This post won’t get into the details of that argument, besides recognizing that Bitcoin would qualify as “decentralized enough” and that Bitcoin has been building out the Lightning Network to support more decentralized applications built on top of the network.
Ethereum possesses a more robust application ecosystem from which to discuss the varying levels of application decentralization, but Bitcoin’s achievement as the first decentralized blockchain paved the way for the entire industry.
For more basic context on Ethereum specifically and blockchains overall, Bruno Lulinski, co-author of this paper, wrote A Simpler Guide to Ethereum that goes over an introduction to blockchains, DeFi, NFTs, the Ethereum community’s decision-making process, and the future of Ethereum scaling solutions.
The Different Components of Decentralization
Decentralization is essential to the value proposition of several parts of the broader crypto ecosystem, so it makes sense to view the meaning of the term relative to the area in which it applies. The elements of decentralization are both discrete (i.e, “is the project decentralized in this specific area?”) and related (i.e, “how does decentralization in one component impact a project’s attained decentralization in another area?”).
Since decentralization in the context of one component means something different from decentralization in the context of a different component (while at the same time sharing underlying activity that impacts the other categories), projects need to consider each component to be able to function as intended.
The components of decentralization are broken into three broad categories that relate the effectiveness of decentralized systems across three axes; technical, economic, and legal decentralization.
As Miles Jennings stated in his in-depth piece discussing principles and models of decentralization, “The effectiveness of these decentralized web3 systems will depend upon their security, economies and parity of information” – each of which corresponds to one of the three listed components.
Technical decentralization – A global permissionless infrastructure layer and the applications built on top of it require credibly-decentralized technical underpinning. The underlying blockchain provides the execution layer for the on-chain components of the individual applications, but the applications themselves still require their own technical decentralization in the form of permissionless clients for interacting with the underlying smart contracts, user-owned data (and ease of data portability), and decentralized governance of the smart contracts by a broad group of participants in the form of a Decentralized Autonomous Organization (DAO).
Questions to ask when considering technical decentralization: How are these systems designed? How are upgrades made, if needed? What processes backstop the execution of upgrades (i.e, Compound’s 48 hour timelock)? What blockchain underpins the application, and what tradeoffs does that blockchain force onto the application? Can users easily ‘ragequit’ the system, i.e, can users exit the system and use (or build) different methods of interacting with the core protocol?
From the perspective of determining the decentralization of the blockchains themselves — how many clients are there, and what’s the distribution amongst clients used for miners/validators? How can individual participants verify the authenticity of the given blockchain, and how difficult is it for an individual to participate in that verification process? There are manymore ways to consider the technical decentralization of blockchains.
Ultimately, technical decentralization is the necessary foundation upon which economic and legal decentralization can occur.
Economic decentralization – Public blockchains create the opportunity for a reimagining of the economic interaction between the developers of an application and the users and adjacent stakeholders around that application. In the ‘traditional’ pre-blockchain world, companies are incentivized to view their users as a source of value extraction, primarily in the form of user-generated content or the corresponding data of the end user, which is then transacted between the company and willing advertisers behind the scenes.
Blockchains allow for systems that are not reliant on central leadership, allowing for the balancing of incentives between developers of the application, contributors to the application, and users of the application. These economically-decentralized structures are basically a new generation of open-source software communities, but with embeddable, transparent economies. In an economically-decentralized ecosystem, contributors can participate in the value-creation of the application while receiving compensation for their contributions.
Questions to ask when considering economic decentralization: How is the underlying token of the application designed and distributed? How was the airdrop designed, and what considerations were made by the early project developers to prevent centralized ownership of a majority of project tokens? How are early investors and project contributors compensated, and what do token lockups for all parties look like? How do distributions of the DAO treasury work, i.e, how are funds distributed to initiatives and/or working groups intended to further the project’s mission?
Legal decentralization – Beyond the technical mechanics and the economic benefits of decentralization are matters of regulation and legality, including taxation, liability, ownership, intellectual property, reporting and privacy. Although U.S. securities law is an essential area of analysis in determining how decentralized systems may make use of digital assets, it is not the only area of law impacted by the decentralization made available through public blockchains.
Although decentralization exists in the current legal system – most obviously in the form of general partnerships – there is significant question as to how the default rules established for participation and responsibility can be fairly applied to decentralized systems that are exceedingly dissimilar from the activity giving rise to existing law.
Going beyond the superficial similarity to existing rules and laws, the decentralized activities available through the blockchain represent significant changes in concepts like equity, ownership and control. These differences underscore a different relationship and responsibility than the activity giving rise to current laws and regulations and create significant uncertainty in how decentralized organizations that exist on the blockchain will be treated.
As these activities are capable of creating taxable events and acting in a way giving rise to litigation, it is expected that taxation and liability will soon be matters of equal prominence with securities law when considering legal decentralization.
Early projects necessitate some form of central leadership and planning to define the project’s purpose and provide crucial activation energy. These teams might retain some influence in the soon-to-be-decentralized project, but the level of influence retained could have a significant impact on whether the project is considered decentralized from the perspective of regulators and other government authorities.
Levels of Influence
Securities regulation stems largely from a desire to prevent information asymmetry amongst market participants. While there is no defined standard for the concept of legal decentralization, the levels of influence that early project contributors retain in the decentralization process, as well as the transparency of information amongst participants, will be pivotal to determining whether a project is legally-decentralized.
Many have written high-quality overviews on the topic of decentralization from the perspective of securities regulation:
Questions to ask when considering legal decentralization – how much influence does the early project team have, and where does that influence come from? Does their influence stem from outsized voting power retained in the supposedly-decentralized organization, or from their voice in communal decision-making processes? How much influence do early investors have? Can community members be held accountable by other community members, and does the project depend on the efforts of a central authority? Do different stakeholders have asymmetric information based on the structural design of the organization? More on legal decentralization in the securities law section below.
The Benefits of Decentralized Systems
As discussed above, the term ‘decentralization’ is itself a reflection of the term ‘centralization’. Looking at the attributes that a decentralized system might have, like censorship-resistance and distributed decision-making, makes it easier to visualize the concept.
Censorship-resistance is the idea that no single governing authority can unilaterally make the decision to restrict another participant’s actions in a network. Historically, coordination between humans has relied on some levels of trust.
Trading goods between two people requires the trust that both parties will actually deliver their goods to the other, and agreeing to some sort of truce or treaty between conflicting nations requires trust that the other party will continue to abide by the agreed-upon treaty.
Immutable code deployed on decentralized public blockchains sets the foundation for censorship-resistant, privacy-preserving innovation. These censorship-resistant systems are not yet completely un-coercible, but they act as a necessary foil to the institutions and platforms we’ve come to rely on outside of crypto (world governments, social media platforms, etc.). If the infrastructure layers (the blockchains themselves) weren’t decentralized, it would be simple for a powerful government to shut it down — just find the party responsible for the network and coerce them.
Decentralization makes this difficult, as China’s Bitcoin ban demonstrated, because censoring sufficiently decentralized systems requires coordination outside of the scope of most governments; just a few months after the China Bitcoin mining ban, several underground mining operations emerged in China to fill the gap left behind by the ban.
NFTs have been used by individuals to preserve information in the face of authoritarian governments as well – however, these NFTs still required anonymity from the individuals to avoid direct coercion from their government. Even democratic governments like Canada have recently expressed a willingness to exert powers of censorship by coercing financial institutions to financially censor some citizens.
Power and Influence
Other countries, like Ukraine, have effectively used the censorship-resistant quality of public blockchains to fund their defense when cross-border money transfer companies initially capped transfers of money to Ukraine (these caps for transfers to Ukraine were later relaxed).
Tech giants like Apple, Facebook, and Google have scaled to huge amounts of power and influence, throwing them (willingly or not) into public debate about the interactions that happen on their platforms (and the processes that guide their frequently–controversial decisions). Twitter is frequently used by governments to communicate directly with their constituents, and offers a great example of the benefits of decentralization – in 2018, Twitter removed access to a variety of APIs that independent developers had used to build applications on top of Twitter.
A decentralized system would be censorship-resistant to decisions like Twitter’s. In fact, transparent, unopinionated rules to participation are by themselves a flavor of censorship-resistance that blockchain-based applications naturally inherit, since by default, code deployed to public blockchains is open-source.
Even if Twitter’s former CEO, Jack Dorsey, had committed to an open protocol and long-term neutrality (as he later lamented), the promise of a censorship-resistant Twitter will always fall flat on long enough time horizons — it’s the natural game-theoretic conclusion. Open-sourcing code and permitting user ownership of private data are concepts that are fundamentally opposed to the business models of corporations built on closed protocols which rely on the data of their end users to generate financial returns for their shareholders.
Resilience to Attacks and Decorrelation
Vitalik Buterin argues that decentralized systems are more resilient to attack and less likely to accidentally fail than their centralized counterparts.
Critically, decentralized systems are generally more expensive to attack due to the absence of sensitive central points of failure for attackers to target — an attacker couldn’t just infiltrate the Ethereum Foundation and press a big red “HALT” button (since there is no big red button), and an attacker couldn’t overpower Buterin and force him to shut down the blockchain (since Buterin doesn’t have that type of centralized power, despite being the initial founder and specwriter).
At the application level, trust assumptions, key management and security practices will differ, which means different levels of resilience to attacks. The Layer 1 blockchain’s resilience to attack will still provide a credibly-neutral, permissionless infrastructure for application developers to build from.
Decentralized networks also tend to breed duplicative systems, leading to more robust security. Tim Beiko, one of the lead coordinators for the Ethereum developer community, recently called this benefit of duplicative systems “uncorrelated failure modes” on Farcaster. The general idea is that multiple solutions, — i.e, different client implementations, different approaches to a specific problem, or just different schools of thought — reduce the probability of catastrophic failure across the stack.
(Beiko technically uses “decorrelation” and “uncorrelated failure modes” as a replacement for the term “decentralization” in this context, because of the difficulty of quantifying decentralization. We are using “uncorrelated failure modes” as a benchmark for a sufficiently-decentralized system. Therefore, from our perspective, a decentralized system would necessarily have uncorrelated failure modes, but either way, the sentiment is the same.)
An example of catastrophic failure at the hands of correlated risks is the 2008 Global Financial Crisis, where the risks created by the rise of collateralized debt obligations, credit default swaps, and frothy lending practices were improperly underwritten by ratings agencies. This highly-tangled web of risks led to correlated failures as homeowners defaulted, leading to lender defaults, leading to counterparty defaults, leading to havoc.
Underwriting systemic correlated risks is difficult and, in complex interconnected systems, can lead to disaster. Decorrelation of systemic risks via the open borders of blockchains can help mitigate these risks and reduce the surface area of attack vectors.
“If you had asked a normal person in 2007: “How would it affect your life if it turns out that investors have mispriced the super-senior risk in synthetic collateralized debt obligations built out of subprime mortgage tranches,” that person would have said “I have no idea what you are talking about, but I can’t imagine how that collection of words would affect me.” But it did.” – Money Stuff, May 12, 2022, Matt Levine
Transparent Incentives and Distributed Decision-making
While a shareholder in a public company might successfully petition the board of directors to include a shareholder proposal in an annual proxy statement, boards have some leniency in which proposals they have to include for discussion, and many large tech companies (Facebook, Snap, and Google, for example) have dual-class share structures that give insiders a supermajority of voting power, denying any significant outcome from stakeholders.
While not solely a tool for distributed-decision making, decentralized systems do enable governance transparency that has the potential to improve the equity and effectiveness of critical decision-making processes. There are good-faith critiques of distributed decision-making systems, including the necessarily-centralized focus required by early project teams, the tragedy of the commons, and the voter apathy that can manifest in these types of horizontally-distributed decision-making processes.
Governance of these distributed decision-making systems is a complex topic in and of itself (which will be discussed in a future DAO Research Collective piece). However, proponents of decentralized systems argue that transparently-recorded actions are one of the main benefits of decentralized organizations.
Critiques of Decentralization
Despite occasionally being misconstrued as such, a decentralized economy built on public blockchains is not a replacement for all centralized entities. Instead, it is an expansion of structures that will allow decentralized and centralized organizations to interact in ways that previously were not technologically feasible or practical. Still, there are several good-faith critiques of decentralization in the context of public blockchains, which are addressed below.
Only Perfectly-horizontal Structures Can be Considered Decentralized
To some, the idea of decentralization means perfectly unhierarchial structures, completely unstructured and without guidance or leadership. Some critics of decentralization argue that any influence exerted in a decentralized system by a small number of participants proves that the system is centralized, or that any attempt at structure preserves centralization.
This claim has been used to say that Ethereum isn’t decentralized. But as described throughout this piece, decentralization comes in various forms and must be viewed through specific frames of reference to distinguish between different types of decentralized systems.
Kevin Owocki of Gitcoin discusses this in a Twitter thread where he points out that decentralization can refer to “decentralized governance via a permissionless token” rather than a “chaotic loose web of individuals”.
In Anticapture, Spengrah writes that “conflating decentralization with permissionlessness is one of the most common mistakes in the DAO space.” Spengrah discusses the concept of anti-capture, a framework for how networks of humans can design systems resistant to governance capture by bad actors. Capture-resistance governance is a more reasonable goal for decision-making for projects that can’t be reduced to completely non-human programmatic functions.
Not Enough Recourse for Consumers
One concern about decentralization is that decentralized entities won’t provide recourse for consumers. The critique is that many (or most?) consumers will not care about self-custody of their assets, and many will be happy to claim ‘decentralization’ as a virtue of web3… until their Bored Ape gets stolen. If users lose their assets because of bad-faith actors or fat-finger mistakes, how will users get their assets back? How can that process work in a decentralized world?
In March 1933, as the U.S. recovered from the Great Depression of the late 1920s, President Franklin Roosevelt gave a speech regarding the banking crisis and the advent of the Federal Deposit Insurance Corporation, which insures depositors in banks for up to $250,000 in assets. His first sentence is demonstrative of the situation at the time:
“I want to talk for a few minutes with the people of the United States about banking — with the comparatively few who understand the mechanics of banking but more particularly with the overwhelming majority who use banks for the making of deposits and the drawing of checks.”
The Needs of Users
At the time (and still to this day), many didn’t understand the inner workings of the banking system. In the near future, many still won’t understand or care about the mechanics of the underlying blockchains, but solutions will exist to fill the need of the consumer (in this case, wanting to own a Bored Ape or participate in a DAO) without the consumer needing to understand the specific mechanics of the system. Through the process of iteration, society learns and improves. The same will happen to crypto.
Insurance will become more robust, and either the users or the protocols themselves (or both) will have programmatic insurance coverage for these types of damages. The user experience on wallets will improve (see Argent or Rainbow for a great wallet user experience), making it less onerous to self-custody, while still retaining the benefits of self-custody.
The fear of losing a 24-word secret recovery phrase will become irrelevant in a world of social recovery wallets, and solutions will continue to spring up to fill the needs of users who want to enjoy the benefits of decentralized systems while still ensuring reasonable forms of recourse for potential assistance in difficult situations. Most importantly, centralized organizations will coordinate with decentralized organizations to provide these solutions in a way that preserves the benefits of decentralized systems while creating a user experience that fits the desires of their individual user base.
Centralized Entities Prove Centralization
The centralized entities that exist in the crypto world — like Celsius, a centralized exchange — are often used as a demonstration of how the crypto world is not truly decentralizing. This claim appropriately targets the projects that claim decentralization as a selling point to attract users to what is a clearly centralized (by any definition) project, including many of the recent catastrophes in crypto (Luna and Celsius, for example). These should be critiqued as such.
But as described in detail throughout this piece, decentralized systems are not just completely horizontally-distributed systems and instead, there are several individual components to consider when judging the level of decentralization of a project. Crucially, the critique that “centralized entities within the system prove centralization of the system” often ignores the idea of data portability. Mudit Gupta, Chief Information Security Officer of Polygon, called data portability “the ability to be decentralized.”
Centralized systems can exist and create value for end users by making it easier to interact with permissionless blockchains, but ultimately, blockchains give users the ability to exit with their own data. If OpenSea, a centralized NFT marketplace, decides to censor a subset of the NFTs that are sold on their platform (by not displaying them on the OpenSea user interface), or if OpenSea decides to start charging higher fees to users, users can simply stop using OpenSea and move to another NFT marketplace.
OpenSea doesn’t actually hold user NFTs — OpenSea is just a venue for displaying and transacting (importantly, a venue with lots of liquidity, which makes for a more efficient marketplace and better price discovery, but a venue nonetheless).
Traditional internet companies don’t give users the flexibility of data portability because they’re not incentivized to do so, but blockchain-based applications necessarily have data portability embedded into their operations. While centralized crypto companies can create efficiencies (like organizing liquidity, providing customer support, and standardizing user interfaces), the user’s ability to exit the system provides a check on any centralized entities’ power over the system and ultimately, over its users.
Ineffective Governance and Potential Plutocracy
The critique that decentralization will beget ineffective governance might be the most truthful critique of the ecosystem right now. As of early 2022, governance of decentralized organizations is largely ineffective across the board — participation is low, and pure coin-voting, as most decentralized organizations have trended towards, has a variety of embedded issues that might create more plutocratic systems than the prior status quo.
The DAO Research Collective’s Governance paper will touch on many of these issues, and Fred Ehrsam of Coinbase / Paradigm wrote a prescient 2017 piece on blockchain-based governance systems that highlights some of the benefits and issues of on-chain governance, as well as future approaches. Ultimately, it remains to be seen whether decentralized governance can be as effective (or more effective) than traditional centralized governance systems.
In Part Two tomorrow: How Select Projects are Decentralizing
Bruno Lulinski is the author of “A Simpler Guide to Ethereum”, a guide for understanding the different components of the Ethereum ecosystem. David Kerr is the Head of Research at the DAO Research Collective, and the principal consultant at Cowrie LLC.