Viewpoint: The Web3 debate should focus more on types rather than the degree of decentralization

Original Authors: Divya Siddarth, Danielle Allen, E. Glen Weyl

Original Title: “The Web3 Decentralization Debate Is Focused on the Wrong Question”

Translation by: Eva, Chain Catcher

Web3 advocates believe that the scale of decentralization is unprecedented. Excessive centralization hinders coordination, undermines freedom, democracy, and economic vitality, and decentralization is a remedy. However, the term itself is too vague to serve as an ultimate goal. Getting the work done requires the right kind of decentralization, and we are concerned that Web3 has so far been on the wrong track.

We are more concerned about the degree of decentralization rather than the type. Focusing on the degree of decentralization leads Web3 advocates to mischaracterize the reality of existing centralization and the possibilities of pure decentralization. On one hand, existing "centralized" systems are not as excessively concentrated as Web3 advocates typically describe. "Traditional" banks delegate many activities to local branches, and even central banks are often consortia. Architecturally, "centralized" clouds are rarely so concentrated in practice; they are usually distributed across a range of geographic areas to train large machine learning models in a distributed manner.

On the other hand, many Web3 critics point to the extreme inefficiencies brought by proposed decentralized architectures and the inevitable re-emergence of "centralization" in Web3 (NFT platforms, currency exchanges, wallet providers). Furthermore, broadly aiming for greater decentralization also has significant limitations and trade-offs. For example, narrowly defined technical decentralization faces the contradiction between resisting censorship and embedding values, which often leads to degraded functionality or some centralized decision-making, as seen in content moderation mechanisms on decentralized social networks.

Thus, in a functional system, the degree of centralization and functional feasibility have (soft) limits. Rather than engaging in a misguided debate about whether next-generation technologies should be centralized or decentralized, it is better to discuss how to best achieve the ideal model of decentralization. And to clarify what we want to gain from decentralization.

The value of decentralization lies in genuinely empowering people to act decisively in their social environments while providing the necessary coordination mechanisms across environments. This stands in stark contrast to the current technological environment, where decision-making power over information, computation, moderation, etc., is increasingly held by authorities "far removed" from the relevant communities— for example, platform content moderation processes attempt to operate across communities and cultures but fundamentally fail in both respects. In this context, decision-making is detached from the application environment and made by those with no direct interest in the matters, preventing them from leveraging the rich distributed information.

Our view of decentralization is about coordination. It emphasizes solving problems through the collaboration of "local" units that gather in the social environments most relevant to current decision-making. This is not a new idea. The federal system in the United States, including local governments, state governments, and the national government, fundamentally derives from this principle of universality, much like the establishment of open-source repositories and similar wiki-structured information aggregations. The key is that these local units are composable— modular and interoperable, essentially capable of being "stacked" to a larger global scale, allowing decentralized systems to effectively address issues requiring centralized coordination. This model is called composable local control.

Composable local control distributes decision-making, leveraging core principles of markets and democracy. Those closest to the issues typically have the most knowledge and the greatest stakes, and it is through the aggregation, collaboration, and filtering of knowledge that the best decisions can be made.

Universality is the decentralized architecture and type that makes composable local control possible. However, the dominant trajectory of Web3 is unlikely to achieve this and may even contradict universality. Permissionless blockchains are established as distributed redundant ledgers, where storage and permissions are allocated by anonymous economic mechanisms and accessed through fungible, tradable resources (such as computation and tokens). This architecture is optimized for a highly narrow set of problem sets and, by its nature, cannot interface with the rich economic and social networks that need to be addressed. Purely financial systems have a well-documented history of concentrating wealth, information, and power, and the current Web3 ecosystem has taken these attributes to extremes. Thus, the benefits of redundant distributed ledgers are at odds with the affiliated networks and advocated forms of decentralization.

We remain optimistic about the potential of Web3. However, to realize this, we must take steps to make Web3 a network of networks rather than a ledger.

Most typical crypto projects, such as Bitcoin, currently achieve "decentralization" in what we call "distributed redundancy": globally, openly, consensus-based storage of a common, homogeneous dataset in many places. Distributed redundancy relies on three factors.

• Maximizing the removal of data from the social environment. (All interactions are reduced to transactions recorded in the ledger, and the external environment cannot be reflected in the technical architecture.)
• Solutions aimed at achieving universality. (The focus on "global" applicability requires all solutions to be applicable in all cases.)
• Reliance on global consensus and redundant verification using fungible resources for access. (Decision-making mechanisms are constrained by tokens or computational puzzles; those with more financial resources have more of these resources.)

Why are so many people so eager to pursue redundancy and universality? Theoretically, redundancy aims to prevent attacks. However, as we have seen in recent supply chain challenges and the concentration of most Bitcoin mining in a few pools, market efficiency tends to concentrate activities in super-large centers, which are often very vulnerable to shocks and disruptions (e.g., local Covid lockdown policies) or located in jurisdictions that may be susceptible to geopolitical risks (e.g., China and Russia). Effective and secure redundancy requires intentional compensation, choosing diversified "hedging" risks rather than merely the lowest-cost suppliers. But achieving such hedging requires tracking the regional and network relationships that pure financial systems overlook.

In stark contrast is the desirable type of decentralization, universality, which focuses on:

• Bringing data as close as possible to the social environment of creation.
• Connecting and integrating multiple solutions through mechanisms of coordinated alliances and interoperability.
• Leveraging and expanding relationships of trust and institutions both online and offline.

Perhaps the original network of networks, the TCP/IP-based internet, was designed as an affiliated system from the start, precisely for security and efficiency reasons, and can be said to have achieved unprecedented success. More examples include:

1. The ActivityPub standard for social networks and related applications like Mastodon.
2. A series of research projects from Ink and Switch, including architectures for "locally prioritized" computing and interoperability designs between different programs.
3. A series of socially local identity systems, such as Spritely, BrightID, and BackChannel.
4. Federated learning and broader privacy-preserving machine learning.
5. Mesh networks.
6. Data cooperatives, cooperatives, and trusts.
7. Wikipedia and wiki-based content structures, more broadly.
8. Community content moderation systems, such as Reddit.
9. Community-first cloud computing, file storage, and time-sharing.

Unlike distributed redundancy, universal redundancy often leverages trust to enhance efficiency rather than reducing efficiency to eliminate the need for trust. For example, in community mesh networks, communities initiate decentralized wireless networks through locally installed shared nodes and antennas. Creative economic incentive design is crucial for the sustainability of such networks, but these incentives are embedded in social relationships rather than serving as substitutes for social relationships. Similar principles are the foundation of recent blockchain alternatives.

The following illustrates the distinction between universality and redundancy in some areas that have potential implications for Web3.

Identity and Reputation

Commitment: Web3 promises to liberate identity and reputation from the control of a few large tech companies, allowing for communication, transactions, and governance with "self-sovereign" identities.

Redundancy: As the fundamental data architecture of Web3, pseudonymous ledgers are not suitable as a basis for identity or reputation. Given the ease of establishing multiple accounts on pseudonymous ledgers, false attacks (or multi-identity attacks), where individuals seek undue influence over a platform by controlling multiple identities within it, are common. Solutions predicated on redundancy focus on universal, decontextualized unique cryptographic identifiers. Removing context leads to reliance on "universal secure" identifiers based on biometrics/general traits, which often raises at least as many concerns as the centralized protocols they replace.

Universality: Trust is a fundamental component of identity, and most interactions that rely on identity mechanisms for proof or verification are more about relationships (identities of employees, citizens, students, platform contributors) than about universal identity. Since the early days of the internet, network-based identity authentication methods (often referred to as "trust networks" or "IP trust") have envisioned verification based on strong but often informal trust relationships; examples of recent protocols built on this framework include Spritely, BackChannel, KERI, Āhau, and ACDC.

Data Empowerment

Commitment: Web3 claims to allow data creators to "own" their data and profit from it while protecting their privacy.

Redundancy: The typical vision of data ownership centers on the concept of private property of data in "personal data storage," which can be freely traded and connected to "markets" through DeFi structures. However, this structure is unlikely to facilitate data authorization in a narrow sense for several reasons:

• Most data is relational (e.g., emails between people, shared genetic data among family members, social graph data), thus the concept of private property fails. If anyone can block a transaction, the data becomes unusable; if any individual can authorize a transaction, competition arises as each data holder tries to sell in front of others.
• Most uses of data rely on aggregation, limiting the bargaining power of individuals without collective organization, much like industrial-era workers needed collective bargaining.

Universality: An emerging model of data management assistance combines social and legal structures such as data cooperatives, collaboration, and trust with privacy-preserving and enhancing technologies for data processing, such as federated learning and secure multi-party computation.

In these models, cooperative organizations responsible to and socially connected with data subjects (ranging from local governments to worker cooperatives to credit unions) collect and manage socially entangled interpersonal data. These organizations can negotiate with companies and other entities to establish guidelines around the use of shared data. For example, a credit union could act as a steward of member data, exchanging specific insights only with startups that establish loan refinancing tools or public sector agencies aimed at improving financial policy, while maintaining the privacy of the underlying data and adding value to the ecosystem and passing benefits to members. Such stewards could further interconnect with networks of other credit unions for better leverage and benefits. Similar structures could be used for various needs, such as Covid-19 contact tracing or tracking carbon emissions, releasing significant public benefits while protecting individual and community decision-making.

Organizational Innovation

Commitment: Web3, particularly decentralized autonomous organizations (DAOs), promises flexible, lightweight, accountable organizations and the empowerment of peer-to-peer, fully inclusive communities.

Redundancy: So far, fully automated organizations have failed due to the inability to specify relevant contingencies in error-free code. Flexibility and automation are in fundamental tension, as permanently automated processes are highly susceptible to oversight errors or mispredictions and cannot adapt when faced with situations not covered by the initial code. After all, automation is about universal and repetitive rules; flexibility requires the opposite. In human cognition, flexibility comes from judgment and creation of specific cases. Therefore, DAOs heavily rely on governance, but in the absence of identity infrastructure, they primarily depend on a one-token-one-vote structure, which is easily susceptible to attacks, such as the risk of venture capitalists controlling 51% of tokens. Relying on informal DAOs is often Web 2.

Universality: Partly due to dissatisfaction with current DAO structures, platform cooperativism, exit communities, meta-governance, RadicalxChange, and other related movements have been developing tools such as community currencies, soulbound tokens, [quadratic voting](#:~:text=Quadratic voting is a collective,voting paradox and majority rule.), new democratic deliberation tools like Pol.is and Loomio, and new ways to fund emerging democratic accountability organizations, such as secondary financing and Gitcoin. These tools focus on community engagement and empowerment, uniting organizations to build larger-scale cooperation rather than acquisition or purely financial contracts. While certain functions are most effective when automated, these processes achieve adaptability that is crucial for organizational operation.

Web3 has sparked important discussions about decentralization. However, it is time to harness this energy to achieve the best outcomes of decentralization: universality, not redundancy— a network of networks, not a distributed ledger.

In the context of the current Web3 ecosystem, many goals can be achieved. Non-transferable soulbound tokens, retained by the original issuers and thus non-financialized, have great potential to allow for social identity and community autonomy, even in cases where privacy is not a primary concern (e.g., things a person puts on a public resume or includes in a tweet). In such a soulful ecosystem, the transferability and transparency of these trust-based, non-transferable tokens can not only create a more personality-based network but also enable innovative voting and community governance protocols.

In fact, many such protocols have been thriving in experimental forms that are not entirely secure, such as four-dimensional voting and funding, voting and collective funding mechanisms that allow for more nuanced preference expression and aggregation. Multi-signature accounts are empowering communities with asset management authority. There are a series of serious attempts to establish data alliances that heavily rely on blockchain. It seems possible that various privacy-enhancing technologies can be integrated with the existing Web3 ecosystem, further extending these technologies and striving for more affiliated structures.

More ambitious projects suited for universality have greater potential in addressing meaningful coordination challenges. These projects may face some obstacles in achieving scale:

• Neither serving the current profit-seeking incumbents' power nor encouraging speculation based on their global revolutionary commitments to globally fungible currencies.
• Given the intention to collaborate with and leverage existing social structures rather than attempting to replace them, the implementation path is more long-term and socio-technological.
• Less directly connected to the current energy surrounding blockchain, thus benefiting less directly from the resulting hype.

However, we see these challenges as a call for coordinated, multi-sector investment. While this path may not be as straightforward, it holds greater potential for social transformation. Today's internet developed through multi-sector cooperation among the U.S. government, academia, industrial research labs, and internet service providers, supporting affiliated decentralization on the basis of a public mission. The major coordination challenges of our time, from crisis response to global governance, require carefully designed universality to succeed at scale. We believe the elements of universality outlined here can provide the foundation for such a system.