I still vividly recall, like many other participants, the electrifying atmosphere of ZuConnect Decentralized Day last November. It was a full day of engaging panel discussions that delved deep into the heart of our mission. We began by drawing insights from the experiences of community organizers like Zuzalu and Liberstad. Our journey continued to weave through the lessons learned from architecture, sociology, and the ongoing evolution of protocols and applications.
These cross-discipline topics were thought-provoking. We embarked on a journey that spans across themes not commonly discussed together, and here are 3 areas that are especially helpful to me.
Trust in a Decentralized World
In most decentralized networks, users are free to join with anonymous or pseudonymous identities, and we often need a way to identify people we trust and maintain accountability in the network.
For projects like PGP and Secure Scuttlebutt, this is solved by the Web of Trust. This concept hinges on a simple yet profound idea: if someone I trust, trusts another person, then I can probably trust them too. It's a notion we apply in our daily lives, relying on our network of acquaintances to gauge trustworthiness.
When making decisions collectively, such as allocating matching funds in a quadratic funding round or ranking content within a community space, we must aggregate trust from everyone and guard against malicious actors employing Sybil attacks. This requires different approaches, and one common method is calculating an account's trustworthiness based on its connections within the network. It's akin to determining the "eigenvector centrality" of a network in graph theory. Well-established algorithms like EigenTrust in P2P file sharing and Google's PageRank are variations of this concept, making it battle-tested and well-understood.
A hackathon project led by @jpren and his team proposed an intriguing alternative: starting with a trust circle and democratically expanding it. Newcomers would mint a non-transferable token representing group membership only after gaining attestations from existing members. The criteria for "enough" attestations is based on group size, increasing the requirement of attestations as group size grows, whether through simple rules such as the square root or more intricate ones considering network structure. It's a captivating notion that begs exploration, particularly in comparing its resilience against Sybil attacks with the traditional eigenvector centrality calculation.
But what I am most excited to explore, is the idea of “pseudonym parties”, in which participants gather physically and receive certificates for being a real human. It was proposed in the early days of the internet to solve the conflict between anonymity and accountability and could be simplified now with non-transferable tokens (or SBT, Soulbound Tokens) on-chain. It can be scaled in a federated way, where several parties can take place at the same time while acknowledging the legitimacy of each other. Other than its effectiveness (that is, still in theory) in preventing Sybil attacks and mitigating fake news, it also provides a chance to bridge online and offline relationships, and enable people to form close-knit groups (I will explain why that is desirable in a sperate article).
Digital space and pattern language
Problems such as reputation are clearly defined in their objective, and the solutions should converge over time. For social networks, which impact our lives directly, there are many divergent problems. Questions on designing social networks for the benefit of individuals and collectives might only have indirect answers and without much consensus.
But when we start to view social networks as digital spaces, we suddenly tap into the rich experience and knowledge of architecture and urban planning, where such divergent problems have been discussed.
One method, or school of thought, called “pattern language” that originated from architecture is particularly insightful for our purpose. It is how we traditionally build our homes collectively: using a common vocabulary, based on patterns proven to work, to describe the relationships that we want to see in our shared dwellings. It is also what designers and builders can consciously document and articulate, to make it easier for people to participate in building a cohesive, poetic, and lively environment.
This mental model has affected the thinking behind wiki and scrum development, and now people are using it to understand community dynamics, study digital public spaces, and speculate a social network in plurality.
Many patterns in architectural designs could be borrowed directly, especially in issues related to community dynamics, and on how to integrate private and public spaces. For example in the seminal work A Pattern Language by Alexander et al., one pattern "Community of 7000" posits that "individuals have no effective voice in any community of more than 5,000-10,000 persons." Does it translate to the realm of digital communities? How does community size impact the facilitation of freedom of association?
Together with design such as Web of Trust, network centrality, or pseudonym parties, these could be the patterns that together weave a dynamic, plural, yet interoperable “social media landscape”.
The Internet as A Social Playground
In our quest to design alternatives using Web3 technologies, we must not overlook software and protocols that have fostered meaningful connections despite not being classified as traditional social networks.
A great example is GitHub, which is now used by over 100 million developers. It brings the social element to open source collaboration (remember the tagline “social code hosting”?), and in turn, forged meaningful social connections through those collaborations. It started when the open source movement was taking traction, and by making collaboration easier, it grew with this expanding and creative community.
Another example is email, which if counted as a social network (why not?), is larger than any other one. It is used by almost every internet user, and functions as the de facto identity standard. The reason it is widely used as identity is revealing: not because it was designed to be so, but because it is so successful and ubiquitous as a communication channel. Similar stories can be told about many other identity standards, such as mobile phone numbers and, to a degree, mailing addresses.
Email is also a good lesson on protocol design. Like many protocols now, it was intended to be permissionless, decentralized, and resilient. As email became more popular, spam and spam detection became a bigger problem. Spam detection relies on identifying patterns unique to spam, which is cheaper if you can train one model and serve many different users. This drives email to centralization, contributing to the current situation where only a handful of email providers dominate the market, and very few people running their own servers. As we will see later, dynamics like this might still be crucial in the development of newer protocols.