We need to talk about the social graph

We need to talk about the social graph

By Philip Sheldrake


March 22, 2022

Who doesn’t love a good concept?! Concepts are the fundamental building blocks of thinking, of designing. While there are plenty of things in the mix when it comes to contemplating system design, if the primary concepts remain unchallenged and unchanged from what came before, then the outcome will likely look very familiar. If you want system change, start with changing the paradigm — the system of concepts and patterns that form the worldview.

By way of a quick example, if the economy of your new system picks up on the concepts collectively known as capitalism — e.g. private ownership, capital accumulation, scarcity — then perhaps it should not be a huge surprise when your new system turns out to be capitalist too. New code. Same concepts. Familiar outcome.

Technological decentralization isn’t magic dust. Merely decentralizing the technical structures or components manifesting a concept is no guarantee of different outcomes; technological decentralizing doesn’t even guarantee decentralization.

So with that said, this post is about a concept known as “the social graph”.

I put the phrase in quotation marks to convey that it’s considered a BIG thing. To help you get in the right sort of mindset for this post, I invite you to say the phrase again in your head — or even better out loud — in dramatic Hollywood voice-over style.

🎥 🎬 I’ll wait 😆

... So let’s start this movie at the beginning.

⚠️ Spoiler alert: there are two cuts with quite different endings and so then quite different implications for the part you might play.

A very brief history of the social graph

The measurement and study of networks of relationships among members of a social group took off after the First World War culminating in Jacob L. Moreno’s groundbreaking 1934 book on sociometry, Who Shall Survive? (see Freeman 1996).

While studying truancy at the New York Training School for Girls, Moreno developed the sociogram, the now famous representation of people and their relationships as nodes and edges. Commonly referred to as the social graph these days, here’s an example graph from his book and its more modern presentation (Martin Grandjean, BY-SA 4.0).

Evolution Of Groups Who Shall Survive page 36 .png


What was Moreno’s goal? Well, here’s something that might interest you from the foreword to his book (courtesy of the Internet Archive):

foreword to Who Shall Survive?.png

While we might appreciate nearly nine decades later that interpersonal relationships might not be quite so compliant, nor the pursuit of “efficiency” quite so mechanistic, that’s not the point. We find out here that social graphs were conceived not merely to understand groups but in fact to help control them. As Moreno himself noted:

“... this superimposition of a mechanical-social network upon a psycho-social network produces a situation that takes society unawares and removes it more and more beyond human control.”

Fast forward ⏩ to Facebook’s popularization of the social graph from 2007 integrating this property into its business model, and it’s no wonder many of us wish to digitalize things differently. But let’s not make this entirely about Facebook. Let’s take a look at some decentralizing initiatives and their founding concepts in chronological order, and then we can talk about those concepts.

In solidarity

I’m going to take a short visit to a chapter in Google’s history and then onto some dweb (decentralized web) projects. As I’m going to be critical, I want to be clear about something up front.

We want to shout out RESPECT ✊🏼 to anyone innovating in the dweb context with the aim of making the world a better place for everyone. For example, dsocial (decentralized social) should be a means to some most welcome ends, not least in prompting Emperor Zuckerberg to renounce his throne and engage with society as one of us rather than one above us. Nevertheless, if dsocial was easy, the likes of Facebook et al would already be little more than Wikipedia entries for historical reference. But here we are.

Dsocial is complex and it’s going to need a heap of innovation and cooperation, so the more ideas and the more experiments and the more conversation, the faster we can all get to where we want to go 😍. Criticism is offered only with that shared goal and shared effort in mind.

Google social graph API, 2007

Google aspired to produce a sort-of-but-not-really-decentralized social graph with its Social Graph API project, 2007 - 2012. It was sort of decentralized because it relied on everyone ‘out there’ publishing structured data about themselves and their relationships on their own websites and blogging profiles that the company could then ingest to spit out an API anyone might use to quiz the graph. Here’s a brief presentation (PDF) from one of the project leads at the time.

The assembly of the graph and the provision of the API are centralized, but then so are the likes of Infura and Alchemy and aspects of the AKASHA framework currently.

Clearly, the project inherited one concept without question, the concept of the social graph. One assumes Google was intent on profiting with its Social Graph API or more likely, in line with its typical modus operandi, because of it, per Facebook’s social graph API today. We will never know.

The project was shelved when adoption fell short of expectations. As Facebook was IPOing and achieving a userbase of one billion monthly active users in 2012, I’m guessing only a teeny tiny fraction of that userbase also published the requisite structured data some place, and of those that did, it was likely set-and-forget.

AKASHA, 2015

The AKASHA Foundation has pioneered dsocial since 2015. Our early experimental products published a social graph to the blockchain — as the saying goes, we’ve been there done that — but our latest incarnation does not.

We are learning by doing. Openly. And we aren’t taking any conceptual ideas from earlier digital innovations and initiatives for granted for fear, as I indicated above, of merely repeating history. This might mean we keep some facets centralized as a public good for now until we can crystallize new conceptual models and put them to the test with community in community. That process is well underway.

Project Liberty, 2021

Are you ready to create a web that works better for all?

The homepage continues:

Project Liberty includes the development of a new infrastructure to serve as a foundational layer for the next generation of the web. At the core is a groundbreaking open source protocol called the Decentralized Social Networking Protocol (DSNP) ... The key to change lies in the power of combining our social networks in one unified social graph that is free from corporate control.

The proposition could not be clearer: replace privatized social networking platforms with “an open architecture” featuring “one unified social graph” that is owned by everyone as a public good. Rather than having to “surrender ownership and control” over our data, the Project Liberty team believes that “giving people ownership and control of their own data is the first step to break the cycle and shift the balance of power.” In short, the project aims “to reclaim the social graph”.

The project inherits three dominant concepts without question:

  • the concept of the social graph
  • the concept of treating data as property
  • the concept of the bureaucratization of human identity.

I’ll make one last observation here if only because it made me smile and made me look forward all the more to cooperating where we can. Its progenitors present four elements they believe distinguish Project Liberty from anything that has come before:

  • The protocol is built to be a public good
  • The proposition is not reliant upon traditional VC business models
  • It places people at the center of the protocol’s design
  • It’s a bold vision focused on achievable innovation.

Which all sound very familiar to me and my AKASHA Foundation colleagues! 😁

DeSo (previously often referred to as BitClout), 2021

The DeSo initiative presents itself as “the decentralized social blockchain”, and describes social media as “one of the biggest markets for blockchain”.

It’s not entirely clear to me why social media should be described so prominently in terms of it being a market other than to recognize the imperative to deliver ROI for early investors — the homepage lists no fewer than fourteen VC firms. It could be argued that the primacy of commercial motivations in the context of social networking has already been shown to be ... well, shall we say ... sub-optimal.

Nevertheless, putting this urgent need for DeSo to one side, what can we learn about the initiative’s conceptual grounding?

The DeSo initiative’s documentation (especially here) and a fleeting reference to the social graph in a blog post appears to confirm that the project inherits two dominant concepts:

  • the concept of the social graph
  • the concept of the bureaucratization of human identity.

Lens, 2022

The Lens Protocol is a permissionless, composable, and decentralized social graph that makes building a Web3 social platform easy.

Easy? Bless those marketing communicators! “Social” includes human beings, so never easy.

Lens renders a singular social graph publicly on a blockchain, and the introductory overview talks in terms of people owning their data and owning the links between them with shared ownership of the resultant monolith. So just for clarity:

Users will own everything.

The project inherits three dominant concepts without question:

  • the concept of the social graph
  • the concept of treating data as property
  • the concept of the bureaucratization of human identity.

Conceive to conceive

The word concept has its roots in the Latin for conceiving, i.e. creating, forming, devising, generating. To rephrase my opening point here, we are very much less likely to get anywhere new if we don’t create or generate anew.

There’s a critical meta implication here too given our context.

What if the ongoing facilities for all social networking participants and communities to conceive wonderful new generative futures pivots in some way on the fundamental conceptual choices we make right here right now? Are we not, in this moment, conceiving a social system for conceiving? If a dsocial system falls short of being generative in this way, is it degenerative?


Project Liberty, DeSo, and Lens inherit three concepts to varying degrees. Social graphing is the main focus here, so let’s address the other two very briefly.

The propertization of data is a category error. It is not property. Moreover, to try to treat it as such is unethical. I couldn’t put it better than the European Data Protection Supervisor who likened the possible emergence of markets for personal data with markets for live human organs. To find out more, I can point you to a series of three blog posts on the topic: The misleading name, metaphor defiance, and awesome potential of "personal data”.

The dominant computer science conception of identity is inherited entirely from the bureaucratization of identity, of which legal identity is a well known example. Just to be clear, this is not good news when it’s extended into every aspect of life. Its dominance is not surprising however given that bureaucracy, in the form of governments and large companies, was the IT industry’s first customer. But there is a chasmic difference between the bureaucratic approach to identity and how human identity works as a sense-making capacity in human community. For more on this topic, check out the generative identity website.

Of course, it is not really possible to separate interpersonal data and identity from social graphing ...

Your graph or mine?

Human identity might be described as a dynamic, contextual, multiple and subjective social location of persons in relationships with others for the purposes of sense-making and cooperation. It is about being and becoming, together. On the other hand, bureaucracy must render each and every citizen legible to the state, to the system. The bureaucratic machinery applies a singular, rigid, enduring and non-contextual identifier. You are in fact just a number as far as the state is concerned, and the accelerating process of digitalization is advancing the bureaucratic concept adeptly everywhere you look. 👆🏼👇🏼

Facebook necessarily renders its users legible to its system. The Facebook social graph is not yours nor mine but the Facebook system’s. The graph is singular and reifies the informational contexts as required by the system’s encoded goals rather than your needs or preferences or values, or mine. This stands whether the social graph is operated by a for-profit company, a non-profit company, or as a public good (or is that a public-not-so-good?) For the sociologists out there, the social graph is deep and wide and stubborn structure.

The social graph’s role is transformed with use and with growth. It begins as nothing more than a miserably (or perhaps beautifully?!) imperfect description of a rich and diverse human network but, in the total absence of alternative digitally-accessible graphs or stories, it is soon confused for “the truth”. The system wields its graph to determine who ‘knows’ who, who is interacting with whom, and to determine what’s important, what’s trending, according to its inevitably imperfect guesses at your contexts. We are nothing more to the system or then to each other (”social” anyone?) than a weak signal. We no longer constitute a “we” — adoption of the singular social graph concept sets the system up to atomize us and demarcate us for the purposes and assumptions encoded in its architectures and algorithms. Just for the avoidance of doubt, this purpose cannot constitute meaningful human connection by definition. Graph connection trumps that.

It is frustrating to have no direct agency over the social graph in the instance of Facebook but at least the distant and opaque megacorp in charge must comply with regulations our democratic representatives develop on our behalf (with some jurisdictions more on the front foot than others), with concerns for privacy being foremost of course. Only the Facebook system gets to ‘see’ the whole graph so your privacy could be respected in principle but for the accompanying business model requiring the company to farm and sell your attention.

Attempts to write a monolithic social graph to a distributed ledger introduce a different technology ‘on the back end’, but the social graph concept remains unchallenged and unchanged. The mechanics of the protocol necessarily renders you legible once again to the system. When this is also in plain sight (in the absence of prohibiting regulation), the individual’s response will be the same response as always when caught in a panopticon — self-censorship. As and when commercial motivations are embedded as deeply as for Facebook, it will very likely and very quickly start to look all the more remarkably like Facebook.

A modern-day Pink Floyd (ref) might sing ...

🎼 It’s no laugh — you’re just another node in the graph.

Rather than one graph assimilating billions of people, imagine now what we might actually want from the digital components of our cyborg selves to help us achieve our goals and better live up to our values in combinations with others; the very definition of autonomy. And perhaps, given that our multiple selves in this respect are reciprocally defined by and co-evolving in relationships with others’ selves, the idea of a billion small graphs is a more real, more natural, more decentralized and innately more human pattern more likely to produce outcomes that may be described in terms such as dignity, agency, autonomy, intelligence, flourishing, and well-being. The ROI will be immense, just of different and more valuable kinds.

Does this feel like the makings of an edgy concept?

Playing your part

Which combination of concepts you chose to pursue or support will come down to the part you’d like to play and the parts you’d love to see your friends and family play, and so then the ‘edit’ you prefer.

In one edit of the movie, the system, monolithic and stubborn beyond the scale of any meaningful or contextual governance, collated the weak signals relating to the thicket of global super-wicked problems and reflected them back at you and your friends and family, but lacking autonomy and the freedoms to self-govern and self-organize at human scale, the changes came faster than community could respond. It’s a tragedy. Game over. May the next dominant species properly grasp the lessons Nature has to teach.

In the other edit, the end credits begin to roll as plural, autonomous, cyborgian communities determine their own micro-structures and delight in the generativity that allowed them to rise up together to the movie’s thematic storyline with an unexpected freedom of mind and resilient collective intelligence. Participating as nature, these transhumans live to tell their stories another day.

Make it happen

If you are as excited as we are for the possibilities here, join us in conversation and in collaboration. Everything we do is open source and offered up as a public good. Take your pick from one or all of these:

— Join Ethereum World, a first alpha instance of social networking built with AKASHA

— Join the AKASHA Conversations

— Join our open design sprint: designing for moderating decentralized social networks

Keep in the loop.

Subscribe to our newsletter.