Verifying identity as a social intersection

Co-author of Radical Markets Glen Weyl invited me to review Verifying Identity as a Social Intersection, co-authored with his colleague at Microsoft, Dr. Nicole Immorlica, and Stanford University's Professor Matthew Jackson.

The topic is so-called digital identity, a term that could be mistaken for how personal and group identity is manifest online, but actually relates to how we might employ digital technologies to transform society's accommodations of and approaches to identity.

It is not a challenge that anyone might describe as readily "solvable," as my recent webinar for SSImeetup makes plain. If anything, it is a fine exemplar for H.L. Mencken's witticism:

For every complex problem there is an answer that is clear, simple, and wrong.

This paper is an important contribution towards navigating this complexity appropriately. I do however identify a major problem — and therefore opportunity — relating to the conceptualizations of identity the authors have made their object. It appears they are intent on engineering for the trickiest yet perhaps the most societally-important conceptualization, but then present it as a solution to a more mundane conceptualization, and one that desperately needs the balance of the former to mitigate its innate harmful potential.

I finish with a brief explanation of the AKASHA Foundation's work here. As you can imagine, this is core to our purpose.

About "identity"

Brubaker and Cooper (2000) argue that "identity" is in crisis, a devaluation of its meaning caused by a proliferation of varying conceptualizations and applications. "The term is richly — indeed for an analytical concept, hopelessly — ambiguous."

Amongst its uses they note:

to highlight non-instrumental modes of action (i.e. particularized — e.g. characterized by religious or ethnic identity)
to focus on self-understanding rather than self-interest
to designate sameness across persons or sameness over time
to capture allegedly core, foundational aspects of selfhood (relevant for example to a government seeking to distinguish one individual from another)
to deny that such core, foundational aspects exist
to highlight the processual, interactive development of solidarity and collective self-understanding
to stress the fragmented quality of the contemporary experience of the "self", a self unstably patched together through shards of discourse and contingently "activated" in differing contexts.

Stable and unstable. Concrete and malleable. Like and unlike. Within and without. Singular and multiple. Individual and collective. Separate and connected. Whole and partial. Enduring and varying. In being all things, "identity" apparently fails to distinguish itself.

I interpret the authors' invoking "the modern state" in the opening line as qualifying their working conceptualization of identity as one setting out to satisfy a relying party's need to recognize that this is Alice again, and by corollary assisting Alice in identifying herself as Alice again. My reason for explicating the other conceptualizations above is that they are not clearly delineated — as a bullet list might otherwise convey — but interwoven. Designs for one have emergent consequence for others.

The authors call on the properties of Georg Simmel's conception of social circles.

To situate Simmel's thesis in the most efficient way, albeit somewhat simplistically, you might like to imagine what identity was like before the modern state set out to bureaucratize it. Who you were was defined in terms of who you knew, what you did, and where you lived and worked. Given the corresponding lack of mobility back in the day, it was so unlikely that anyone else shared these properties that one hardly need bother with family names.

Adam's daughter Alice. Bob the blacksmith. Claire over in Clarence Court.

The authors' central claim relates to the core features of Simmel's work:

if such features can be formalized and accessed efficiently, they are a powerful foundation for many identity applications.

They support this assertion by eliminating other options: centralized, "big data" platforms, and self-sovereign identity.

They aim to supplant centralized identity systems, finding these:

highly insecure
thin — reducing an individual to a small dimensional (binary) object; and
artificial — bearing little resemblance to the 'real' person.

The authors note the attempts by the "big data" platforms (naming Facebook and Google) to address thinness, but this has been co-emergent and intimately linked with surveillance capitalism, or "panoptic identity" as the paper puts it.

As for self-sovereign identity (SSI), the authors acknowledge that current definitions are vague, but in general find the prospect to be thin, insecure, and artificial. I would go further, if only to explore a reconciliation of SSI with the paper's development of Simmel's social circles.

Adopting a conceptualization of identity as a social construct — and what good can identity serve should you find yourself all alone on that fabled desert island? — how can it fundamentally emanate from oneself in the way "self-sovereign" intimates? The only basis for anything that might be called self-sovereign in this context is a self that reliably condenses its defining relationships informationally and some how reflects that out to the world personally, and verifiably to some degree. If it didn't sound so oxymoronic, would that be socially-self-sovereign?

Web of Trust

The authors identify the "Web of Trust" as the paradigm most closely related to their own. So do we. You can find AKASHA's report from the Rebooting the Web of Trust meetup in Barcelona this March here. From that post:

If you're not familiar with the phrase 'web of trust', it was coined in the early 90s by Phil Zimmerman, creator of Pretty Good Privacy (PGP) encryption software.

While it originally related to a specific technical execution, the term is invoked more generally today to describe the prospect of a private and secure peer-to-peer / person-to-person / agent-to-agent network, i.e. one that does not require a centralized intermediating service. And needless to say, the aspiration is to make it scalable and a beautiful user experience to drive mass adoption, requirements that haven't yet been achieved — hence the need for a rebooting.

Trust, just like identity, is a tricky concept to pin down, but I don't think we need to explore that here. Just to say, trust is not a static thing. It is established and grows and shrinks and dies and rekindles over time. More a verb in nature than a noun. And if trust is dynamic, so then is a web of trust.

Verb-like

With such dynamism in mind, we consider various conceptualizations of identity in terms of being more verb-like or just plain noun-like. Birth certificates, driving licenses, and passports for example are noun-like — they're not intended to be dynamic, by design. They are artificial, as Phil Windley notes so achingly well:

Descartes didn't say "I have a birth certificate, therefore, I am."

There are many scholars working on the topic of our multiple and fluid identities, linking them to many natural phenomena in our societies, not least one's psychological and possibly physical safety. For brevity, I will reference just one typification.

After Foucault, Stuart Hall conceptualized late-modern identity negatively as not signalling the "stable core of the self, unfolding from beginning to end through all the vicissitudes of history without change." (PDF) As a founding figure of British Cultural Studies, Hall addresses the implications for group-level identity. He argues against the idea that one might find one's identity hiding amongst others coalescing around a common history or ancestry. The individual and so the group, the group and so the individual, cannot attain a unifying identity, rather identities are "fragmented and fractured; never singular but multiply constructed across different, often intersecting and antagonistic, discourses, practices and positions."

In a recent article for the Omidyar Network's Good ID project, I put it less elegantly.

What do I share for example, socially speaking, with the little chap tottering off to his first day at school in the 1970s? Indeed, I feel different from just a few years ago, and fully expect to feel different again in the future. In fact, the way I conceive my own identity can vary from one context to another in the very same day. That's living. Being. Becoming.

As noted in the SSI meetup webinar I referenced earlier, the art here will be ensuring that such essential human qualities are not obliterated by systems for noun-like digital identity as such systems erase the frictions that have kept noun-like concepts in check to date.

Mistaken identity

Having attended diligently to the matter of assembling social identity from the qualities of redundancy, sociality, and intersectionality inherent to Simmel's conceptualization, the authors suddenly remind the reader of their opening reference to the modern state by asserting (5.6.1):

In an identity system, it is important that each individual corresponds to a single identity.

We might read that as "in a particular noun-like identity system ..."

On the other hand, Simmel himself conceives of identity (or "individualities") as verb-like.

We are all fragments, not only of humanity in general but also of ourselves. We are amalgamations not only of the human type in general, not only of types of good and evil and the like, but we are also amalgamations of our own individuality and uniqueness — no longer distinguishable in principle — which envelops our visible reality as if drawn with ideal lines. However, the view of the other broadens these fragments into what we never actually are purely and wholly. The fragments that are actually there can scarcely not be seen only juxtaposed, but as we fill in the blind spot in our field of vision, completely unconsciously of course, we construct the fullness of individuality from these fragments. (ref.)

... the procedures play out in the individual soul as well. In every moment these processes are of so complex a kind, harboring such an abundance of manifold and contradictory vicissitudes, that identifying them with one of our psychological concepts is always incomplete and actually falsifying: even the life moments of the individual soul are never connected by just one thread. (ref.)

Having journeyed through their design thinking inspired by Simmel's natural and essentially human articulation of identities, verb-like and beautiful in their complexity and existentially vital, the authors switch back to a noun-like conceptualization. This involves a certain irony, as I will explain.

The disadvantage of noun-like sociotechnical identity systems, beyond the traversal of national borders and the payment of taxes and suchlike, is that they are so utterly rigid as to make their broader, frictionless application a danger to society.

The disadvantage of verb-like sociotechnical identity systems is they require such a sophistication of design and lengthy explanation to affected parties that many technicians 'working on identity' find it difficult to appreciate such conceptualizations exist let alone reflect on ways they might be realised.

This paper appears to combine these disadvantages. Attacking the verb-like challenge is to be welcomed, of course! I'm arguing that it shouldn't then be trained on a noun-like instantiation.

Interpersonal data

At AKASHA, we consider verb-like identity as co-emergent with interpersonal data, a term we employ deliberately to distinguish it from the term personal data as defined by the General Data Protection Regulation (GDPR).

This isn't the place to expand on interpersonal data — you can find out more in the SSImeetup webinar and in this post on the RadicalxChange Medium publication.

By design, interpersonal data is privacy preserving and socially meaningful, and has the innate qualities — at least when elevated to information — of redundancy, sociality, and intersectionality.

At the conclusion of the Twitter thread with Glen inviting my response here, he hints at a bigger vision, of which this recent paper constitutes one part, that "would replace money with a nonlinear mixture of identity and trust." This is exciting and we hope AKASHA might collaborate. Our RadicalxChange post draws parallels between interpersonal data, money, and open money, concluding:

Markets don’t transform personal data so much as interpersonal data transform markets.

Verb-like identities, collective intelligence, and new forms of currency. Bring it on!

Reference

Immorlica, Nicole, Matthew O. Jackson, and E. Glen Weyl. “Verifying Identity as a Social Intersection.” SSRN Scholarly Paper. Rochester, NY: Social Science Research Network, April 20, 2019. https://papers.ssrn.com/abstract=3375436.

Image credit: Akio Takemoto, CC BY-SA 2.0