This is the final part in a three part series by RadicalxChange and Wireline on decentralized social identity, focusing on its transformative potential. Part One laid out the requirements, while Part Two discussed its applicability.
Beyond near-term “practical” uses, functioning social identity systems have truly transformative potential to transform the organization of technology, society, politics and the economy. We now consider a few, increasingly ambitious examples.
First, consider blockchains, a popular distributed database technology. Such systems offer rewards to those who store and ensure the reliability of the database. Most systems for administering blockchains rely on “proof of work” or “proof of stake” to establish that those claiming such right and rewards are not fraudulent. The first relies on demonstrating that one is actually performing the computational work required to store the system, while the second relies on proving that one holds a certain amount of financial value within the system. Both are highly plutocratic, as they literally reward those who already have access to the financial resources needed to purchase computers to do the work or own the relevant resources. One could imagine, however, alternative systems based on “proof of identity”, decisions about database validity and rewards for participating in them are run on a one-person-one-vote or other identity-based system. Humanistic identity systems would enable such protocols, making more democratic governance of blockchains feasible.
Yet, the improvements a social identity system could bring to social data structures go far beyond changes to the governance of blockchains. In fact, most directly, an effective social identity solution could itself offer an alternative data structure. Central to the conception of blockchains is a fully public ledger, visible to all, with wallets controlled completely privately by a single individual. Every data is thus either fully public or completely private. As noted above, this is not a natural or logical data structure for many applications, as most data is and should be partially public, shared in a circle of collaboration but not beyond that. A social identity system could realize a much more accurate model of textures of data management IRL than blockchains can by directly storing and representing the social relationships that underlie the storage and validation of data.
Such a higher-fidelity representations of social data could transform our political economy and society at its foundations. First consider democracy. Current high-stakes democratic institutions are administered almost exclusively by large-scale centralized institutions such as nation states. Attempts to incorporate democratic governance based on humanistic principles, such as one-person-one-vote, even in relatively casual ways, online or cross borders have run into severe problems, with bot networks and other manipulations corrupting the signal from systems like Facebook like buttons or citizen polling by parties like the Pirate Party and the Partido de la Red. The Democracy Earth foundation was created to help overcome these challenges, but the greatest technical sticking point in implementing their solutions continues to be a flexible, humanistic data structure that can accommodate credible, high-stakes democratic voting in a range of contexts independent of fixed institutions.
It is hard to overstate the potential impact of enabling such a broadening of democratic institutions. It could allow more democratic governance of international corporations, realizing fundamental goals of major global political movements. It could enable the creation of new democratic entities representing, for example, environment features or open source software communities. In the long term, it could dissolve the hold of nation states on the public imagination as the only legitimate governance institution.
Beyond democracy, social identity systems could reshape the data economy. The centralized private control and profit-oriented exploitation of the great bulk of data linked personal identifiers is among the leading social concerns of our time. Yet the alternatives to the status quo with the longest history and most public prominence are beset with paradoxes. These broadly fall into two extreme categories of individualist approaches and collectivist ones.
Individualist approaches focus on the notion of “personal data” or “personally identifiable information” and aim to maximize individual control or sovereignty over these data. Precisely what this means can vary wildly, from an emphasis on privacy and an inalienable right of revocability of data (as emphasized in the European General Data Protection Regulations) to a more property-like vision popular in the blockchain community where sovereignty is interpreted as an absolute right to irrevocable sale. Much of the debate we hear is between these perspectives; with privacy advocates warning of a spiraling commoditization of personal space from “data as property” and property advocates arguing that the privacy-first perspective would destroy the potential for data-based artificial intelligence systems.
Yet these two apparently conflicting perspectives share many of the same paradoxes. What is “personal data”, and personal to whom, given that essentially every datum generated by online interactions is shared between interacting agents? Who has the right to sell or delete this? And of what economic or privacy value is revoking individual data anyway? Given that AI superpower platforms require huge aggregations of data to produce compelling new products, any such platform is likely to have so much power of an individual data contributor that a property right in data exercised unilaterally is likely to offer little bargaining power; and given that most patterns can be mostly gleaned without any individual datum, it is not clear that individual revocation can even much enhance privacy.
The natural alternative perspective, the collectivist one, imagines some historical quasi-democratically-governed entity (such as a nation state) appropriating to it the sovereignty over data currently claimed by the large platforms and then collectively and “democratically” determining appropriate use, economics and privacy. Yet these proposals are always vague on precisely what data whom should assert rights over. The whole concept of the internet was to cross traditional borders, and relevant datasets do that with ease; furthermore, given the volume of international interactions, who has sovereignty over what? And is it democratic for the citizens of one polity that happens to be large, like the US, California or Europe to determine the treatment of data around the world? Given the increasingly nationalistic behavior of many countries, such a scheme does not seem much better in terms of the exploitation it will allow than current surveillance capitalism, and many would argue that the Chinese experience suggests it could be significantly worse.
A workable social identity solution seems a natural substrate for an alternative to these two problematic and paradoxical poles that one of us has been advocating for and calls “Data Dignity” or “Data as Labor”. The idea is data sovereignty should lie with democratically accountable communities of interacting individuals whose data are naturally related for productive purposes or by the social relationships that define the sharing of their data.
A social data and identity structure seems like the natural substrate for allowing people to form “Mediators of Individual Data” (MIDs) for several reasons. First, the social sharing of data that it imbeds is one of the central motivations for MIDs in the first place; formalizing it offers a natural basis for finding groups who should join a MID together. Second, by providing a natural basis for internal democracy of a MID in a flexible way. Third, a primary role of MIDs is in guaranteeing the quality and validity of the data it governs and potentially sells collectively. The social validity and verification goals of a social identity are a natural basis for ensuring such quality. Finally, in any social data system, software and services will have to be supplied to the computers on which the system operates, potentially providing a backdoor in which centralization (of that software supply) can reintroduce many of the weaknesses of centralized solutions. MIDs offer a natural provider of such software more consistent with the decentralized architecture that both Data Dignity and social identity seek.
Finally, and most ambitiously, social identity offers the potential of enabling a new political economic infrastructure that combines and moves beyond the democratic and capitalistic applications above. Recent research in political economy suggests new organizational structures that combine the benefits of democratic governance typically associated with nation states with the dynamic, entrepreneurial spirit typically associated with capitalism. In particular, the “Quadratic Finance” or “Liberal Radicalism” mechanism proposes that governments, philanthropists or other public-interested institutions provide matching funds that are directed by individual contributors to form public good providing institutions.
The precise formula is too mathematical to describe in this blog post, though see here for a nice exposition. But some basic features are crucial: namely, small contributions are match more than large ones and contributions to popular causes are matched more than ones to causes with only a few donors. As such, it has a “democratic” character, though unlike standard democracies it is flexible, driven by emergent demands and incorporates financial value rather than being based on a principle of one-person-one-vote.
Given these properties, the necessity for a reliable identity system should be clear. Absent the ability to clearly distinguish two contributors each giving a dollar from one contributor giving two dollars, the above approach is clearly infeasible/can easily be attacked and undermined. In fact, while there have been several recent implementations of Quadratic Finance, with some perceived if difficult-to-measure success, they have run into problems with participants creating fake accounts to exploit this feature.
Yet this challenge clearly runs deeper than identifying physically separate human beings. It seems quite clear that members of the same family should not be treated quite as “separately” as should people who have never met. The former are likely, separate bodies or not, to have deeply related interests and to “collude” in the sense that they already internalize much of one another’s interests. With decreasing intensity, the same logic might apply to long-time friends, neighbors, co-workers, co-religionists, co-nationals, etc. In short, given that extra-mechanism cooperation can lead to the exploitation of mechanisms to encourage cooperation based upon the assumption that participants are selfish, the more cooperative individuals are likely to be inclined to be with each other, the less they should be treated as separate selfish agents from the perspective of the mechanism. Conversely, very “diverse” participants who are unlikely to have much sympathy for one another should be treated as very separate and thus highly subsidized.
Of course, to have any hope of implementing such a plan, one would require quite rich information about social relations and social structure, embedded in a reasonably incentive-compatible way. While ambitious, there is reason to hope that a social identity system might precisely provide this kind of information. If so, I could provide the basis for a fundamental re-imagining of political economy.
Since the advent of trade and monetary exchange, there has always been a powerful line of social critique of money: namely that it abstracts from important social relations and reduces everything to a single, linear, fungible metric. While economists have long struggled to make sense of this critique, the logic above offers a relatively clear formulation.
For public goods or other increasing returns contexts, which arguably constitute most economic value in a modern society, efficient governance mechanisms rely heavily on identity, collusion resistance and funding that is non-linear in contributions. Yet to work properly, such systems need to rely heavily on social information. Thus, most funding should not take place through standard, linear pricing mechanisms, but instead through a rich, non-linear mixture of value and social identity. Thus, standard global currency, and cryptocurrency systems based on ideas related to money, are not the right bases for running an efficient modern political economy. Social identity thus opens the door to re-imagining the foundations of political economy.