Exploring apparent polarities in the digital identity space

[draft 2]

I'm working on a paper on "Exploring apparent polarities in the digital identity space" and I need all of your help! I thought I'd post drafts here on my blog and invite people to comment on the blog or to write on their own blog about it (I'll put a Technorati cosmos link at the end of this post in lieu of Trackback for Blogger). Any suggestions for amendments or additions (lots of room for those so far!) welcomed. Even just links to interesting resources for each section would be great.

[I intend to at least touch on WS-Federation, Liberty Alliance, Sxip, Identity Commons, Shibboleth, FOAF, Ping ID and various client-side solutions.]

UPDATE: I realise that the scope of this paper makes it impractical to deal properly with the characteristics with each of the identity initiatives listed above. Rather, I will focus on the fundamental principals and issues of identity networks for the most part.

Incidentally, the paper was inspired by hearing this identity-focused IT Conversations discussion with the Gillmor Gang — Steve Gillmor, Doc Searles, Jon Udell— and Phil Windley.

Anyhow, here's my evolving draft...





1) Introduction


The world wide web's original promise was of an interconnected machine-facilitated mesh for humans to collaborate within and share information across freely. The emphasis on openness and mutual independence in the development of the web's underlying transport, connection and publishing technologies (TCP/IP, HTTP and HTML respectively) allowed for ready participation and diverse innovation within the resultant digital fabric, leading to an ongoing explosion of creative, social and commercial applications on the web.

The radical openness and flexibility of the internet's key technologies is clearly vital to the ongoing success of the web. Yet at the same time, the absence in the web's fabric of an intrinsic respect for the boundaries people wish to place around their digital identity have lead to a polarization in the evolving web, where castles of structured coherence (web sites) float in a chaotic ocean of hyperlinked information exchange.

The core technologies of the web are designed with communication between machines, not humans, in mind [explanation for non-technical people here]. This is all well and good—the machines need to be able to process the information on the web on our behalf. But lacking in these core technologies, however, is a representation of the boundaries we place around our digital identity: an enshrined respect for what we wish which people to be able to do with which of our digital "posessions".

The fabric itself of the web must embody that principal of respect for human identity for the web to become truly human-like. It's really that simple. I would like to explore in this paper the practical issues of how we might move (and are indeed already moving) towards realising that vision, working to dissolve the apparent polarisations of technological, business, governmental and user priorities in order to evolve an holistically-integrated network that honours the human identity of each of its users.


2) Background


Microsoft tried to unify digital identity services with Passport, which provides users with a single sign-on across member web sites. But people didn't wan't to put all their identity eggs in one Microsoft basket.

So if not Passport's single, centralised database, then what?

Virtual and meta directories (used primarily within corporate intranets) are structured approaches to aggregating data from diverse stores across a network.

Conversely, the emergent Blogosphere facilitates a fluid sharing of ideas and content (via RSS newsfeeds) between bloggers, seemingly functioning in a way analogous to a neural network.

[any suggestions for how to integrate a discussion of Web Services / SOAs here?]

A digital identity network that allows its users to choose the "boundaries" around their digital identity—the relative closed-ness or openness, relative to other users, of stores of data attributable to themselves within the network—has the potential to offer the best of both worlds. Such a network will be neither essentially closed and centrally-regulated like a corporate intranet, nor globally open and predominantly unregulated (aside from the interventions of various national governments) like the blogosphere. It will have a structure that emerges organically and at many granular levels of community from the attributes of the digital identities of the people who use it.

To succeed, however, digital identity networks must address issues of trust, privacy, governance and convergence around shared language(s) for describing information—issues that arise when a network is neither centrally controlled nor radically deregulated, but rather emulates the complex dynamics of physical-world human relationships and communities. These are the challenges that organisations such as WS-Federation, Liberty Alliance, Sxip, Identity Commons, Shibboleth, FOAF, Ping ID and LID are grappling with.

How might we begin to weave these various initiatives into a coherent digital identity meta-network ("ID meta-network" for short)? As one starting point for discussion, I suggest the following axioms for the technical characteristics (as opposed to legal or regulatory aspects) of an ideal ID meta-network as provisional frames of reference for considering various specific aspects of the identity space. For the sake of clarity I have expressed the axioms in the first person:

Axiom 1: I may determine who may do what with which of the data attributable to me (my "digital possessions").

Axiom 2: I may nominate individuals, groups or organisations as proxies for me in negotiating and managing the interactions with my digital possessions by other individuals, groups or organisations.

Axiom 3: I may only interact with the digital possessions of other individuals, groups or organisations according to their wishes with regards to Axioms 1 and 2.


3) Emerging pictures


User vs. enterprise priorities?

Commercial web services have traditionally favoured strategy of data lock-in for users, treating user data as a proprietary asset of the enterprise rather than belonging to the user and merely loaned to the enterprise. Unscrupulous practices such as selling user data to spam mailers have proliferated in the absence of integrated identity networks that allow the reliable tracking of data exchange within the network.

Users have understandably often been reticent to entrust more than a minimum of their data to all but the most reputable commercial web services, aside from the ubiquitous requirement for a valid email address to authenticate their account.

Doc Searle's Data Silo metaphor

An ideal ID meta-network will embrace the status quo of data lock-in while simultaneously incentivising openness and transparency. The basis of competition amongst large organisations will naturally shift away from the effectiveness of their lock-in strategy to the added value they can provide to users who choose to "lend" a subset of their data to them.

Individual vs. role-based identity?

Andre Durand's piece

Building from the individual up

Role-based reprentations of identity are effective as "shorthand" for the complex set of relationships an individual participates in within a given context within the ID meta-network. However, all relationships in the meta-network are also understandable in terms of an each unique relationship an individual has with other individuals, groups or organisations according to Axioms 1, 2 & 3.

Distributed vs. hub-and-spoke authentification and data-exchange models?

role-based authentification
client-side solutions
FOAF
LID
Identity Commons i-brokers
SXIP homesites

A flexible federation model for authentification and data-exchange would allow people to choose, through mutual agreement: (1) one or more authentification points within the meta-network (these points could belong to themselves or to other individuals, groups or organisations) and (2) with which third-parties those authentification points could share which of their data.

Open (source code, APIs & standards) vs. closed (commercial and proprietary)?

Protocols—SAML, gateways etc.

There can be room within an ID meta-network for a whole spectrum of approaches with regards to standards and source code licensing. Users are likely to accept a trade off between the added value a service within the meta-network provides and the ring-fencing around it. However, the mutually-beneficial network effect inherent in opening up source code, APIs and standards should ensure that competition between services is waged progressively higher up the software development "stack": at the user-experience level rather than the level of the network and data-silo "building blocks", which will become increasingly commoditised.

Privacy vs. security?

Resolving the conflicting priorities of privacy and security may prove to be a thorny issue for an ID meta-network. On one hand, national governments are reacting to a perceived threat of "terrorism" by seeking increasingly far-reaching powers to track intelligence on groups who they suspect of planning to act against their interests with violent means; on the other hand, this very extension of government powers is giving rise to concerns that basic human rights such as the freedoms of speech, self-determination and religious worship are being progressively impinged upon. This conflict of priorities will surely apply to an ID meta-network, and would seem unlikely to have a purely technical solution, in that the more robust the privacy controls within the meta-network are, the more potential there is for groups to operate beyond the reach of government intelligence agencies.

But is our ID meta-network really significantly more of a security risk than the existing internet? In practice, according to Marcos Moulitsas Zúniga, insurgents already have ample recourse to technical and practical ruses to avoid discovery of their online communications with one another: "The troubling truth is that terrorists rarely have to be technically savvy to cloak their conversations. Even simple, prearranged code words can do the job when the authorities do not know whose e-mail to monitor or which Web sites to watch".

And on the other side of the coin, an ID meta-network that facilitates the respectful co-existence and interaction of diverse communities across the world could actually help us to build the very bridges across gulfs of belief, language and nationality that will enable us to evolve as a species beyond our destructive habits.


Fluid vs. structured community?

Human beings exist in relationship and community with one another and with the rest of the world. On one hand, this state of relationship and community has a fluid quality: the ever-shifting compositions within and between spheres of our life (our living place, work place, family, romantic and social ties, shopping habits, hobbies and so on) ensure a constant stream of unexpected interactions (serendipitous or otherwise) with people, places and things. On the other hand, at any given time, each of these various dimensions of our life has a particular structure, which may have a stability and coherence that is only upset by a relatively significant event.

We can think of this structural stability of as a kind of social phase-locking. For example, our role and colleagues in the company we work for are typically the same from week to week, only changing if we or they are sacked (or perhaps relocated) or quit. And the threat of chaotic disruption, uncertainty and loss of expertise incentivises sticking with a particular employee (or job), even if they or it may seem far from ideally suitable. Similarly, the living place of our family members may vary (as may the harmoniousness of family relations!), but typically we consider them our family for life—ours or theirs. The ties of blood and shared experience discourage members of all but the most dysfunctional families from completely disowning one another. Moreover, a change in one dimension of our life can trigger a change in another: if an elderly family member dies, we may quit our job to look after their spouse.

This phase-like behaviour of communites, where a tendency towards homeostasis or phase-locking is counterbalanced by a propensity to tip into (often chaotic) phase-shift given a sufficiently significant stimulus (and thence, sooner or later, into a new homeostasis), is nicely analogous with certain mathematical and physical processes: non-linear equations may tend towards a particular solution, termed a "strange attractor", for a given range of inputs, then tip into a phase-shift chaotic pattern of results beyond that range; and chemical substances transform from solid to liquid to gas with increases in temperature in a similarly non-linear way (consider the way ice rapidly starts to melt as its temperature rises above 0ºC, and water in turn suddenly starts turning into steam at 100ºC in a boiling kettle).

The phase-locking of human communities seems to happen at multiple levels of granularity or scale. The military juggernaught that was Japan turned on a penny after total defeat in the Second World War away from a religiously-fuelled and reckless total militarism to become a peaceful manufacturing giant; Protestantism spread like wildfire across Europe after hundreds of years of a progressively corrupt, commercialised and politicised Catholic hegemony; the Descartian reductionist, quasi-objective paradigm continues to hold sway over the world of science decades after research into quantum physics demonstrated the limitations of such a model. It took decisive events in the former two instances—the dropping of the Bomb and the nailing of Luther's Theses on a church door respectively—to trigger phase-shift away from status quos that had clearly been failing the majority of their participants (the non-elite) for a considerable time. I wonder what it will take to shift science into a holistic paradigm?

But stable and circumscribed community structure is perhaps an ethically and morally neutral thing in itself. High degrees of persistence, self-coherence, organisation and focus are surely indispensible attributes of groups of people who wish effect a common aim, and have made possible some of the great cultural and scientific breakthroughs in human history: the sequencing of the human genome, . Furthermore, as the arguable failure of the Communist movement (and counterculture experiments in "free" community living, which I experienced for myself as a seven year-old) to provide a model for radical egalitarianism would seem to demonstrate, human beings have a proclivity to impose control on their own and others' freedom, however lofty the ideals they set out with.

A global soup of unrooted individuals


Awkward diversity v. interoperable uniformity?

Email, discussion boards, instant messaging (IM), and more recently blogs, wikis and a plethora of related social web applications have facilitated the evolution of a social dimension in the web. These applications have allowed individuals and groups to communicate with others and to publish and share their writing, still images, music and video.

However, one of the hallmarks of the existing social web is the predominantly ad hoc and often fragmented nature of the community structures it facilitates. Each social application genre (blog tool, wiki etc.)—and to a lesser extent each application within a given genre—imposes its own hard-wired ontology onto its users through the data structures it employs and the metaphors of its user interface. It follows that the formation of relationships and community groupings within each application must take place within the limitations of that application's ontology. And as each application type's ontology tends to differ significantly from the others, and so the data-sharing interoperability between social applications tends to be limited, people have to create anew the online dimension of their relationships and communities each time they sign up to use a new application.

Perhaps the strongest counter examples to an ad hoc and fragmentary tendency in social applications' ontologies are those of RSS syndication and open APIs. RSS (and, to a lesser extent, its sibling syndication format Atom), has become a de facto standard for the open syndication of web-based content [explanation here], providing a "mould" with which blocks of data may be built, then shared and combined across diverse applications, much like Lego bricks may be built into an infinite number of models. Open APIs likewise allow social applications to selectively open up their data store, via information-processing services, to third-party developers, who can then build added-value applications on top of these "web services". The ecosystems of applications constellated around Flickr and del.icio.us are examples of the value of open APIs in extending the reach and functionality of social applications.

RSS syndication and open APIs are facilitating the evolution of the web into a truly service-based network, and as such have great value. Yet it is clear that neither RSS nor the current crop of open APIs can provide the basis of a digital identity meta-network. RSS data is public data, accessible by anyone via a URL like an HTML web page. RSS therefore cannot support identity-based access filtering. Applications deploying open APIs, on the other hand, typically filter access to their users' data through their API(s) by requiring the submission of a user's login and password. In order to have application A automatically share my data with application B, I must entrust my login and password from the application A with the application B—a clear security headache.





Online vs. offline?

Relationship vs. branding?

Hughtrain etc.

IP anarchy vs. big business?

Creative Commons
Falling cost of copying is inversely proportional to rising strength of IP law

Naming conventions

i-names
extended DNS

Distributed vs. centralised governance?

Chaordic model
Role for national government?

Identity web as platform

open APIs
situated software
i-brokers: identity-based information processing services?

Resolving the Identity Catch 22

initiative from enterprise?
initiative from users?
incentivising self-centred participation for mutual benefit

Identity of objects

Spimes

The Digital Divide

Global Voices