Mobile: +61 (0) 414 488 851
Email: swilson@lockstep.com.au

Remember that Digital Identity is a metaphor

The seminal Laws of Identity define a Digital Identity as a set of claims made by one digital subject about itself or another digital subject.

Crucially, this definition leaves "identity" as a metaphor. It's quite different from the way we casually use the word "identity" day-to-day as if it were a thing, like a label.

There is a presumption online (largely un-examined) that identity can be lifted from one context and freely applied in others. So despite the careful framing of the Laws of Identity - that digital identity is about sets of claims and critically context-dependent - many people still carry around a utopian idea of a singular digital identity. The archetypal online identity metaphor is the passport. The belief in the possibility of a universal digital passport has been a long standing distraction, and terribly unhelpful, because there is actually no such thing, not in the sense the word is used by technologists!

Ever since the early days of Big PKI, there has been the beguiling idea of an all purpose credential that will let its bearer into any and all online services, enabling total strangers to "trust" one another online. Later Microsoft of course even named an early digital identity service "Passport", and the word is still commonplace in discussing authentication products. The idea is that the passport allows you to go wherever you like, yet the concept that the metaphor alludes to doesn't exist.

A real world passport simply does not let the holder into any country. To begin with, a passport is not always sufficient; you often need a visa. Then, you can't stay as long as you like in a foreign place; some countries won't let you in at all if you carry the passport of an unfriendly nation. You also need to complete a landing card and customs declarations specific to your particular journey. And finally, when you've got to the end of the arrivals queue, you are still at the mercy of an immigration officer who usually has the discretion to turn you away based on any other evidence they may have to hand. As with business transactions, there is much more to border control than identity. So if we could create the universal digital identity, we would do well to call it something other than "passport"!

Metaphors are more than wordplay; they are used to teach, and once learned, simplistic mental models like “electronic passport” can be deeply unhelpful. The dream of all-purpose digital certificates derailed PKI. When they tried to implement "digital passports", they turned out to be unwieldy, riddled with fine print and excessive identity proofing, and very rarely could such certificates be used anywhere on their own. So the passport metaphor is lousy. Yet with "open" federated identity frameworks, we're unwittingly repeating many of the missteps of early PKI, largely because people aren't coming to grips with complexities obscured by faulty metaphors.

The well-initiated appreciate that the Laws of Identity and earnest schemes like NSTIC all involve a plurality of identities tuned to different contexts. Many federated identity supporters expressly deprecate a single all-purpose cyber identity. Yet NSTIC especially is easily confused by many with a single new ID; a crazy number of press reports represent it as an Internet "driver licence". The misunderstanding is actually exacerbated by the strategy's own champions when they use terms like “interoperable identity” without enough qualification, and casually suggest that a student in future will log in to their bank using their student card.

The Laws of Identity teach that identities are context dependent. That is, you cannot expect that an ID issued in one context will operate seamlessly in another. If we recall the formal definition of digital identity and set aside the passport metaphor, it's actually obvious that identities don't easily interoperate.

Consider the set of claims made about me in the context of my employment; my corporate digital identity might comprise my employee number, position and department, contact details, and company role, which together amount to my employer's imprimatur to represent the organisation. On the other hand, I were enrolled at university, my student identity might consist of my student number, faculty, the stage of my course, and my eligibility to get into certain labs and access certain online collections. What do these respective sets of such claims say about me in other contexts, say banking or healthcare? Very little. I can identify as Steve Wilson in my company and Steve Wilson at university, but any interoperability of these identities across contexts only happens at the attribute level, where the identity metaphor breaks down.

Post script

"Interoperability" is actually a curious omission in the Laws of Identity. The interoperability of atomic claims like date of birth, home address, credit card number, student number or SSN is almost trivial; some services recognise these claims and have business rules that use them, while others don't care about them. But the "interoperability" of a rolled-up set of claims like "Steve Wilson is employed by Lockstep Pty Ltd." makes almost no sense. The set of claims that make up that digital identity says a lot about me to a Relying Party doing business with Lockstep, but my corporate identity means nothing to retailers, doctors, my personal bank, the police or the video store.

Posted in Security, Language, Identity, Federated Identity

I just don't get Levels of Assurance

IDAM practitioners and government authentication policy makers settled on a generic categorisation of transaction risk and of quality-of-enrolment. Let's recap: the idea is to characterise the seriousness of a transaction in terms of "Levels of Assurance" (LOAs) -- typically one of four levels -- and then match the LOA of the party you're planning to do business with. LOA schemas are codified in NIST SP 800-63 and described more loosely in the Australian National Electronic Assurance Framework (NEAF).

The idea of LOAs can be traced to risk management methodologies and standards like AS 4360 and now ISO 31000. These approaches involve gauging the severity and frequency of anticipated adverse events, and combining them to deduce a rolled-up risk rating for each event on an ordinal scale, like {Negligible, Low, Medium, High, Extreme}. Examples given in the NEAF documentation use consequence-severity tables lifted straight out of AS 4360 (see Table 3, p15 of the NEAF Framework document).

A powerful feature of these risk driven approaches is that each enterprise is empowered (in fact expected) to create its own internal calibrations of adverse events. Severity can be gauged in different ways, by referencing monetary losses, health consequences, political impact and so on, and the most appropriate frame will depend on the business environment. Organisations also set their own policies for what level of risk is acceptable for each anticipated threat. So some will not tolerate residual risks that are worse than Low, while others will live with Medium risks on a case-by-case basis with special contingency plans. Risk management standards allow that different organisations have different risk appetites.

But what LOA advocates seem to forget is that, as a result, risk determinations made under ISO 31000 and the like are not transferable between organisations. Simply saying that a certain event (for example compromise to a user account) has a risk rating of “Medium” tells someone outside the organisation nothing at all about the details of the threat, its impacts, its expected likelihood, nor how it might be mitigated.

And yet the Levels of Assurance paradigm has us pick and choose externally issued identities based on a genericised rating of LOA 1, 2, 3 or 4. There really cannot be any definitive assurance that all "LOA 3" credentials for instance issued by all IdPs are equivalent, nor that they will satisfy the detailed needs of all Relying Parties conducting "LOA 3" transactions.

In other words, you cannot pigeon hole risk.

And I've seen repeatedly the rather farcical situation where RPs looking for say LOA 2 aren't quite satisfied with a given IdP's idea of LOA 2, and they seek to haggle over a special "level two and a bit". Generic LOAs are supposed to save time with generic levels, but in reality RPs and IdPs still do local risk management -- as they must -- and as a result, negotiate detailed authentication arrangements case by case. Don't get me wrong on this point; it is entirely proper that RP and IdP get together like this to hammer out their authentication arrangement. My criticism is that the idealised LOA bucket is irrelevant; it doesn't save time; it doesn't save anyone from having to do detailed case by case risk analysis.

The idea of discrete standardised LOAs was based on schemas that are used to communicate risk within organisations. They do not work for communicating about risk between organisations. It's misleading to represent discrete assurance levels, for it leads people to oversimplify the necessary matching of Identity Providers's offerings to Relying Parties' needs.

In fact, the very notion of "Identity Provision" is misleading, as I discuss here.

Posted in Security, Internet, Identity, Fraud, Federated Identity