About Making Data Better episode 9, “The Bigger Step: Want a Single Digital ID?”
The presumption of a singular ID
We’ve been writing about the new Australian Government Digital ID System (AGDIS) and talking about it the Making Data Better podcast. We’re reading press reports, speaking to commentators, and finding some ambiguity about whether this thing is might be a new ID or attribute.
The Finance Minister Katy Gallagher, responsible for the Digital ID Bill, says she is not in the business of producing a new national ID. We believe her.
Yet many still believe AGDIS must be producing some sort of new attribute. It could provide a new way to do KYC checks when opening a bank account, or proving how old we are when we buy alcohol online.
A lot is left to the imagination: Does AGDIS really mean a new sort of attribute, a de facto ID? No, and the spectre of a singular ID distracts from the true potential of Digital ID to fix what’s broken.
In our latest short conversation, George and I try to clarify what AGDIS could be delivering.
How do we think about the prospect of new attributes?
We’ve been celebrating the fact that with the Digital ID Bill, the government is not talking about “digital identity” anymore.
But the term Digital ID in the public mind could be construed as a new identifier or national ID, and many don’t like that idea.
The good people of Australia, the US, the UK, Canada and New Zealand — the Common Law system countries — are famously allergic to National ID. So, there are some optics to be concerned about.
But much more important is the question of how relying parties use any ID, be it old or new. It’s the relying parties that actually make use of IDs and take a risk on them. We’ve seen over and over around the world that the success of digital identification is in the hands of relying parties.
The critical problem with a new digital ID is that the relying parties’ existing systems don’t know how to consume it.
They will not have seen it before. They won’t know what to make of it.
Even if lawyers and risk officers gave it the all-clear, a new ID needs to be incorporated into subsystems and business processes that have operated for years around driver’s licenses, birth certificates and Medicare numbers. Subsystems include onboarding routines, user interfaces, databases, risk assessments, and call centre processes. They would all need updating for a new ID.
The need for flexibility in identification
KYC (Know Your Customer) has a bad reputation and it’s far from perfect, but it works in familiar ways, drawing on sets of well-established IDs which can be put together in different ways to onboard new customers.
Every bank actually does KYC differently. KYC conventions allow for variations.
This is critical. Think of the challenge of “low doc” customers — individuals without the affordances of passports or driver licences who still have the same right to participate in the finance system. To accommodate customers with diverse personal documents, banks can vary their KYC processes.
Banks know their own way business environments and rightly have the discretion to customise their identification processes.
When KYC draws on lists of established IDs, it allows for flexibility, and compensatory controls when identification pathways need to be varied.
But if we throw a new attribute into the ecosystem — especially one that is supposed to be better or preferred — then we force the banks back to square one to work out what to do with it.
We know perfectly well from the failure of grand Identity Federations that trust isn’t transitive. That is, the work done by one party to manage its risks is not going to satisfy other parties.
Instead of any new IDs or bundles of attributes, we need to make better data available to established identification processes. The Australian Digital ID System can do that simply through verifiable credentials that make it impossible for ID data to be replayed by fraudsters.
Let’s move away from plaintext presentation of IDs and adopt device-bound presentation just like we have with our chip payment cards and digital wallets, where hardware and software work together to make each presentation unique and non-replayable.
The golden opportunity we have now comes from so many people having computers in their hands that can present and release data in a far more secure way than filling out a form.
So in conclusion, we understand why people can suspect that Digital ID going to be a new ID, but it’s not. No new ID is needed. To fix online identification, we can conserve all the IDs we know today and make the move from plaintext to device-bound smart presentation.