Identity, verification and blockchains

It’s been a while since I last blogged. Since then, one of the things I’ve been spending a lot of time on is the concept of ID&V, especially the challenge of joining the “ID” (identity) and “V” (verification) elements. I’ve been able to spend rather less time on blockchains and shared ledger technology, but following a couple of excellent sessions at the Barclays Digital Conference last week, I have been thinking about the implications of how virtual identity transforms real world identities and the implictions once that's translated into blockchain like systems. These are more than just technology implications, there will be some important philisophical and legal implication from however virtual identity is created and defined.

Fortunately, there’s been an increasing and rather dramatic increase in interest in the blockchain as a source of identity and I’m able to bring these two of my interests together in this post. If you’re not familiar with how the idea of a blockchain based virtual identity might work there’s lots of good material, but a couple I’d recommend as starting points are:

• “How do I know it’s the digital you?” Back in April Chris Skinner wrote an excellent post summarising discussions at the Innonvate Finance Global Summit on his Finanseer blog.
• “Putting “identity” on the “blockchain”. Parts 1- 4”. The always very readable Dave Birch goes through the design considerations of how the blockchain might be used to create a transferrable, sharable but privacy protected identity.
• “ID2020” . I’m huge admirer of the work John Edge and others are driving, aiming to create a digital identity that is independent of government. The idea of uncoupling identity from the state has huge potential, especially when you consider refugees, failed states and the more kleptocratic/ tyrannical states. Given the huge possible implications for Financial Services, Chris Skinner provides a good starter summary here “The impossible dream: a digital identity for everyone”.
• “R3 Corda: What makes it different”. Really, I’d recommend almost anything by Richard Gendall on blockchain and distributed ledger, but here he sets out very clearly some of the distinctions between Bitcoin type systems (where the point is not to reveal the participant identity) and other blockchain use case such as R3’s, where protecting the participant identity is less important than ensuring transaction integrity, confidentiality and security. He provides a more detailed walk through in his excellent earlier April article: “Introducing R3 Corda™: A Distributed Ledger Designed for Financial Services”.
• Simon Taylor “10 Things You Need to Know About Blockchain This Week” Always a good read and also well worth listening to his weekly 11 FS Podcast.

What interests me most is where the blockchain (or distributed ledge, the two are not automatically synonymous) needs to interact with the “real” world and therefore be linked to a unique person or other unique real world and potentially physical identity. This challenge is one that banks world-wide have dealt with for a long time – “how does the person in front of them (physical or virtual) relate to virtual or abstracted assets?”.

 Traditionally the ID&V remote interactions in financial services has been based around a log-in and passwords. The principle is that is if the password is only something the customer could know, and the password is kept secret by customer and bank, then the person entering should be the customer.

The challenge with this is that the password measures what the customer knows and then equates that knowledge to the person entering that data with having that identity. In many circumstances this log-in based proof of identity is quite adequate.

Where it can get interesting is when you look at the discrepancies between banking channels. In contrast to remote channels, the branch channel too often relies on only a signature and the teller’s common sense as verification methods.  Due to publicity, there is a tendency to see the voice and contact centre channel as particularly vulnerable. In practice, though, (and when well-designed) voice is one of the more secure channels and branch is one of the more vulnerable. Dave Birch provides a good summary of some real identity frauds exploiting by branch security here: ( "Strong Consumer Authentication with Gloria Hunniford, Gold Membership and Gary Munro" ) By contrast the voice channel authentication password requirements need be no less secure than on-line or app based banking and with biometrics, the voice channel has the potential to be significantly more secure (see my last post blog post “What does HSBC’s adoption of voice biometrics tell us about trends in identity, customers and contact centres?”).

I think the challenge comes when we start looking at some of the more ambitious projects for identity through blockchain and shared ledger. The idea is a fascinating one, that with your birth, you will a get a blockchain or shared ledger identity (from whom/ how varies a bit) and as you go through life key data (such as academic qualifications) will be irreversibly added to it. If you’re not familiar with the thinking then Wired published a good introduction in January this year, “How the blockchain will enable self-service government”.

The challenge I see is firstly relating this virtual identity to real, physical individuals and secondly allowing that permanent, virtual identity to be changed, recovered or hidden or altered (which is generally what blockchains are built to avoid happening).

The second challenge is perhaps the most important. I would argue that we should assume that identity theft will continue and evolve. It is not a new issues (it goes back to Martin Guerre and beyond)  but in an increasingly virtual world the barriers to it happening have lowered and that not everyone can be assumed to keep their identity safe for their entire lifetime.

Identity theft is more than just capturing a person’s log-in details, at its worst a person’s whole virtual identity is taken over and sometimes even the physical identity is challenged (e.g. if a stolen virtual identity is used to create new photographic or biometric ID).  It can lead to more than just bank accounts being emptied, the stolen identity can be used to take out loans and conduct criminal activities

Therefore, all identities on a block chain need to somehow be re-set and potentially have any fraudulent transactions they are tied to removed.

This has further implications when we think of managing identity through a life-time. There may be things (such as changing nationality to escape persecution or changing gender) where leaving behind records of past activity might be deeply undesirable. Yet a large point of a blockchain based identity system is to provide an accumulated identity, built up over time and spread across many points.

Returning to the first point, there’s no doubt that the lack of transferable, authoritative virtual identity is a significant problem and one that needs a long-term solution. The challenge is that if we create a virtual identity that is one source of all proof, how do we keep that identity tied to the real person?

One suggestion is that the blockchain incorporates biometrics or (rather better) is linked to a secure store of biometrics (though that seems to negate some of the point of the distributed blockchain by inserting one or more trust providers into a trustless system). On the face of it, this seems a reasonable strategy. Biometrics are unique to individuals and, if built into the blockchain, are then locked in and provide an insoluble record.

Yet someone who’s worked with biometrics, this causes concern. Biometrics are very good at verification but few would pretend they are infallible in every situation. Blind faith in any technology is something I’m always concerned about and “biometrics”, while very good, are not a magic bullet.

A very specific concern of mine is that biometrics are often referred to as if they were all the same, but there are many different biometrics and they are not all equal or equally accessible. For any system, the ease of access to the user’s biometrics (e.g. things like voice or fingerprints) needs to be traded against the confidence level (e.g. the near certainty provided by a genetic print from material captured in lab conditions). All can provide a high confidence level, but just how high do you want it to be if it was your only proof of existence for translation into a virtual world?

A separate concern is that if we are talking about an ID for a lifetime, then biometrics may not be so absolute. For example, over a lifetime voices do change. We now tend to understand gender as less absolute than previous generations did. As medicine explores the idea of gene therapy, genetic identity may not be quite so permanent (and if that resolves some hereditary illnesses, we will be all the better for it).

Perhaps instead we would be better making sure identity we translate into the virtual world was more fluid, more adaptable and keeping verification distributed?

In the real world we take for granted that different audiences see different aspects of our identity. For example, the identity we use for paying tax is not necessarily the one we use on social media and even within Social Media a LinkedIn profile may be very different from a Facebook profile.

So is a blockchain based identity practical?

I think the answer is “yes” but it needs to be designed very carefully. It may also need to start with the sort of areas that can manage it best and already need a virtual identity (such as land registration or financial services transactions). I’m very sympathetic to the idea of a citizen owned and managed identity, but I think that is several steps further down the road. Identity is so tightly bound to bigger (and philosophical) issues, such as legal rights and ownership, that I cannot see these being re-engineered quickly. It’s also critical that some of these big, underlying issues are debated publicly, rather than just driven by technology, but that’s perhaps a future post.