Kim Hamilton Duffy Interview (Co-chair of W3C Credentials Community Group and Architect of the Digital Academic Credentialing Infrastructure at MIT)
Self-Sovereign Identity allows individuals to control their personal data. To fully control it and the relationships where his or her identity is being used in. Choosing who gets to “see” it and how much of it they get to “see”.
Self-Sovereign Identity “privileges individual ownership of credentials, rather than custodianship of credentials by a software provider or issuing institution”. (source)
That’s according to Kim Hamilton Duffy, Co-chair of W3C Credentials Community Group and Architect of the Digital Academic Credentialing Infrastructure at MIT (Digital Academic Credentials Initiative).
For her, the path towards a truly decentralized identity management system is through Decentralized Identifiers (DIDs) and Verifiable Credentials. She sees these as fundamental elements of Self-Sovereign Identity, promising to address the shortcomings of existing decentralized credentialing solutions such as Blockcerts, a blockchain-based credentialing solution that Kim developed in collaboration with the MIT Media Lab in order to solve the problem of decentralized verification.
According to the W3C DID specification, “DIDs are URLs that relate a DID subject to means for trustable interactions with that subject”. They “enable the controller of a DID to prove control over it and to be implemented independently of any centralized registry, identity provider, or certificate authority”.
In an Self-Sovereign Identity ecosystem, individuals make claims about their identity, using DIDs, and those claims are “rendered tamper proof through digital signatures” stored on the blockchain. These claims can be verified anywhere, anytime. DIDs and Verifiable Claims allow for “persistent, independent digital identities” with increased privacy and security.
Kim considers DIDs an important tool for the proliferation of Self-Sovereign Identity. Managing cryptographic keys is a cumbersome process. DIDs make it easier for an individual to “retain ownership of their identifiers over time”. They offer cryptographic strength while factoring in the full lifecycle of keys, including expiration and revocation”. Decentralized Identifiers help prevent a “situation in which all of a person’s data is tied to a single individual identity profile” by allowing an individual to have as many DIDs as he or she may wish in order to “curate their identity profiles and increase their privacy”. (source)
DIDs also benefit institutions and organisations who issue or verify identity. Their decentralized nature makes identity always available for verification. As opposed to a system where identity is in a centralized database that may be rendered useless if it becomes offline for any reason (or, in a worst case scenario, destroyed).
On the Use Cases for Decentralized Identifiers, you can read what Kim considers the 15 required features for DIDs.
“The Blockcerts standard was published under the MIT open source license in 2016 so that any institution, vendor, or researcher can use it to build their own applications for issuing and verifying claims on the blockchain”. For Kim, it’s extremely important that identity solutions rely on open standards in order to achieve “maximum interoperability and portability of documents and data, without sacrificing privacy or individual control”. To her, openness and standards compliance are essential.
Kim believes that through open standards, DIDs and Verifiable Credentials, there is a possibility to evolve the identity management paradigm to one that preserves the privacy, security and self-sovereignty of the individual. With Blockchain opening the door for the possibility of true individual control of personal data.
We had the opportunity to ask Kim Hamilton Duffy a few questions:
What are your responsibilities and goals as the Architect of the Digital Academic Credentialing Infrastructure at MIT?
For context, the Digital Credentials Initiative is a university-led effort to develop a learner-centric digital credentialing ecosystem. I joined the initiative to drive the technical architecture and prototype/implementation rollout with the technical working group. Our initial focus is on standards, requirements, and shared infrastructure. We’re not defining competing standards; we’re identifying existing standards that suit our use cases, and extending/adapting them where necessary. So this effort complements existing credentialing standards (such as W3C Verifiable Credentials) and well-known vocabularies/taxonomies in the EDU/OCC space.
There are two characteristics that make this effort special. One is our participants’ commitment to include a broader range of perspectives and expertise. We felt that emerging decentralized credentialing standards showed a lot of promise, but that there were many open questions and gaps (not just technical — policy, governance, and more) that needed to be addressed. So for us, it was important to lead the effort with a well-rounded set of stakeholders.
The other interesting characteristic is our ability to strongly advocate for learner use cases. The learner side often gets deprioritized in existing credentialing systems, resulting in limited ability for learners to access, store, and use their credentials across systems (as an example, credential exchange protocols are still in early phases of development). This initiative is positioned (and committed) to drive these standards and requirements forward – and even develop reference implementations if the market is not providing them.
In the field of Digital Identity, what is the question that people should be asking more but aren’t?
It’s critical to have clearly defined use cases when dealing with digital identity. I think many efforts start that way, but then get muddied by adding on — almost as an afterthought — higher stakes use cases that are not well-understood. When I say “higher stakes”, I mean the stakes may be higher for the populations involved (e.g. displaced persons needing access to resources) or the nature of the claim itself (e.g. containing more private information). The risk is that a poorly-informed “solution” can do more harm than good.
One reason is that “identity” is so overloaded and potentially all-encompassing. If instead, we’re precise about the capabilities we are building, we may not need to “go to ‘identity’” (phrase borrowed from Steve Wilson). Further, pushing for use case clarity, as well as continually learning from intended users, allows us to build systems for our users (as opposed to systems that are imposed on them).
What needs to be true for SSI to achieve mass adoption, and what uses cases you think will gain early traction?
First, a caveat. SSI on its own is not something we can reasonably ask people to adopt — it’s an idea, with a confusing name at that, due to baggage associated with the word “sovereign”. (For clarity, I’d like to point readers to Christopher Allen’s 10 principles of self-sovereign identity and also Philip Sheldrake’s Generative Identity – beyond self-sovereignty). That said, those of us working in the decentralized identity space have not done a great job communicating the value-add or articulating concise use cases that, if solved, would actually warrant adoption.
An example: an SSI advocate might lead in with “imagine a Facebook but where you control your data”. Here are the problems with pitches like that:
- – Control and privacy are not features many users will pay for. In fact, some users accept (or at least claim to accept) that their data is already for sale so they might as well get paid for it. The downsides of thinking of data as property is more thoroughly analyzed in Elizabeth Renieris’ “Do we really want to “sell” ourselves? The risks of a property law paradigm for personal data ownership”.
- – Many users believe (rightly so, in many cases) that they must choose between usability/convenience on one hand and privacy/control on the other. Users are accustomed to the conveniences of centralized systems, and we need to work harder on UX.
In sum, we need to (1) improve usability and capabilities (particularly in “exceptional” cases requiring recovery of control), (2) develop interoperability standards currently missing in the decentralized identity stack, and (3) focus on compelling user-focused scenarios.
So there’s a lot of work to do, but there are some use cases that can obtain immediate traction. Those involve public claims (i.e., without sensitive data) that are improving efficiencies of existing workflows. This includes educational and occupational claims (equivalent to what you would post on LinkedIn), and government/business registries. As an example of the latter, Samantha Chase, founder of Venn Agency, is doing some interesting work around safety credentials backed by British Columbia’s Verifiable Organizations Network and OrgBook BC. This approach improves transparency and efficiency around safe workplace claims, which can further benefit companies through reduced costs (in the form of insurance discounts, for example).
Specific roadblocks other people in this space should look out for?
In the decentralized identity space, discussions around GDPR (and similar emerging privacy protections) have focused too much on rationalizing existing technical choices. It seems like every discussion about GDPR jumps to a debate about whether a hash of PII on a blockchain is acceptable. While I’ll steer away from that specific question, I think this misses the point, which is that designing systems for privacy and individual control offers exciting architectural challenges. It’s an opportunity to design systems more responsibly, which can lead to cleaner architectures that limit liability and exposure.
In my mind, GDPR has been generous in its rollout, and as long as system designers are being mindful about how user data is handled, and documenting decisions along the way, then they’ve made tremendous progress. And while it’s important to stay up-to-date as the consequences of such regulatory frameworks emerge, having documented your decisions along the way will make everyone’s life much easier.
What are your hopes for the digital identity field in the future?
I’d like for us to develop meaningful ways to include more diverse perspectives and expertise in our decentralized identity-focused groups. In our eagerness to develop core building blocks, technical folks (myself included) have sometimes inadvertently created sandboxes that exclude essential perspectives. We need to improve how we communicate about what we’re building. But more importantly, we need to actively engage, listen to, and accept leadership from people with a broader range of backgrounds and experience.
What are the books you have recommended most to others?
Because many of the writings I find interesting are not yet in book form, I’m going to include some blogs as well. Here are the writers/writings I keep coming back to:
- – Elizabeth Renieris’s blog
- – Kaliya Young and Heather Vescent, who have also written books on SSI:
- – Joe Andrieu’s Functional Identity primer
- – Christopher Allen’s Path to Self-Sovereign Identity
- – Philip Sheldrake’s blog
- – Phil Windley’s blog
- – Steve Wilson’s blog. He blogs at constellationr.com, and there’s not a clear entry point to his blog, so use the search. Here’s a great example of his content.
- – Tim Bouma’s blog. There are some remarkable decentralized identity efforts happening in Canada, many of which are led by Tim. Tim’s blog is a good way to stay informed of these efforts.
Tykn is a digital identity company. We are now about to launch Ana, a digital identity management platform that allows organisations to issue tamper-proof digital credentials which are verifiable anywhere, at any time. If you’re keen on reading more we suggest you check out our Blog. There are interviews with Daniel Hardman, Elizabeth M. Renieris and many more. There’s also our Definitive Guide to Identity Management with Blockchain.