The Verifiable Claims Task Force

A Task Force of the Web Payments Interest Group

Verifiable Claims Telecon

Minutes for 2016-05-31

Gregg Kellogg is scribing.

Topic: Introduction to New Participants

Kiara Robles: I’m interning at BlockChain; I’ve been involved with the design workshop and 2020; hoping to attened MIT workshop.
Adam Lake: Started with Digital Bazaar last week.
Dave Crocker: I’ve been working on internet protocols and architecture for a long time, mostly email. Went to New York and found myself interested in your architecture and ways to improve it.
Joe Andrieu: I came because of IIW; I’m between startups but writing a book on requirements modeling.
… The UN summit made me think I might have something to contribut here.
Adam Madlin: I’m on the Accelerate team on the DHS SIR team on blockchain-based identity. I’ve been working a long time in government standards.
… I spend most of my time working on cyber-security standards for fed and state governments
Drummond Reed: I’m founder of Respect Network; we were awarded a DHS grant on blockchain identityh.
Jason Law: Building a distributed legister dedicated to distributed identity.
Bob Bonomo: I’m an independent consultant focusing on Block Chain and working on standards.

Topic: Self-Sovereign Identity Architecture

Manu Sporny: For the past 4 years, starting in WPCG, we’ve been talking about identity and exchanging claims (“attestations”)
… These concepts have been around for a while, and been in the “Identity Ecosystem” over the last 10-15 years. Mainly in IIW and other forums.
… Recently, it’s been called “self-soverign identity”. At the ID2020 event and rebooting web-of-trust workship, the concept of an architecture surfaced again.
… During this, dcrocker and others took a look at our work. We’re going to show the W3C members in a couple of months, so we’re trying to get it in good shape. It’s not done, but it’s a first cut.
… We need to quickly come to conensus.
Dave Crocker: I want to stress my own view of archiecture diagrams is that they need to be simple and easy to understand.
… Ballancing this, they need to be complete enough.
… On average, diagrams are overly complicated. I was delighted with the diagrams from this work.
… In the NY discussion, we looked at ways to enhance it (not replace). We explored this over hours and Manu circulated some diagrams based on these talks.
… I realized that no where in the diagrams is there an indication of the identiy itself. Maybe there shouldn’t be, but we should think about that.
Manu Sporny: Here are the documents that we worked on at ID2020 - again *VERY PRELIMINARY / JUST FOR DISCUSSION*
Drummond Reed: I would argue that this is exactly what self-sovereign identity is about - there is NO BOX for a self-sovereign identity because there is no conventional "identity provider"
… I also find it useful to have diagrams which work at different levels. For example, public behavior vs private behavior. In email, there’s the user agent and mail transfer agent. Looking at RC 598, UA and MTA are still there, but there is also quite a bit more.
… The fact the original is still valid is still useful.
… For example, the repository is internal to the holder, so it might not need to be shown in the basic architecture.
Manu Sporny: For those who don’t know, Dave is one of the people who created email standards.
Christopher Allen: I was intregued by was that in the original doc, we had DID in the center, and then it was moved over to be a number of different oracles. Can you explain?
Dave Crocker: I expect this to be controversial. A side point: the diagram has DiD and similarly the other box had “ledger”. One of the challenges is to distinguish between abstactions and implementations. “Ledger” has taken on the notion of an implementation approach. Similar with DiD.
… The other actors in the system need to query for information; there are different types of queries. The identiy itself, history of claims, other dynamic data. The ledger reference refers to a growing journal, but has data static when created. One might want to query dynamic data.
… The idea is to merge all of these into a single abstraction, which were listed in the box out of NY.
Manu Sporny: The Oracle box includes ledgers, generators, registry and file stores.
Christopher Allen: I’m intregued by this change as it relates to smart signatures and can make use of a variety of different oracles. It’s an interesting new way to think about it.
Dave Longley: +1 For this path
Manu Sporny: Good news is that there’s not screaming to keep going down this path. In the coming weeks, we can refine these diagrams with the goal of fanalizing in the 3rd week of june. That’s when the WPIG would start reviewing, and see if we want to take to an IG meeting on July 1st.
Dave Crocker: I think that’s a realistic timeframe. Let me encourage to contact me directly offline.

Topic: Decentralized Identifiers

Manu Sporny: In this group we’ve known that the decentralized identifier problem is very difficult to solve, but required for user-centric/self-soverign identiy.
Christopher Allen: Link to 10 principles of Self-Sovereign Identity
… The key think is the understanding that we need to have an identifier that can’t be taken away from an individual through force or coersion. We create the concept of a “WebDHT” (a hash table with history), but that’s getting into implementation details.
… A higher level requirement is that we have an identifier that can be instantiated and owned so that it can’t be taken away.
… We’ve discussed block chain, journal, DHT a number of times, but we should wind back and look at requirements.
Drummond Reed: It was a very intense discussion on DiD architecture. I have some notes freom that discussion. We got to concensus on goals of the architecture. We came down to two key goals: define the structure of an identifier that can serve as a discoverable key for a value.
… The value part is the “DiD object”: define the structure of an object that can provide cryptographic proof of DiD ownership and update permission.
… Also, for the DiD Object to provide pointers to other sources of claims and other peer DiDs. For purposes of privace protection, a person (or principal) may need to have multiple/many DiDs to keep different personas separate.
… From the standpoint of claims-based ID, the concept of a DiD-Tuple is a genesis claim, and is the starting point.
… most discussion was over a “public DiD”, so that you can refer to it to discover,lookup and verify a DiD. This corresponds to an “omni-directional identifier”. However, from a privacy consideration there are limitations. There are also cases for private DiDs shared by a limited set of holders, thus a “uni-directional identifier".
… There was discussion around DiD discovery. Could use ledgers or oracles as peers, which provides independence but not a known starting point.
… The second bucket was “addressable ledgers”, was to treat them as part of the URI space (ledger of ledgers). This would make the DiD itself addressable, but has questions of portability.
… The third option was meta-protocols to allow ledger interaction.
… The last option as the “index ledger”, the idea that by concensus there is a single DiD registry, where the object may be stored elsewhere, but there is a master index.
… We’re interviewing individuals intersted in this area and want to contribute; Anyone on this call is weclome to participate.
Christopher Allen: We’re punting on naming, which has caused problems in the past. This should be reflected in the archiecture. Names are credentials not identifiers. People want the right to revoke credentials that are harmful, and this may have not been thought of enough.
… Secondly, on the meta-DiD discussion, we wanted to be able to move the root identier from one chain to another. Self-soverign users would have a choice of what chain to use, as well as the ability to move it later. We think we have some good answers by having every chain have it’s own DiD.
… These things are hard, but we think we can say it’s possible with this architecture, but don’t want to have to prove it first. The intent is to have a design which allows for portability, but this may be a 2.0 feature.
Manu Sporny: The DiD stuff is a long-term need for verifiable claims work. Today, in the current architecture, we’re just using URLs. If you want to use it today, you must bind the claim to a URL, which places the domain holder in control. We’d like the DiD stuff to replace that.
Dave Longley: Notes that, technically, it's still a URL if the scheme is "did" instead of "https"... this is still a URL: "did:abcd1234..."
… There is a system up already used in demos. The issue is that it’s just backed by a regular DB, it needs to be distributed.
Dave Longley: It's just dereferenced in some other way (TBD)
… We don’t need it to be ready by the time W3C membership votes, but we should have a plan.
Drummond Reed: One of the time-frames is based on the 6-month DHS contract. Our job there is to perform as many interviews as we can, trying to do 20-30 by the end of June and produce a sample architecture and prototype.
… We anticipate that a spec should/would come out of this. I’d like to get through the heart of the work this summer and be able to implement/test in the fall.
Christopher Allen: Block Stack has implemented various aspects of this in their own architecture. We’re working with them to bring alignment; in effect, they offer DHT and could come into alignment with the overall DiD and self-soverign architecture.
… In oreder to do DiDs property you need a proof of existence, which was also worked on at the design workship, that will make the work easier to move forward.
… There’s interest in a standard way to do this, possible using linked data sigiatureas as a way to do this.

Topic: Verifiable Claims Use Cases Review

Les Chasen: In regards to the interviews that Drummond mentioned, we are currently running a survey re "applicability of blockchain tech" here Feel free to take it and indicate whether you are interested in a further in-depth interview.
Manu Sporny: At ID2020 we also broke out to talk about use cases. They looked at Shane’s document for potential W3C vote and had some thoughts about it.
… I wanted to introduce Joe Andrieu to talk about this.
Drummond Reed: Also, this a reference to Kim Cameron's Laws of Identity (with the law about omnidirectional and unidirectional identifiers):
Joe Andrieu: I wanted to be sure that my writeup aligns with Shane’s work.
… Our first impressions was that we wished the human value was more prominent. It should discuss how system addresses real human problems.
Manu Sporny: We're looking at this document:
… For example, easily humanly recallable titles. I appriciate that there is confusion over this. Things like “the lazy doctor” in UC 4.3.1. We tried to come up with a short name capturing the human value of each use case. A lazy doctor doesn’t maintain credentials, but a bad university has credentials pulled.
… I’d like to give some examples of requirements modeling (not engineering).
… I have some tools and techniques that can make these use cases easier to absorb, to capture essential interactions.
… One of the notes is that there is a lot to digest, and there was tension between a rigorous specification/detail, but this is at odds with the process.
… This puts the focus on a really good architectural diagram.
Manu Sporny: Speaking for myself, I think that would be fantastic. You’re group was forced to do it in an hour, which is really good. The W3C Membership may spend 5-10 minutes on it before moving on. The UC doc ends up driving requirements for a future WG.
… People make two passes of UC, 5 minutes to vote to proceed, and the second set is people actually doing the work, who will look with a comb to determine if work is in scope.
Shane McCarron: You said you had trouble getting your mind on the difference between scenarios and use cases. Note that it was structured as requirements, with supporting use cases.
Joe Andrieu: Use Cases are overloaded to mean a lot of different things. But sometimes UC is in the problem domain, or the solution domain.

Topic: Update on Work Items

Manu Sporny: There are a number of documents that still need to be updated.
… I’m going to send out another request on the survey.
… Also, we’re hoping that Adam will work on AC or Architecture summary.
Adam Lake: I’m getting up to speed on working on both. We should be on track for 3rd week of June.
Manu Sporny: The W3C advisory committee needs a brief summary so they know why they should vote for this.
Adam Lake: I also have feedback on the other documents, including Use Cases.
Adam Lake: If the authors of those docs want it.
Christopher Allen: There’s a blockchain workship coming up 2 days before the one at MIT
David Ezell: Let’s work on getting a story infront of WPIG ASAP.
Joe Andrieu: Thanks, all. I need to run. I've joined the email list and will focus on something presentable to the group within the week.
Manu Sporny: Next week: heads down on getting voting package together.