Versions Compared

Key

  • This line was added.
  • This line was removed.
  • Formatting was changed.

...

  • Resource owners and those acting on their behalf: what language do we need to support the distinction? Any additional use cases? Any additional insight on technical design patterns?
    • We said we'd try to conclude this decision-making process by this week, and then move on to the decisions about RS actions in contradiction.

Attending: Eve, Adrian, Ann, Andrew, Mark, Sal, Kathleen, John

Kathleen had summarized it in our WG call yesterday that this is an "owner" and an "on behalf of" problem. Adrian asks: Is it clear that there need to be separate credentials for these two roles (something like subject and custodian)? Kathleen's work at HL7 includes three roles: the subject of consent directive (of governed information), the grantor (the consentor who grants access), and the grantee (the one to whom access is delegated). Eve is familiar with a project where UMA is being contemplated where subject and delegate are the terms used, where she suspects (subject+grantor) and grantee map perfectly onto Kathleen's terms, and she knows they map exactly onto UMA's RO and RqP terms.

Let's put the "third term" aside since it seems there's no controversy about the mappings (for the moment), and concentrate on the "split RO" nature of the use cases in front of us.

Do we have to distinguish "who logs in to the AS" vs. "who logs in to the RS only"?  Adrian asks: In our use cases from last week, are there separate credentials? We need to know the exact UMA design pattern we're looking at to know the answer, because otherwise the specific identity ecosystem in place is unknown. Andrew suggests: Is the question really about whether the RSO cares who the ASO is? Let's say a hospital IT department is the RSO and the healthcare insurer (Eve adds: or EHR SaaS vendor or government agency or whatever...) is the ASO.

Going back to the key issue: Adrian presumes that the AS presents a certificate and the RS gets to look at that certificate and see if it trusts it. But Eve points out this is off-track for the subject of this call. Any AS cert is for the whole AS. Let's return to (subject+grantor).

Regarding the terms coming from HL7, FHIR is taking this nomenclature work even farther. Eve wonders: Could subject and grantor suffice as terms for us, even if the subject isn't even literally the data/PII subject? Like, e.g., what if the "subject" created the digital artwork that is being sold and the grantor is managing access to the painting on behalf of the subject because they're no longer legally competent to be party to contracts? In health, a donor or cadaver might not be donating data, but body parts! (?) 

The current theory/hope is that Subject could be a good new term for the "true resource owner" who might not be legally competent or might not be in a position to manage resource access at the AS. (Note that "control" is the verb we use on the UMA marvelous spiral architecture diagram to denote what the RO does at the AS.) If we need to get more specific in any of our model clauses regarding resources that truly contain PII, perhaps we need to clarify in the definition that UMA allows for resources not to contain PII, and then define a more specific term, Data Subject or PII Subject or Resource Subject (as discussed in notes below), specifically for resources that do contain it. (Do we also have to define the term Resource and/or Protected Resource in our model definitions?)

More current theory/hope: Grantor could be a good new term for the "guardian" of a legally incompetent Subject, or designated "proxy"/"agent" for a legally competent Subject, who exclusively manages resource access at the AS on their behalf. [Added after the call:] Eve's theory about the UMA design pattern in this case is this: The Grantor is the only one that interacts with the AS (e.g., logs in there, creates policies, etc.) If the Subject is capable of managing resources in an online fashion ("manage" is the verb we use on the UMA marvelous spiral architecture diagram to denote what the RO does at the RS, but also the verb we use to denote what the RqP does at the client), then in fact what could happen is that the Grantor sets up policies for the Subject to become an RqP to their own resources, accessing them through RS-that-is-also-a-C, and requiring that the Subject use the same AS as the Grantor to enable full Grantor oversight capabilities. (This is a known limitation due to #wideeco realities.)

More current theory/hope: We could replace Authorizing Party with Subject and Grantor.

More current theory/hope: Grantee could be a good new term for Requesting Party.

AI: Kathleen: Share HL7-based definitions of subject and grantor (and grantee) with UMA WG so we can try out definitions for our own purposes. (ISO standard basis for the definition – what is the ISO number? -- can't be shared because they're not free standards.)

AI: Adrian: Work with Kathleen to capture the "joint control" of painting resource example and share with the list.

AI: Eve: Add GitHub issues for the new model definition ideas.

2016-02-19

  • Distinguishing resource subjects from resource owners: Can we develop a cohesive system whereby "resource subjects" without legal capacity can have "authorized agents" acting on their behalf as "resource owners" as required in order to forge "resource registration agreements" for the purpose of UMA's phase 1 particulars? Do the use cases/design patterns provide any insights or challenges here?

...