*Attendees*
Grace Agnew, Rutgers
Steve Carmody, Brown
Jeanette Fielden, Internet2
Keith Hazelton, U. Wisconsin
Tom Barton, U. Chicago
Mike Grady, U. Illinois
Steve Olshansky, Internet2
Scott Cantor, OSU

*Discussion*
Grace has sent the latest draft of her scenario to the list. The scenario is two universities are team teaching a course. One is in the U.K. and the other in New Jersey so there are differences in spelling and terminology. This requires a globally unique identifier so orthographic differences are transparent and trivial. The main entities are the instructors, enrollees and the digital repository. The repository would be Shib enabled and grant access to resources based on authentication and role-based authorization. Course enrollee's can be at either university; they will have access to the resources based on CourseID for just that course offering. It's expected that roles, the course offering, etc. would have unique identifiers. Enrollee's would be able to access and download materials based on enrollment, identified by courseID and time bound by the course offering. Instructors would be able to add, edit, and remove materials. Administrators would have a consistent way to authenticate users not registered at their university but needing temporary access to resources. This could be used to create adjunct instructor access as well. The courseID could enable differential resource access for instructors at the application level. Instructors would have credentials in only one origin site.

Scott suggested emphasizing that the authorization decision is being delegated to the other origin site so people understand what is going on and that institutions will have to come to agreement over it.
Add a header for future more complex scenarios.

[AI] Grace will add a header for future more complex scenarios. Please review Grace's scenarios and provide comments by Friday. Grace will update scenario to reflect comments.
[AI] Scott will try to revise his scenario for discussion at the BoF.

Grace indicated that most everything would be an identifier, not just the course level. The courseID is not meaningful without a course offering ID as well. There will be a need for data dictionaries to arrive at common understanding of roles, offering, course. Scott indicated that another level below affiliation would be needed, because role can't be avoided.

Are there drivers from the Shib side? The some plans to proof out concepts and how they work. Penn State and WebAssign are asserting a student role. There are also projects underway with a Shibbolized Blackboard at U. Texas and U. South Florida. Texas is looking to have it in production by January for courses that have students at multiple campuses. Blackboard has raised the possibility of dynamically creating user objects within Blackboard by presenting an appropriate attribute set from the origin. That attribute set would resemble what the courseID group might develop. These efforts can be regarded as bottom up, with IMS and others coming from the top down. Where they don't meet is where courseID can focus. The London School of Economics has been discussing some proposals as well and will be at the Internet2 meeting.
[AI] Steve Carmody will check with U. Texas and U. South Florida on their Blackboard projects.
[AI] Keith will talk with Barry Ribbeck about the U. Texas work and courseID opportunities.

The courseID group needs to be explicit about defining the scope of this effort and where this group will use existing work and where new work will need to be done. Scoping would be useful to make sure that people don't assume that we're going to attempt the impossible task of evaluating if course A is the same as course B elsewhere.

Most difficulties around identifiers have to do with resolution and not identification. Everyone wants to immediately figure out what the identifier points to. Identifiers are semantic placeholders that represent some agreement between institutions. So we don't have to describe what the identifier is, it's just the way we refer to this particular course offering.

Grace indicated that if an URN syntax is used, the courseID would be the final end of the string, not the entire URN. Keith felt that it might be thought of more as OID's (object identifiers). If you can think of that as a sequence of steps that guarantees uniqueness of the thing at the right end, it leads to fewer problems. The left end of the string could have an administrative hierarchical ownership/relationship.

It's important to unlink issues so that meta-data associated with an object is not confused with being in the identifier itself. It's intended to be the linking common element between an action and an end result and not about what the future actions and results would be. Focusing on a combination of identifiers plus the meta-data that the identifier is linked with would be good.

Grace is speaking at an invitational conference for a federated educational repository next week in Wisconsin. She'll talk about a common robust data dictionary that can be tied to identifiers. She will reference courseID and how it would play through the repository to see if there is interest among attendees in the courseID work.

Internet2 Fall Member Meeting BoF
The BoF will discuss Tom's eduCourse structure and the scenarios. Definitions will be addressed only as raised by Tom's document during discussion. Feedback, what campuses are working on, and comments on requirements will be collected.
[AI] Grace and Keith will meet next Monday in Wisconsin to discuss scenarios and draft agenda.
[AI] Keith will mail out draft agenda for feedback.

There will be no call on October 13 due to the Member Meeting.