Making an Open edX Learning Object Repository

[ormsbee]

Assume the following:

• We want to start by focusing on the small-Item reusability use case.
• We've split the concepts of Learning Package and Learning Context.

What would a centralized LOR look like, and what's the easiest way it could work for end users?

Executing in-place

A number of institutions like MIT operate multiple instances and have difficulty managing their content across those instances, leading to various solutions over time. Instead of trying to facilitate content replication across instances, what if we serve content centrally? Not in the sense that there's only ever one server, but in the sense that an organization could choose to manage all their content in one place.

So what could that look like? Some possibilities:

• An LTI Advantage interface would be a basic requirement.
• We could expose a separate iframe-based API for the anonymous-user use case. This would drastically lower the burden of using very small pieces of content in random places. We would want to have some API layer over this to do things like iframe-resize messages, so this could likely be encapsulated into a web component.
• We'd probably need a way to assign things like attempts, hints settings, weighted score, and other settings at the Learning Context level.
• We may need to create some site <-> learning context mapping so that certain learning contexts only get delivered out of certain domains.
• The authoring interface is in this new web application.
• There would be a runtime dependency between LMS instances and one or more LOR instances.
• Security would be a major concern if we're hosting some shared LOR space. We'd have to make sure frontend code was executed in iframes, and potentially unsafe backend code was executed in sandboxed processes (WASM?).

This could be a pretty jarring authoring experience for course authoring though, particularly if people are bouncing back between the Studio and one or possibly multiple LOR instances.

[jmakowski1123]

Couple thoughts:

• I agree that we'd need to minimize having authors bounce between Studio and one or more Commons. I think we can design solutions and workflows to mitigate this. For example, similar to the way we're thinking about baking a "menu" of LTI apps into Studio, we could consider the same here - ie, baking a search/discovery layer into the Studio authoring flow that would enable authors an easy way to find and pull over the content they need from the Commons.
• Instead of thinking about one Commons per institution, what if there were one central Open edX Commons, with levels of access restrictions at the institution level? For example, MIT wants to make Batch A of sequences share-able only across MIT instances, but they are happy to make Batch B of sequences available for any Open edX author to use in any instance.
• Both bullet points above reinforce the need to continue fleshing stories around descriptive metadata, as that will be the key to ensuring content is searchable and discoverable in a meaningful way. Descriptive metadata would need to be attached to Learning Packages (components, items, sequences). And for imagining potential down the road - outcome data could be attached to Learning Packages too.
• I'm not sure we can (or want) to call this thing a Commons. For further thought -

[ormsbee]

Instead of thinking about one Commons per institution, what if there were one central Open edX Commons, with levels of access restrictions at the institution level? For example, MIT wants to make Batch A of sequences share-able only across MIT instances, but they are happy to make Batch B of sequences available for any Open edX author to use in any instance.

Yeah, I think we'd want to operate a big site that multiple institutions could use. But the software being open source, I imagine that the end state is going to have a few different instances in different places, and that it wouldn't work to assume a 1:1 mapping of Commons instances to Open edX Studio or LMS instances.

I'm not sure we can (or want) to call this thing a Commons. For further thought -

Yeah, I'm definitely open to other names on this one. I've thought of a half dozen others and even the most ridiculous of those was taken. 😛

[ormsbee]

Just replaced Commons with LOR for now, even though this is a bit more expansive than what a LOR would traditionally be.