I attended a very interesting meeting of the UNFOLD group exploring the emerging tools for the implementation of the IMS Learning Design specification. Although the tools look very good and are shaping up, there is still a long way to go. Some elements are yet to be developed, such as a user management tool and authentication interface that are interoperable with other systems and can be integrated.
The meeting raised the question whether Learning Design will ever be handled by teaching staff or if this is in fact a specialist area with specialist tools. This will obviously raise questions for economically stretched higher education institutions strapped for cash and incapable of creating extra posts.
The other thought that came to my mind was how scalable this solution would be and how an institution would manage large amounts of Units of Learning together with Learning Objects and other digital repositories. The ultimate vision I have (perhaps shared with Google) is the possibility of retrieving all institutional knowledge from a single and simple point of access.
Metadata for Learning Objects are a big issue in terms of cost to the maintaining organisation. The ongoing discussion about the structure of the data entry fields, the accuracy of what’s going in the fields, and who is putting it in, causes major headaches to institutions and may also have an impact on interoperability.
The dilemma is that if you have small granules of RLOs, such as e.g. media objects, they have a higher reusability, but cause a sometimes disproportionate amount of work and cost. Larger RLOs on the other hand may be quicker to tag, but are harder to reuse as they are more specific.
The all important thing about metadata is that describing and categorizing a resource makes it discoverable and retrievable. The IMS and other people emphasise this very strongly. While theoretically this may be true, practically it is difficult to achieve coherent descriptive values, and a well-organised maintainance structure – especially in small resource-limited institutions.
I want to pose a critical view here by asking whether the need for lengthy metadata is in fact overemphasised. Are we requiring masses of metadata to compensate for inadequate discovery tools?
I look at the Gmail approach to e-mail as an example for a radical revised strategy. The philosophy of Gmail is *not* to have a folder structure where users categorise their mail, but to provide them with a good and fast search tool instead. Could the same be applied to Learning Object repositories?