He's right on the money when he emphasizes the importance of an open interfaces that permit reuse:
What I realized through the course of my conversation with Raymond was that the most critical aspect of a new Flickr like service is not really an attractive user experience. Certainly, that's essential to help find images and associate data with them; it's also what makes an application desirable enough to initiate use. However, what will make the application ultimately successful is the availability of open services that permit re-use: mashups that encourage integration with other services and content.
It is this feature - the ability to disintermediate the content, at least partially, from the restricting frame of the application - that is the most fundamentally important virtue of Flickr, and should be for any application that is premised on widespread internet adoption and use. Support for re-use enables a plethora of user experiences to be designed and developed. The needs of mobile users are distinct from highly interactive uses in in-world education, and in turn vastly different from many research imperatives.
This reminds me of a comment by Richard Wallis at Talis about a "reading list" (UK speak for course reserves) product that they were coming out with that he (somewhat boastfully) noted, doesn't even have a user interface. It simply provides data feeds and APIs for syndication/integration of content in other contexts like the course management system.
Really, you'd get three main things with an 'academic Flickr' that you don't get with the kinds of curated digital collections done now in libraries:
- network effects of a shared web scale database (shared tagging, comments, search, etc)
- a great best of breed user interface that allows convenient uploading, tagging, organization, image editing, and licensing of content
- an open API to permit reuse of content