A few days ago I posted some thoughts about learning repositories. Alan Levine commented with some detailed points that I think are worthy of a post in their own right.

The problem with learning objects to date is that everything has been focused on building the collections, making piles of objects, and almost nothing on how to create something meaningful with them.

I agree with this. And I'd go even further and suggest that a lot of effort has actually been spent on making fancy (but ultimately empty) containers to put these learning objects in. Implicit in this approach is an assumtion of "if you build it, they will come". Unfortunately, this assumption may not always hold. As a thought experiment, think about your own, your colleagues, and your students attitudes to sharing. Not especially in the context of sharing "learning objects", but sharing in general.

I'm an habitual sharer, but I know I'm unusual in this respect. As a blogger, I'm sure you have some of the same instincts. But sharing is not easy, and often misunderstood. Last year, I was censured by the Dean of Education at the college where I work and study for putting some of my own notes, taken during a teaching course I attended, on the web. I fought back, and gained a grudging acquiescence, but it became clear that implied college policy is that everything, somehow, ought to be kept secret.

How would you and your colleagues feel if your students could easily find (and copy/paste!) past work by other students who had attended your courses, or who had taken similar courses elsewhere? Would this make your work easier, or harder? How rigorous are you (and your colleagues) about researching and ackowledging all the sources of information and teaching materials? Might there be any copyright, licencing or potential plagiarism issues? How confident are you and your colleagues in writing style - "Publishing" is a long way from making a few rough notes for a small and specific audience. Might any of your, or any of your students', work cross any confidentiality boundaries? Who owns the work? Who has the right to share (or to withold) it? What is the commercial value of the work - do you have competitors? Might something in the work be possible cause for a lawsuit?

It seems there are an almost infinite number of reasons not to share. I have a gut feeling that maybe the time and effort spent on making learning object repositories might actually be better spent on encouraging a culture (and an understanding) of sharing. Without this there is a real danger that all these pretty and expensive boxes will remain populated only by self-conscious examples, placeholders, and tumbleweeds.

The problem with "Web as my repository" approach is that the web contains vastly more junk that would need to be sifted through. Busy teachers who are not techies want a more refined collection.

Certainly. But this is not a problem unique to this domain. It is faced by every special-interest group, and is being aggressively tackled by teams across the web. I find it difficult to see how educational object repositories are so special that they should be built in isolation, ignoring (or clumsily re-implementing) work done to improve internet "findability" in general. Why build a repository when you can get the same benefits by just building an index? Things like dmoz, h2g2, and Wikipedia might serve as good models to start with. I can see huge benefits simply from disassociating the the notion of "a way to find things" from "a place to store things". After all, we have the worlds biggest "place to store things" in place and working already :)

The theory also goes that such objects have the capability to be assembled into larger chunks of content, so there would be APIs or meta information that might describe what can be connected. The "authoring" is the assembling of these discrete pieces into a chunk of learning content.

I'd love to find out more about what progress is being made on this. I currently find it difficult to imagine how this might work. Proceding from the assumption that if diverse things are to "plug together" the connections need to be standardised, I would expect to see a lot of evolving draft specifications of formats, interfaces and adapters. I guess I must not be looking in the right places, because I can't find a way to "plug" my colleague's Visual Basic "quiz" program (with its wealth of PowerPoint content) together with my Java Wiki software and the JavaBlogs RSS aggregator service (for example).

The objects most certainly would not be textual- they would be things like Flash, applets, shockwave, etc- stuff that Google cannot readily index. What you describe for some people more more like information objects or just media assets. A learning object might be something like an applet for measuring pH that could be connected to say another code object that can simulate the collection of test materials.

This I understand. Indeed, it was the main thrust of one of my blog entries a while ago. My point this time, though, was that in cases like these it's not the object per se that should be indexed, but its metadata (and, incidentally, the references to it, reviews of it, discussion about it, comparisons with it, blog posts on it, and all the richness of the textual web.) One of the great enablers of the web is its distributed, asynchronous model - anyone can reference, connect, and annotate. Any monolithic repository design looks limited and inflexible in comparison with this.