Metadata, what metadata?

Should a well implemented metadata system be transparent to the user? An obvious question perhaps but a valid one nonetheless and depending upon the context the answer might not be as obvious as you may think.

Metadata is all around us yet how aware are we of its existence? An example. Most of us use Google at some point to search for pages on the web. Google uses metadata, quite a bit of it in fact yet most is not obviously apparent to us, at least not as metadata per se. There's a web page's title, metadata of a kind though you might rightly argue that's part of the page's data rather than metadata. More obvious metadata for any given page might include the date the page was last modified (available via the web server's page cache), the language, mime type (not all objects on the web are HTML web pages; think pictures, video, PDFs, etc.) and a few other bits of information. A web page's URL is also metadata, and that's quite an important part of searching Google, not least because Google creates its own metadata about a web page used to compile its PageRank. All these metadata are used to construct your search results when you search Google. Thankfully you can be blissfully unaware of their existence and still get pretty authoritative results. However, awareness of Google metadata can help you perform more focused searches but generally most of us are happy with the results we get when using the familiar single, simple search box.

Then there's learning object metadata. How aware are we of that, and indeed my main question, how aware should we be? The current learning object metadata specification from the IEEE weighs in around 69 metadata fields, that is 69 boxes in which something could be entered to describe a learning object. Let me say at the outset that you don't have to enter anything into any of these fields, they're not mandatory. But of course if a metadata system is to be of any use you'd have to use at least some of these fields otherwise what's the point. And it's a given that some metadata is essential for resource discovery and reuse, and therefore standardization in what those metadata are is essential. However, depending upon what your learning object is, some IEEE LOM fields will be more useful than others at describing your object. Some groups are working hard at helping to create a consensus as to what are the core fields that a learning object could use to aid resource discovery and reuse.

There's clearly an overhead associated with creating metadata as any data that can't be automatically gleaned from the object itself or created by the system a la Google has to be added by someone. Consequently there must also be a trade-off between the utility of metadata vs the cost of adding metadata. At one extreme no metadata is probably not going to be very helpful (and actually quite difficult to achieve given the inherent metadata surrounding any object placed on the web – see above). At the other extreme a comprehensively completed IEEE LOM record is likely to be too costly for many objects. As a result of this trade-off one of the hot topics in e-learning is trying to identify where the balance is, recognising that it's probably going to be different depending upon the context.

So, back to the question. Should a well implemented metadata system be transparent to the user? And a supplementary question, how can we as learners use this metadata to enhance our learning experience beyond that which was possible before the creation of the IEEE LOM?

Are there any really effective implementations of metadata e.g. the IEEE LOM in learning management systems that are transparent to the user yet sufficiently useful to justify the effort that went into creating them? And the $64,000 question, are these implementations used, and if so how, by whom and to what effect?