I was reading Dare Obasanjo’s blog today in which he disputes the assertion that GDS is equivalent to WinFS, more or less credibly. Specifically, I agree with this: “Full text search is really orthogonal to what WinFS is supposed to enable on the Windows platform,” but do not agree with this: “Technologies like Google Desktop Search solve a problem a few people have while WinFS is aimed at solving a problem most computer users have.”
Either way, I think both have their place, and while GDS has a several year window (no pun intended) to evolve and add new functionality, I doubt very much that the two products will ever match up in terms of their overall intent.
Beyond this debate, however, Obasanjo compares the benefits of WinFS to those of iTunes. That is the ability to process a unique file type – music, in this case – differently than other files via an interface. It accomplishes this via an awareness that music is a file with unique abilities (i.e. the ability to produce sound) and by leveraging the metadata associated with that file. That is, as my interpretation of his argument goes, the underlying task that WinFS will enable – the leveraging of asset type and metadata for new programmatic functionality. I agree, sounds cool.
But I think it also highlights a problem that a lot of people are likely to have with their individual assets. Namely, that a very small percentage of assets have any metadata associated with them.
What’s the problem? Well, to borrow Obasanjo’s metaphor, think of iTunes without any track names, or artist names, or album names. Not really all that useful. And that, as far as I can tell, is the most likely outcome. How many of you have tagged your digital pictures (beyond the exif data inserted by your camera)? How many of your Word documents or Excel files are tagged with information about their contents?
The point here is that I’m in total agreement that applications or filesystems that are aware of both file type and its associated metadata can do some amazing things, but I question how many people are going to be willing to spend the time required for an initial metadata population – and then maintain it going forward.
