I stumbled across an email this morning from the Semantic Web Best Practices list that made me more aware of one of the main things that’s making me hesitate before diving into the Registry project… I’m certain that I don’t know enough about what I’m doing. So I spent a little bit of time browsing the SWBP list…

…and now I’m not sure anyone else knows enough about what they’re doing either. That somehow didn’t make me feel any better.

I just finished building some stairs, but these weren’t just stairs plain old everyday stairs. They’re stairs with drawers and cabinets with doors underneath nestled into the stair steps (I’ll post pictures when the finishing is done). My initial design called for a completely flat front with the doors and drawers flush with the cabinet surface and this kind of cabinetry requires a high degree of precision. But the stairs required a heavy duty infrastructure of 2×4s for strength and support. Modern 2×4s are NOT flat and this is an old house, it’s not flat either, or entirely square and level. So I built the frame for the stairs, built the cabinet face frame and discovered that despite my best efforts, things were a little bit thicker in places, not quite perfectly square in others. Now we’re talking maybe 1/16 inch over 3 or 4 feet in just a few places, which is pretty good for twisty 2×4s and an old house, but it’s not good enough for cabinetry. If I made my cabinet face frame absolutely precise, then over the course of an 8 foot rise and and a 7 foot run (yes, steep stairs) it would no longer fit the structural frame. Things like cabinet interiors and drawer runners were going to require a lot of shimming and they still wouldn’t fit right. So I changed the design and made the drawer fronts and the cabinet doors overlay the face frame instead of nesting flush and now you can’t tell just how out of whack the rest of the cabinet/stairs are because you don’t have the tiny, necessarily imprecise gaps between the drawer edges and the frames to show you.

And this, leaping suddenly to the point, is part of the problem that I have with RDF at the moment. In my humble opinion, the SWBP (and RDF) is trying to impose precision on something that’s inherently imprecise — language and the web.

From my admittedly limited perspective it seems that the Semantic Web community is still trying to make the drawers flush. As a consequence we’re having to do a lot of shimming of unstable and imprecise infrastructure, and we’re expecting that the infrastructure will be more precise in the future. But the infrastructure of language and the infrastructure of the web is already built and it’s not going to get more precise just because we need it to be.

“I suggest that if current real-world usage of a metadata vocabulary seems to be causing no actual operational problems, it might be better to study this real-world usage carefully with a view to learning something about how symbols actually are being used on the Web, than to set out to take great pains to improve it.” –Pat Hayes
http://lists.w3.org/Archives/Public/public-swbp-wg/2006Jan/0139.html

I think my hesitation boils down to being afraid either that I’m wrong about this and simply ignorant (or unnecessarily pessimistic) of realistic efforts to make the web and its semantics more precise, and that what we build will be considered too simplistic and imprecise to be useful (the infrastructure will un-twist); or we’re building something that will be widely perceived as contributing to making the web a more precise place, and expectations will be created that can’t possibly be met (the drawers will fit perfectly).

I spent a lot of time carefully designing the stairs — measuring and re-measuring, creating a computer model — knowing in advance that I would have to overcome the imprecision of an old house and dimensional lumber. In the end, I had to discard much of my careful design because it expected too much of my ability to alter an uncontrollable reality. Much the same thing is likely to happen with the Registry, and hesitating on the edge won’t improve the result. This leads me to the conclusion that diving in is the only way to go, and I should just get back to work.