Agreed.

The rest of this thread is a little bit misguided; the entire discussion is centered around users and writing the tags. Those things are not a problem. It will be as problematic writing embedded code in HTML as it is now writing CSS, DHTML, PHP, JavaScript, etc. We've never been too lazy to incorporate more code before, and we're not all too lazy to incorporate some more tags now, that's ridiculous.

Organizing the incorporation of multiple and related ontologies on the Semantic Web is much tougher, and there's a lot of research being done in the area. It's probably quite doable with some global guidelines.

The problems you discuss here are significantly detramental to the goals of the Semantic Web. I wonder what "basic inference techniques" you speak of. Somehow I still argue in favor of more metadata.