As someone who has spent on an ambitious set of extensions that will probably explode into complete obscurity on release, I am definitely sympathetic to the attempt to expand the visibility / reward / feedback factor for extensions

.
To the cool idea of an extensions census, I would add that it would be nice to (somehow) get author-bloggers to review extensions; i.e., to make discussion of extensions more a part of the "culture" of IF authorship. Especially in the case of Inform 7, with its minimal standard library, extensions are a key component of the system, so the
I7 news page really ought to make mention of at least the most newsworthy extensions. (There doesn't seem to be a TADS equivalent to the news page, unfortunately. I've also found it somewhat difficult to find TADS extensions.).
(For me, extensions have also been extremely important central to learning the software. Want to really get into Inform 7? Write a complex extension, preferably one that has you scraping around in the I6 layer...)
Back to the census: it seems to me that it would be a lot less work for the organizer to simply use the IFDB. It already has a tagging system, and it would be easy to glom onto that. Of course, the tagging system isn't ideal for this purpose, but it's probably good enough. (It might be nice to segregate extension tags from others, though; maybe via a subcategory of authoring system tags?)
Following this model, the census site could be a check-offable list of games to be tagged, a list of official tag names (to prevent orphaned tags that are spelled differently and thus don't group with others), and maybe a discussion board...?
--Erik