Well, it DOES seem to me that anyone trying to find the shadows would be better served with a google search (and probably a lot more likely to start with a google search as well). I'd agree. Anything that removes non-working and/or non-useful cruft is a good thing.
While removing cruft and shortening boilerplate are laudable goals, the _correct_ solution to having published ID's with URL's that no longer work, given that the namespace is under our control, is to FIX THE BROKEN PAGES!
There's no reason that http://www.ietf.org/ietf/1id-abstracts.txt, which has worked for a very long time and is published in lots of places, should not continue to return a copy of the _current_ I-D abstracts file.
-- Jeff