I'm interested in a code dump, even if figuratively it appears like a dump. Why? Because I think the code dump is such a good idea that I don't want the overhead of putting the code in public to overwhelm the contributor of the code (in addition to whatever prevented the author to put up the code for public consumption in the first place). As long as everyone knows what the code dump is for, no need to put too much overhead in the implementation.

It seems to me that a link to the site that used the code and the author of the code (so proper attribution can be given) are enough overhead. A semi-sensible organization of the code (by site that used the code) should be enough, too. So if I want to see the implementation of a neat feature in greenpeace.org, for example, I look at the greenpeace.org subdirectory in the code dump and find a bunch of tar balls that Don uploaded. Fine, I'll download the tar balls and look at the code in my own sweet time, and I'll try to figure out the portion of the code that implemented the feature myself. No need to bother Don to describe what the tarballs are for, or separate the tar balls into "digestible" pieces, because he may not have the time.