I think this should not be a replica of CPAN... This is an opportunity to peer-review code that isn't polished enough to submit to CPAN to be saved for posterity. (Also it's an opportunity to review polished projects, but that shouldn't be the limit for inclusion.) If you catagorize by project, then yes, you'll have a lot of repeated projects that are impossible to choose between. If you catagorize by individual module or script, there will be many thousands of files to catagorize. I'm more in favor of an amorphous, dynamically organized collection, because who really wants to manually sort n-thousand individual scripts into catagories?

The code should definitely be indexed and searchable. Preferably NOT publically on google, because I don't want my grungy five-year-old mistakes being aired in public.

Code should be indexed on either comments and POD, or just on POD. That means the most descriptively documented code gets the best reviews. :-)

Is there any way to replicate google's "page rank" and further weight the result by a "code worthiness" and/or "coder XP rating"? That would help people who are looking for good code to learn from and use.

Finally- incentive to review. I think the XP system is a good one. Reviewers gain XP and grant code worthiness points. Code that's been deemed worthy should be offered upon the altar of CPAN. :-)

Implementation-wise, could this be done as E2, translating directories into nodes and files inside the directories as sub-nodes? (yes, I realize I said "finally" in the last paragraph. But implementation comes after design. :-)