The CPAN is one of the best things about Perl. http://search.cpan.org goes a long way towards making the CPAN searchable, but it could be so much better. The code behind http://search.cpan.org is not open source (that we know of), so we're not able to contribute back to it. Instead of coding a replacement for search.cpan.org, we're proposing a free, open web service which will provide useful data about CPAN distributions, modules and authors.
Take http://cpanratings.perl.org as an example. There is no simple way (which we know of) to query a web service in order to get a module's rating. As it stands, you need to download a .csv file, parse it and then extract the data you're looking for. This would be much easier to use if you could simply query a web service, receive a JSON data structure and then be on your way.
We're just in the very early stages, but if you want to play with it, check out the API documentation.
Please contribute to the spec on tagging.
You can also join us on our mailing list: http://lists.scsys.co.uk/cgi-bin/mailman/listinfo/metacpan (old list archive)
And on irc: #metacpan on irc.perl.org
Initially this project was proposed as a web service which could be used specifically by iCPAN, an iPhone application which keeps an offline version of most CPAN modules. Here's a brief synopsis of what we were thinking.
Here's a loosely organized list of topics which merit further discussion: