I listen to quite the verity of music, and what I often do before I go to bed at night is do a search on an entire genre of music select everything, and hit download. Limewire does a reasonable job of telling me what I already have, and by morning I have added quite a few new songs to my library (many are keepers too ). You may of figured out by now that I really cannot stand to listen to the same old music for too long.

What I propose is a mode that I can drop Limewire into that downloads random songs that I don't already have from a list of genres.

It would do a search for all songs of a genre, then try to download them, one at a time, unless they are duplicates. If it was unable to connect for some reason, it would move onto the next song. Once it hit the end of the list it would move to the next genre in the given list, or start from the top again.

It would have a different hashing algo. with a higher sensitivity to pick out duplicates more often, but would also label more as duplicates that are not. An idea for the algo. would be one that checks the time of a file, and if the time matches, check for matches of words over 5 chars in the title/filename. If both are matches, call it a duplicate and move on to the next song.

Something like this I could leave running all night and have a nice verity of music picked form all over the web by morning. I was thinking of doing this myself, but I am afraid that I cannot tell how to compile the OS X version from source (I am no compiling wizard, and I am used to compiling command line programs without GUI interfaces, ie Apache, MySQL, PHP, etc.)

Ideas? Comments?