I think it's a great idea, and not at all because I came up with it.
The same algorithms that determine similarity between artists, users and
so on can be applied to torrents for a neverending stream of
recommendations.
Have you guys written any of those yet? If not, all you have to do is
reference each torrent (you can't check magnet links for download quite
as easily) in a list of files downloaded by (hopefully pseudonymous)
users, and then determine which torrents are closest by determining the
modes of preference (the modes of the set of files that all people who
have downloaded a specific file have also downloaded). Etc.
On 02/28/2010 05:09 AM, Matija Šuklje wrote:
Dne nedelja 28. februarja 2010 ob 02:05:09 je Alberto Rodríguez napisal(a):
Have you investigated about the legallity of what you are proposing?
Since Libre.FM is planned to handle only free (CC and other copyleft) music
and seems to be very picky about the licenses it allows, I don't see a
problem.
I'll double-check with my professor[1] just in case there's some problem with
making magnet (or torrent) links. But I'm 99% sure this should be OK, because
one of the main intentions of free content license is that the content should
be distributable.
Cheers,
Matija
-.-.-.-
[1] I'm a law student in my final year, the prof in question is a FSF member
and has studied under Eben Moglen[2]
[2] RMS's right hand; the lawyer who actually wrote GPL;
http://en.wikipedia.org/wiki/Eben_Moglen
