Really cool, now all someone needs to do is mash this up with the RedditStorage project (https://github.com/Rossem/RedditStorage) to make this a real live thing.
post the media description, and then people would comment the magnet links for that and everyone would up/down vote based on if that really is the content advertised or something that does not match the md5, low quality, etc.
the only downside is that they archive older threads. defeating the purpose of this research, which is the longevity of data.