The proposals as they stand are simply daft
1) Every ISP would have to buy in hugely expensive deep packet inspection equipment (which might increase latency) and devote a lot of staff time to the issue. The equipment used by some ISPs to traffic shape is usually only capable of identifying protocol, not the actual payload, so even ISPs with some DPI equipment already in place would probably spend a lot replacing/upgrading it. Inevitably the end users, including those who don't know what an MP3 is will end up paying for this in the end via higher subscription fees.
2) The equipment would not be able to decrypt encrypted traffic. This means it would be rendered useless pretty much as soon as its introduced
3) It would end up with false positives. Assuming it identifies copyright files by some sort of pattern matching on the filename of the file gives vast scope for mistakes. What about people who are falsely accused, or had their wireless network hacked or similar. If cut off due to a false accusation then the ISP could well be held liable. This stands true even for the simpler identification technique of a few BPI computers seeding the popular torrents and logging the IP addresses of people who download them (this is how they identify file sharers for the infamous US lawsuits, and we've all heard the stories of them suing elderly folk who switch on the computer once a week to check email, so its hardly infallible)
4) Its probably illegal to do this anyway under privacy Law
Kindest regards,
James Sweet
http://www.zen.co.uk