Clearly the goal is to hinder file sharing by reducing P2P traffic. Bandwidth however is not costly, not in short supply, and it is getting cheaper and cheaper. There would never be any reason to throttle bandwidth – and therefore no need to engage in DPI and monitoring – when internet traffic is not at or near capacity.

I think there is a serious negative consequence – that traffic from "preferred" sources will be expedited automatically because of some ostensible reputation benefit. Service from other less-known sources will have their traffic throttled making it far more difficult for an upstart to compete with a site like youtube even though it may offer some unique and novel advantages.

A better analogy is: would you want the post office to look at the envelope to see whether or not you paid for expediated/overnight service or would you prefer they dump it all in the same 'standard service' bin that everything else goes in regardless of how critical your mail is?

Most service providers aren't considering going into the payload (reading your mail) – and they don't need to. The information they require to identify the application being used is generally part of the TCP header or can be identified through heuristic techniques. Knowing what the application is, is enough to properly sort traffic.

And this isn't necessarily a bad thing. Some application traffic (think voice, video) *must* be delivered expediantly and in order to work, other applications (think file transfer) aren't as sensitive to jitter and latency. Without intervention *all* Internet applications are teated as 'best effort' and are subject to the same congestion management algorithms. Would you like your time sensitive applications to work or not?