Differential QoS - applicable to applications as well as access?

I'm currently doing a lot of thinking about the thorny issue of net neutrality, DPI, and whether operators (fixed or mobile) can charge "upstream" content or application service providers for higher levels of QoS, guaranteed bandwidth and so forth. Much of this fits into the discussion of two-sided business models by organisations like the Telco 2.0 Initiative, of which I am an associate, as well as concepts like "sender-data" from groups like the MEF.

I've long been a skeptic about the practicalities of "hard" policy management - blocking certain applications, or degrading them, especially when done in a covert fashion. Comcast's controversial "sandvining" of P2P epitomised the technical, commercial and regulatory problems involved.

But I've been less-bothered by clear network policies, stated unambiguously by broadband operators in properly-competitive markets. If you're a low-tier BT Broadband user and your BBC iPlayer can't run HD video, you know the reason why and have plenty of choice if you want to churn. Similar situation if you want to run VoIP over a 3G modem - you pays your money, and makes your choice. Yes, often the T's and C's could certainly be a bit clearer and more accessible, but the info is there if you want to look for it.

But the big unanswered question is whether operators can actually *charge* for differential service. Will the broadband providers be able to get media or other Internet companies to pay for guarantees of bandwidth or latency? On what basis? How will it be measured and monitored? What happens if there's a problem? Why would anyone pay for QoS during periods of no network congestion? There are a thousand questions to answer.

But there's one in particular that strikes me as the elephant in the room - what happens if certain large & popular Internet application or content companies play the same game, but in reverse?

Could YouTube deny subscribers of a given ISP the right to get access to HD content, unless the ISP pays a fee? Maybe users would churn to get what they want? What about if Google Maps had lower accuracy for subscribers of the less-cooperative mobile operators, and clearly identified that it was a low-tier service? Maybe Facebook could introduce new features selectively?

This becomes a particularly powerful strategy where users have multiple ways to get to the same application - if YouTube on the iPhone starts to be "better" than YouTube on the PC, we could even see broadband providers competing on "Who gives the best YouTube performance".

The fly in the ointment is the typical strategy of the Internet companies to get as wide an audience as possible to maximise advertising revenue. But at some point, the economics might start to look compelling.

My view is that there are almost certainly some good business models for differential network QoS-based charging.... but operators need to be exceedinly careful about how they position them, as QoS could turn out to be a double-edged sword. There is definitely a risk that potential "upstream customers" may in fact turn out to be "upstream costs", rather than a source of revenue.