Irrespective of the legal situation - which in any case varies by country and over time, I have my doubts about the technical and commercial practicality.
I think it is much more achievable in the fixed world, where the operator doesn't have to contend with the vagaries of radio, and where the presence of a "box" like a gateway gives a much better chance of monitoring what is actually delivered. The WiFi or ethernet connection at the fixed-broadband end-point to a final end-device (PC, TV, phone, tablet etc) also gives a clear demarcation point of responsibility. The operator can say with confidence that their bit of the end-to-end system did its job - and any issues with battery life, memory, device configuration and so forth are your problems. That's much more difficult with a smartphone - if the extra-quality video doesn't work, whose fault is it? And does the video provider still pay?
Nevertheless, whenever I spell out my concerns about differential charging for applications, I get bombarded by vendors (and some operators) insisting that their DPI box can detect absolutely everything, right down to what the user had for breakfast that morning.
Rather than get to an impasse, I thought that as well as commercial, legal and technical reasons, I'd also have a try at logical flaws in the argument - and perhaps highlight some extra opportunities along the way.
First off is prioritisation of the operators' in-house services. Many mobile carriers have their own video streaming, music or other rich application/content platforms. I'm assuming that some measure of optimisation is typically used by the operators to ensure these perform well - obviously it will be easier to test inhouse, and senior management can ensure adequate cooperation between network and application teams.
But.... if "real" quality can only be achieved at the level of manageable network QoS... and if "serious" content providers are willing to pay for it,.... then why not set up an effective structural separation between the services group and the network delivery team? If it actually came out of their own budget and P&L, would the inhouse video content team really pay money to the other department for improved network access? Or would they instead use rate-adaption and other tools to work around the limitations of best-efforts?
I haven't heard of any operators running an internal QoS market, but I'd be fascinated if any readers have anecdotes.
The next thought experiment takes this concept a bit further.
Now, consider the situation once again, where the operator's video content team is willing to pay extra for QoS to ensure their streaming is delivered better than it would be from the open Internet.
And consider that another operator offers network-based QoS - perhaps in the same country, or perhaps elsewhere in the world. Given there's already an "open market" in video streaming via YouTube, Hulu and so on.... shouldn't that operator in-house team therefore be prepared to deliver its content via other carriers' networks? Let's say, for the sake of argument, Verizon providing its video service to users on Telefonica O2 in the UK.
Given that they are in-house teams within operators, surely *they* understand better than anyone the capabilities - and potential differentiation - that comes from network-based prioritisation and QoS? If Verizon paid guaranteed-QoS fees to O2, shouldn't it be able to create and market a class-leading video service that end users would pay for? And shouldn't the O2 network team also think that Verizon's video people are therefore much easier to sell QoS to than, say, YouTube?
In other words.... if operators (and their vendors) really believe that "premium" network QoS can enhance the competitiveness of applications and content, or raise ARPU and improve customer satisfaction... why don't they put their money where their mouths are? If Telco X is clever enough and network-savvy enough to create a QoS- managed service that should outperform YouTube.... why hasn't it happened?
My last point is not about prioritisation, but coverage. Often, the gating factor on overall Quality of Experience is not radio or transport or core resource, it's a simple lack of decent signal. (Yes, I know that in theory coverage is a bit dependent on other users in the same cell, but let's just assume the culprit here is a thick stone wall).
Certainly, Vodafone is marketing a femtocell in the UK under the name of "Sure Signal" - and getting users to pay a premium for an "enhanced quality network" in their home or workplace. While some of the customers are just buying the femto to get any reliable signal at all, there is some anecdotal evidence that a proportion want "better than normal" coverage "5 bars, all the time!" - for example if they live inside, but near the edge of a cell. This tends to support the argument that a certain (smallish) group of users might pay extra for "gold service" QoS, however that is defined.
So then the question is... for an operator wanting to offer an improved *average* experience, both in terms of absolute coverage, and higher performance throughout each cell - is the best and cheapest mechanism really through network prioritisation? Or might it be more effective to do some sort of national-roaming deal with competitors, where the phone switches to a rival's network at a given location and time, if that network has a better signal and uncongested capacity?
Wouldn't it make sense for the mobile industry to have the equivalent of the airlines' interlining and codesharing agreements? In those situations, you can get an end-to-end ticket issued on Airline A, which covers one leg actually operated by Airline B. Your luggage gets "handed off" seamlessly and Airline A takes overall responsibility for end-to-end quality. Airline A benefits from Airline B's better coverage or schedule at a local level, while Airline B gets incremental revenue and traffic from Airline A's better sales and distribution to end users. Interestingly, low-cost carriers like EasyJet and Ryanair generally don't participate in these type of arrangements, only the premium-priced airlines do.
The analogy is simple.
If a customer with the "nameplate" Vodafone service sometimes actually gets connectivity via the Orange or T-Mobile network, would they really care, as long as their average performance went up? And, in fact, might they not even pay a premium for it? Could there ever become a distinction between a "full service network" (which puts you on one of its nominal rivals' cells when it gives you better coverage, but charges you more to cover the wholesale fees), versus the low-cost network which is all-or-nothing, running just its own, silo'd coverage?
Is the answer to better QoS (which customers are prepared to pay for) not another box in the network, but just better/richer wholesale arrangements and national roaming? Can we go further than current development in network-sharing, and have a more generalised platform for deals?
[Note: I know that it's not that simple, either because of regulation or because it takes a finite time to find, register and roam onto another network. But they could be improved by a telecom effort to find a convenient codeshare/interline approach]