Carrier Evolution
The “Internet” never was intended to be the “next generation network” for all communications, despite its apparent suitability for any number of communications tasks. For service provider executives, the issue in the past has been largely about the suitability of “connection oriented” and “connectionless” approaches to networking. For real time services, connection-oriented architectures have clear advantages, namely the ability to support quality of service natively.
Those arguments haven’t typically made sense for bursty data traffic, which many network architects would say works just fine, at less cost, than using connection oriented networks.
And though there is a clear and important distinction to be made between the “Internet” and private IP networks, some will dispute the long term efficacy of trying to provide isochronous “real time” services (voice, fast twitch gaming and video telephony, for example) using IP.

Martin Geddes and Dan York, for example, disagree about the ultimate suitability of IP-based networks for real time services, for reasons related to the very protocols themselves. York, for example, thinks newer protocols such as WebRTC will work just fine. Geddes disagrees.

The disagreement boils down to a fundamental difference of opinion on how well IP-based networks can be made to work. In a sense, it is a philosophical debate (with real world protocol implications) over how well real-time services can be made to work over networks that simply never were architected with “real time” services in mind.

Those of you with an engineering bent will understand this as a “class of service” issue, but also fundamentally a protocol and architecture issue as well. Those of you with some memories of past debates will recognize that the “connectionless” and “connection-oriented” approach to networking is a subject that has not fully gone away.