Greetings. In Evil or misunderstood? Google and net neutrality posted earlier today, David Megginson speculates that a key issue in the now widely-known (but little understood) recent "secret" meetings between Google and Verizon might have been discussions regarding giving particular types of content differing priorities -- what I'd call Quality of Service (QoS) -- rather than negotiations about giving priority to any particular content providers per se. David also argues that content-based priority (e.g., giving video priority over e-mail) may make technical sense but is still a bad idea.

I have mixed feelings about all this. Theoretically, given sufficient bandwidth and well-behaved applications, QoS shouldn't really ever have to be an issue. After all, QoS is fundamentally a mechanism to manage scarcity. In the presence of abundance, QoS basically should be unnecessary.

Of course in the real world we have bandwidth limitations, bottlenecks, and the reality that not all Internet-based apps are necessarily well-behaved in their data usage -- so the concept of QoS probably needs to be on the table for discussion at least.

But it also seems clear that QoS (especially relatively straightforward implementations) has a number of potential problems -- some of which David discusses.

There are fundamentally two ways to do QoS -- either you analyze the traffic (based on actual content, traffic patterns, or both, perhaps using DPI - Deep Packet Inspection) -- and/or you base your QoS decisions on flags generated by content providers themselves.

What makes this all so complicated is that we must assume the presence of "bad actors" on the Net who will attempt to game any QoS system inappropriately.

Also, moves toward pervasive, ubiquitous encryption by default -- a concept for which I'm a strong proponent -- could significantly interfere with "automatically determined" QoS, putting more reliance on accurate tagging of data by content providers. Even traffic analysis can be fooled via various techniques.

The upshot of all this is -- gosh darn it! -- these issues are anything but trivial, and anyone who tells you otherwise is either misinformed or being purposely misleading.

Small wonder then, that trying to discuss these issues publicly, with all the money and emotions tangled up in the topic, tends to have a comfort level similar to that of sticking your hand down deep into the blades of a running blender.

While I don't presume to offer any magic wand solutions to these dilemmas, issues of content prioritization vs. Internet data content types are ignored at our peril.

If we do end up moving toward QoS systems of some sort, the real challenge will be finding ways to implement such mechanisms that do not interfere with users' ability to make full use of encryption, are as minimally vulnerable to unfair manipulation as possible, and that do not create distortions resulting in anticompetitive or other unfair behaviors.

A tall order.

And that's the truth.

--Lauren--