Net neutrality is effectively random

[Commentary] Network neutrality is a program of price regulation, and it has always run aground on the question of service differentiation. Net neutrality advocates have typically been content to apply a uniform pricing structure across the board because they’ve never believed there could be more than one service on the Internet. When Tim Wu wrote his seminal paper, “Network Neutrality, Broadband Discrimination,” it appeared to activists that Internet service was necessarily (or optimally) uniform.

The net neutrality activists of the early 2000s believed that Internet service had always consisted of a “best-efforts” packet delivery service. The hallmarks of this service were first-in, first-out transfers across data circuits that were highly (but not perfectly) reliable. Hence, applications had to be prepared to request retransmission of lost packets, but not overly concerned — this scenario was rare. Similarly, application designers had to be ready to accommodate large variations in transmission delay (latency and jitter). But broadband networks that provide Internet service have always provided other services as well: cable TV and telephone service, most notably. Wu wasn’t concerned about these other services because he regarded them as “not the Internet” and therefore outside the scope of his proposed Internet regulations. It’s now clear that the exclusion of voice and video from the definition of Internet service was a glaring error: Most of the traffic that transits the Internet today is either voice or video.

[Richard Bennett was vice-chair of the Institute of Electrical and Electronics Engineers Standards Association (IEEE-SA) 802.3 standards group]