The Internet After Net Neutrality

As we approach the FCC’s repeal of its Obama-era net neutrality regulations, boosters are panicking. Chairman Ajit Pai has made a bold decision, moving enforcement of unfair and deceptive broadband trade practices to the Federal Trade Commission. The implications are sinking in: Pai has called the net neutrality movement’s bluff and they want to delay the Commission’s vote.

Prophesies of the end of the Internet as we know it will now be tested, and they’re probably going to fail: ISPs are not going to offer pay-to-play fast lanes to websites. This has long been clear to networking geeks because the web comes nowhere close to using the speed ISPs make available to it already.

The Web Performance Gap

The average speed of America’s fixed-line broadband connections is 75 Mbps (according to If ISPs controlled how quickly we receive content, we would expect web pages to load at speeds close to these measured averages. But they’re nowhere close.

According to the FCC’s measurements, typical web pages won’t load faster than 12 – 15 Mbps regardless of network speed. Changing from a 15 Mbps plan to a gigabit subscription will make gaming and video editing faster, but it won’t do a thing for mainstream video streaming and the web.

No web-based service is going to pay ISPs to speed up parts of the Internet that consistently outperform their own equipment. Some apps will benefit from – and pay for – higher quality, but not the web.

Speeding up the Web

When Internet companies want to deliver their content faster they pay content delivery networks. CDNs provide more servers placed close to end users than small companies can buy on their own. They do this because overloaded web servers are the speed bumps on the road to a good Internet experience. But net neutrality advocates have taught us to blame ISPs even when they’re not at fault.

ISPs have never offered web acceleration for a fee. They could have offered such a service at any time prior to 2015, but they didn’t because it makes absolutely no sense. ISPs can’t “accelerate” web traffic, they can only forward it to users as fast as they receive it from websites, which isn’t very fast.

Former FCC chairman Tom Wheeler failed to conduct a thorough, impartial analysis of the advocacy claims animating his decision to impose bright line rules on broadband networks. Incidents of misbehavior have occurred, but not so frequently as to warrant pre-emptive regulations. The practices he banned can all be beneficial in certain contexts, rhetoric notwithstanding.

Crying Wolf

The death of net neutrality is an existential threat to the pressure groups who’ve claimed the Internet owes more to federal regulators than to technical innovators and risk-taking entrepreneurs. If the Internet works better next year and the year after than it does today, we’ll know the net neutrality movement has been crying wolf.

By the same token, if it collapses we’ll applaud them for telling us the honest truth and rush to restore Wheeler’s regulations.

Calling the Bluff

In erasing the 2015 order, Chairman Pai is closing a policy debate that has raged in Washington for 15 years and in the engineering community since the 1970s.

Few net neutrality advocates realize that the current Internet regulation argument is an echo of historical engineering discourse about where to locate control points – traffic lights – in computer networks. When computers were slower, it was necessary to choose between more lanes on the information freeway and metering lights.

Creative Synthesis

The technical debate came to an end in the ‘90s when semiconductor chips became fast enough to do both. Modern Ethernet is a synthesis of 1970s Ethernet and its competitor, the IBM Token Ring. The policy community didn’t get the memo on this development and the pressure groups don’t appear to care.

The truth is that networks need to be extremely fast, highly reliable, and very well-behaved. But networks are only one part of the Internet, and not the most troublesome or dangerous one at that. Content, malware, and surveillance are the real trouble spots.

Having it All

We’re paying the price for placing too little emphasis on the health of the overall Internet and too much on the facets – such as broadband speed – easiest to measure. Net neutrality didn’t cause fake news, inattention to the social consequences of user-generated content did.

Net neutrality’s creators and protectors – chiefly, law professors Mark Lemley, Lawrence Lessig, Tim Wu, and Barbara van Schewick – guessed that the Internet’s ideal regulatory paradigm might be inherent in its design. This guess was wrong, but it took 20 years to disprove.

Let’s not be distracted by shiny objects any more. The Internet still has tremendous promise as well as serious problems to solve. Making it better through continuous experimentation should be the top priority.

See my research paper: “You Get What You Measure: Internet Performance as a Policy Tool” for detailed breakdowns of Internet performance factors.