The Myth of Internet Fast Lanes

Should three unelected bureaucrats be able to reverse three other unelected bureaucrats on vital social, political, and economic questions? This is the haunting question for Internet policy in the United States.

This method appears to be perfectly acceptable to many partisans as long as the most recent vote aligns with their political values, but not at all acceptable to others. Rep. Marsha Blackburn, chairman of the House Communications and Technology Subcommittee, isn’t thrilled about it.

According to media sources, Blackburn proposes to pass a law settling the regulatory status of Internet Service Providers for the foreseeable future. It’s an uphill battle because it will need at least a little bit of Democratic support in the Senate.

A Nasty Battle from Day One

Keeping the net neutrality controversy alive is important to pressure groups Fight for the Future, Free Press, and Demand Progress because it’s their only issue. So it comes as no surprise that they’re complaining about Blackburn’s bill before they’ve read it. Free Press calls it “fake legislation”, everyone’s favorite aspersion.

Partisan bloggers are also fanning flames: TechDirt’s Karl Bode lashed out on Twitter against Washington Post writer Brian Fung for daring to acknowledge an innocuous statement by free market think tank leader Berin Szóka. Szóka’s offense was pointing out that the bill is a long way from a law.

Apparently, the bill fails to mention – let alone ban – the vexing practice of “pay-to-play fast lanes”. It also fails to provide the FCC with the wide-ranging power to sanction ISPs for engaging in violations that aren’t actually banned by any up-front rule. The latter is the sort of thing the FTC does, so let’s focus on the lanes question.

It’s All About the Lanes

An awful lot of nutty stuff has been written about net neutrality on both sides; so much that it’s often hard to tell the the serious from the crazy talk. So let’s rely on a senior advisor to the FCC chairman who imposed Title II on ISPs for the first time to explain what that order did. Here’s Gigi Sohn writing in the New York Daily News:

[T]he rules prohibited broadband providers from blocking or throttling online content, and barred them from charging online companies a second fee to reach their customers faster; known as “fast lanes.”

Because Blackburn’s bill bans blocking and throttling, Sohn and her comrades will now be forced to admit that these practices have never been a serious problem. According to Sohn, the lanes are vital because no startup can flourish without fast lanes:

Without the protection that net neutrality offers, broadband providers can charge you and your competitors for fast lanes…If you can’t afford to get in the fast lane, you’ll be relegated to the slow lane. That means that your big competitors’ websites will load faster, while yours will load slower. Studies have shown that Internet users will wait no longer than about 3 seconds for a website to load. Life in the slow lane means fewer customers, fewer revenues and profits and likely a quick demise.

The important part of this statement is the unstated assumption: Sohn presumes that ISPs and ISPs alone control the speed at which web pages load. This bit of rhetorical misdirection sails right by many bloggers, journalists, and pundits because they’ve seen it so many times they don’t question it. But it’s categorically false.

Except it Isn’t

As we’ve pointed out many times, Sohn’s unstated presumption is inconsistent with every last piece of data we have on web page performance. Despite the fact that the average download speed on America’s broadband networks is 75 – 80 Mbps, web page speed is 12 – 15 Mbps.

Web page load time varies a lot by site: the Washington Post loads half of its front page in less than two seconds, while the New York Times doesn’t reach the half way point (where the screen is full even though the lower part of the page is still loading) for more than eight seconds in WebPageTest.org’s simulations. The Los Angeles Times reaches the halfway point in three seconds, but the Wall Street Journal takes more than seven.

None of the sites is paying for a fast lane. Each is accelerated by a content delivery network (CDN) provided by a neutral party, but their performance is radically different. The factors that account for the speed discrepancies are page design, the number and nature of the ads embedded in the pages, and the number and power of the servers that host the pages on their various CDNs.

While Sohn’s claim that users prefer fast web pages to slow ones is correct, her diagnosis that speed is in the hands of ISPs could not be further from the truth.

Trends over Time

If ISPs were in control of web page load time, we would expect to see a strong relationship between broadband and web speeds, but no such relationship exists. Broadband speed steadily increases over time, but web page performance jumps all over the place.

Between 2010 and 2017, the Post’s above-the-fold time decreased from six seconds to less than two, while NYT’s increased from four to six. Both sites benefited from faster broadband, but they made different choices about page design, ad density, and CDN performance that were much more important than ISP factors.

Since 2010, ISPs have placed a major emphasis on performance. Comcast brags that its network is the fastest in the nation, offering speeds of hundreds of megabits per second. Verizon, AT&T, and the others also market their services by speed tiers.

This holds true for both wired and mobile packages, and even the satellite carriers tout increasing speeds. On average, US broadband speeds increased 35% per year during this period, while web speed only increased 5% per year.

Limited Choices

The ISP emphasis on speed is, by itself, enough to negate any possibility of ISPs offering “fast lanes” to web-based services. Because web sites don’t use the capacity that’s already available to them, the ISPs would need to move most of them into very slow lanes – no more than 10 Mbps – in order to speed up other web pages for a fee.

This would mean that all the consumers who pay for higher speeds would have a nifty little class action against any ISP who offered to speed up the web for a fee. Consumers would obviously win this suit, which would make ISPs very unhappy.

The ISPs’ only choice for website “fast lanes” would be an offer of CDN services. This development would create a very interesting competitive dynamic between ISPs and neutral CDNs.

The ISPs could play games with neutral CDNs for price and performance, but that sort of tussle would most likely be beneficial to consumers and to websites: both would benefit from greater choice, lower prices, and higher performance. And if the ISPs went too far, the FTC would be likely to examine the market for anti-competitive practices. Some management practices are hard to detect, but slowing 75 Mbps connections to 8 Mbps isn’t subtle.

The Framing Matters

Commercial conflicts between large CDNs and large ISPs have a very different character than the David v. Goliath tussles that Sohn and comrades have talked up for the last decade. Sohn tries to troll up the drama, but the only likely conflicts pit relative giants against each other.

Nobody talks about the importance of net neutrality regulations to CDNs but this is the only place where fast lanes can possibly happen for websites. There are other apps that need fast lanes – the genuine real-time apps such as gaming, augmented reality, and video conferencing – but the Title II fans don’t care about them.

So let’s bring on the the debate about fast lanes. The fans of old-fashioned, Stanford-style, “all packets are equal” logic will lose that one every time, and Silicon Valley is utterly unprepared for it.

The Valley’s lobbyists – the Internet Association and Engine – are so clueless about the subtleties of network engineering that they still claim it impossible to speed up one stream without degrading another. That’s so far from the truth it’s not even wrong; it’s just crazy.