This coming Thursday, Feb. 26, the Federal Communications Commission will formally pass its fourth set of regulations on broadband Internet services. It’s a sobering thought.
The Internet is working well, so it’s not obvious that the FCC needs to help it. American companies own 10 of the world’s 15 largest websites (Google, Amazon, and Facebook to name an obvious few); the United States has greater access to advanced cable and fiber networks than any large country except Japan; it was the first to deploy advanced 4G/LTE mobile networks; it has more smartphones than anywhere else in the world; and it exports more digital goods per capita than any other nation.
These facts are indisputable, so they’re simply disregarded by the Internet regulation advocates campaigning for net neutrality. Among the arguments they use to make their case are that some foreign cities and small nations have built extremely speedy residential networks; many of these offer Internet services for a fraction of U.S. prices; rural American communities have slower and less reliable networks than cities do; and many older people have no interest in venturing onto the Internet at any price.
A core problem with these arguments is that they are, in truth, unrelated to net neutrality.
The FCC says it’s not passing new rules in hopes of improving the Internet but to preserve it as it is with “light touch regulations.” The agency is taking action because courts have voided all but a sliver of its three previous sets of rules. And President Obama raised the stakes by publicly urging the FCC to impose the “strongest possible rules” on the Internet to fill the regulatory vacuum.
The new net neutrality rules were reportedly formulated by a “shadow FCC” operating within the White House that actively thwarted the will of FCC Chairman Tom Wheeler, the former investor and lobbyist who heads the agency. This has sparked congressional investigations into whether the White House improperly interfered with the FCC, which is officially independent. It has also prompted Congress to hold hearings about a pair of identical “Internet openness” bills in the two chambers that endorse net neutrality but clarify and reduce the FCC’s power and control over the Internet.
Rarely has a wonky feature of telecom policy generated such drama. The FCC rules will certainly be challenged in court, and Congress will just as certainly press forward with its efforts to shield the FCC from White House control. Internet policy will become a prominent issue in the 2016 presidential election, nearly 50 years after the Internet was born.
A brief history of the Internet
The Internet traces its roots to the 1960s, when researchers in America and Britain invented “packet switching,” a nimble method of managing network capacity. The telephone network manages capacity by subdividing the total pool of bandwidth into pieces just large enough to handle a phone call and handing them out on demand. Even in its modern, digital form, the telephone network does pretty much what it did when Lily Tomlin’s Ernestine put through calls with plugs and jacks on a switching panel.
Packet switching does something so radical that most engineers doubted it would even work: It assigns all of its capacity to each user on demand, for short periods of time just long enough for the sender to transmit 1,500 bytes of information or less. This information unit, the packet, contains the addresses of the sender and the recipient, so the network is free to deliver it as it sees fit.
It might subdivide the packet into smaller pieces and send them in parallel over different paths, common practice on the French CYCLADES network that predates the American Internet. Or it could combine packets with other packets and send them over a very high capacity path. It could even send identical copies of the packet over multiple paths at the same time, which would be useful if, for example, a nuclear attack had knocked out portions of the nationwide network. Thus, packet switching replaces telephone operators with computers, harnessing computer power to deliver an unprecedented degree of flexibility and performance.
The first large packet network, the Defense Department’s ARPANet, was deployed in 1969; it became the largest and most important part of the Internet in 1983. The early Internet was restricted to government contractors and research institutions and bound by an “acceptable use policy” forbidding commerce, but the Clinton administration privatized the Internet and opened it to commerce in the mid-1990s.
Because cellular data, digital subscriber line service, cable modems, and wi-fi were invented at roughly the same time as privatization, the essential elements of a communications revolution were in place by the turn of the century.
One network to rule them all
The virtue of packet switching is its ability to support a wide range of computer activities over a variety of media, which is why we can stream TV shows, send email, and surf the Web from laptops, smartphones, and iPads over mobile networks, DSL, cable, or wi-fi without sweating the details. But packet switching does have a drawback. While it can support a wide range of content retrieval activities with a high probability of success, it can be dismal at guaranteeing the rapid delivery of any individual packet. Because each packet takes the network’s entire capacity, packet networks are fully congested for the duration of each packet. If multiple users wish to transmit at roughly the same time, each must take a number and wait for a turn.
Because a movie consists of roughly a million packets over the span of a couple of hours (or more, if the film features Hobbits or conspiracies), the delivery time of each individual movie packet doesn’t matter much. Phone calls or similar applications such as video conferencing can, by contrast, be frustrating over the Internet. Skype and Vonage generally work well, but very frequently users encounter inaudible signals. Internet telephone calls work better on high-speed connections, but even at the unusually high rate of 100 Mbps, the laws of probability occasionally produce moments of chaos. Video streaming services aggravate the problem by delivering packets in clumps rather than with uniform spacing.
Network operators can take affirmative steps to improve the quality of Internet calls without degrading the perceived quality of movies. Because Oliver Stone’s "Nixon" runs three and a half hours, a broadband Internet service can delay the Amazon Instant stream of many of its two million packets by a fraction of a second in order to deliver a Skype packet before it expires, some two-tenths of a second after it was created. Broadband networks handle some packets that must be delivered in tenths of a second to be useful as well as packets that aren’t relevant for minutes or hours. So it is absurd to treat all packets equally.
Inflicting a tiny delay on a movie is of less consequence than, say, erasing part of a conversation. Given a choice, users would certainly like their broadband service providers to do this for them. So why don’t they? You may think the failure of broadband Internet service to meet Skype users’ needs is driven by greed or indolence, which is what critics say. But the answer lies in another quarter altogether.
Regulation vs. innovation
The hallmark of the White House’s broadband Internet service regulations is network neutrality, which is the idea that all packets should be handled equally. While you may prefer your Skype packets to take precedence over your movie packets on the occasions when the two are in conflict, the service provider can’t do this without looking into the packets to see what application created them. As Harvard law professor Lawrence Lessig said in his book Code, the most important rule of net neutrality is that the network must operate blindly, without bias or favor: “Like a daydreaming postal worker, the network simply moves the data and leaves its interpretation to the applications at either end.”
Lessig proteges Tim Wu, coiner of the term “net neutrality," and Barbara van Schewick, the FCC’s most dedicated lobbyist, have modified Lessig’s rule in modest ways, but a prohibition against preferential treatment remains intact in both of their current formulations.
Broadband Internet service providers understandably do not want to offer services that break the law and popular notions of Internet correctness. Some foreign Internet services manage packet streams intelligently, but mainstream U.S. firms have shied away from providing such services since the FCC’s infamous 2008 ruling against Comcast on a complaint brought by Vuze, a video streamer that produces a version of the BitTorrent protocol.
Vuze complained that Comcast was interfering with its users’ abilities to share movies on personal computers on the Comcast network. The accusation was at least partially true, but for reasons the FCC failed to uncover, Comcast reduced the amount of upstream bandwidth BitTorrent clients were able to use in a somewhat crude attempt to deal with problems BitTorrent caused for Vonage users.
Comcast offers a telephone service of its own which is unaffected by Internet applications because it uses a dedicated path from the home to the telephone network. Vonage shares capacity with other Internet applications, and Comcast found itself in a pickle in 2007 when it was deluged with customer service calls from disgruntled Vonage users who suspected Comcast was degrading their service to make them switch to Comcast's voice service.
Comcast knew that interfering with Internet telephony was forbidden since the FCC had already sanctioned Madison River Communications, a North Carolina phone company, for blocking access to Vonage in 2005. The Madison River action had been based on the Internet Policy Statement devised by former Chairman Michael Powell that was never adopted as a formal rule.
When Comcast engineers found that the Vonage complaints were coming from neighborhoods with high BitTorrent use, it reduced the bandwidth BitTorrent could consume by injecting forged packets that told BitTorrent to disconnect some of the dozens of upstream connections it maintained. Comcast didn’t prevent BitTorrent users from sharing their movies with others, but it did reduce the amount of bandwidth they could take. The system employed by Comcast was a stopgap until the deployment of faster, next-generation broadband equipment that had been delayed for reasons over which Comcast had no control.
The FCC ordered Comcast to stop disconnecting BitTorrent streams. The company had already done so before the agency issued the order. But Comcast then sued the FCC and won because the agency's action was predicated on an unpublished rule. Following this court decision and Congress’ refusal to pass a net neutrality bill before the 2010 election, the FCC published its third iteration of Internet regulations, the Open Internet Order of 2010. This was challenged by Verizon and largely voided by the D.C. Circuit in January 2014 on jurisdictional grounds, which brings us to where we are now.
80 years of communications law
The FCC’s mission is to implement the Communications Act of 1934, that defines the general framework of “communication by wire and radio” within the U.S. For historical reasons, the law is organized into several technology-dependent categories, known as “titles."
Title I relates to information services, which are those that store, retrieve or process information — in essence, anything that’s done with a computer. Title II covers the common carrier wired telephone network, Title III is for the mobile voice and data network, and Title VI is cable TV.
There is limited cross-pollination between the titles. Voice services provided over mobile and cable networks are classified as Title II services because they interface with the telephone network, but voice services provided over the Internet are regulated under Title I.
In addition to the six titles, Section 706 of the Act deals with promoting “advanced telecommunications capability,” which is generally taken to mean broadband networks. Section 706 orders the FCC to survey the build-out of broadband networks and take necessary deregulatory actions to speed up deployment if it finds progress too slow.
Congress last updated the law in 1996, when it passed a Telecommunications Act that tried to create competition for local phone service. It failed to achieve its principal objective, but it is most remarkable for its almost complete failure to mention the Internet.
Because there is not a Title VII on regulating the Internet, the FCC has jumped through hoops to find the legal authority to regulate it. When phone companies initially offered DSL service, they volunteered to have this new service regulated under Title II. They gravitated in this direction to secure common carrier immunity from liability for the unlawful behavior of users, such as piracy and libel. Title II also meant making lines available to competitors at prices set by regulators, but this provision didn’t hold up in court.
Cable Internet took a different route. It had been classified by the U.S. Court of Appeals for the Ninth Circuit as a common carrier in 2000 before any FCC ruling on its status. In 2002, however, the FCC issued a ruling that it was actually an information service because Internet services process substantial amounts of information (packet switching is a rich exercise in information processing by itself). The Ninth Circuit wasn’t happy with the FCC’s judgment, but the Supreme Court ruled in 2005 that it was the agency’s call to make.
Justice Antonin Scalia, who had worked on telecommunications policy in the Nixon White House, dissented, much to the glee of today’s leftist net neutrality advocates. But a dissent is a minority opinion.
The 2008 FCC order in the Comcast matter was quickly overturned because the Internet Policy Statement wasn’t an actual regulation. Most of the 2010 Open Internet Order was overturned because it attempted to impose a Title II regulation on a Title I service, the stipulation that broadband service providers “prophylactically” ban the kind of packet discrimination that permits phone services to co-exist peacefully with video streaming.
But the FCC scored a major victory in winning court approval for a creative reading of its Section 706 authority to accelerate the build out of “advanced telecommunication capability” by pre-emptively banning the use and/or sale of network management capabilities that would allow a diverse array of applications to co-exist on common networks. Venturing onto thin ice, the court bought the FCC’s contention that networks should strive to become faster but not smarter.
The tortured path to the 2015 order
The court offered the FCC two paths to a legally valid net neutrality rule. It could rely on its Section 706 authority, as it had in a previous order on mobile data services, or it could attempt to reclassify broadband Internet service under the common carrier title. Classification is not supposed to be an exercise that the commission undertakes with an eye to expanding its own power. The FCC is, rather, supposed to examine the nature of a service and determine where it best fits within the confines of the law. Classification should be a factual judgment, not a policy choice.
The law is actually more complicated than this, because the 1996 Act gave the FCC a special authority known as “forbearance” that allows it to refrain from applying specific portions of an existing title to a new service. Thus, it has a choice between assigning broadband to Title I (information services), Title II (common carrier telephone networks), or to a subset of Title II left over after some portions of the law are excluded by the forbearance power. Some commenters describe Title II with forbearance as a ransom note, since it cuts and pastes parts of the law into an entirely new configuration. But there’s no doubt that Congress did in fact give the FCC this power.
Largely because Title II has spawned so much litigation — there have been literally thousands of cases in the courts and before the FCC — Wheeler was reluctant to apply it to broadband networks. From February to June of last year, his speeches indicated that he would use the FCC’s new Section 706 authority to fashion a third way on net neutrality that would refrain from “prophylactic bans” on smart network management practices in favor of a more permissive approach. He signaled willingness to permit bargaining between content providers such as Netflix and communication apps like Skype and Vonage on the one side and broadband Internet service providers on the other.
Wheeler has been a long-time participant in the Internet policy conferences held at the Silicon Flatirons Center at the University of Colorado Law School in Boulder, arguably the premier venue for Internet policy discourse in America. Tim Wu introduced the notion of net neutrality at a Flatirons event; Powell unveiled his Internet Policy Statement in Boulder; and the center’s director, Phil Weiser, co-wrote a paper laying out a “Third Way on Net Neutrality” after Wu and Powell had expressed their positions. Wheeler’s Notice of Proposed Rulemaking issued last May owes a heavy debt to the Weiser paper.
Through the summer and early autumn it appeared that Wheeler would draft a new Open Internet order relying on Title I and Section 706 consistent with the third way. This would permit bargaining for customized delivery services as long as baseline network quality was sufficient for users to enjoy the benefits of common applications such as web surfing, email, and video streaming. Common applications would need to work well before providers could offer special treatment to leading-edge applications that we’ve yet to see on the Internet, such as immersive video conferencing, high definition voice, and accelerated gaming.
Unfortunately, cracks began to appear in Wheeler’s resolve after he briefed Democrats on Capitol Hill just before the commission’s notice was published. Someone, either an FCC staffer or a member of Congress, leaked a distorted version of the notice to the Wall Street Journal. Press speculation led to the creation of an Internet meme to the effect that the FCC planned to permit Internet service providers to collect ransom from small businesses under threat of relegating them to “slow lanes” if they didn’t pay up. This story proved to be a boon to technically illiterate bloggers and journalists assigned to the Internet beat. Many bloggers announced that the FCC proposed to allow the Internet providers to extort small businesses. Sadly many credulous readers believed them; many filed comments with the FCC protesting what was an entirely fictional plan.
The Netflix saga
The effects of the vindictive leak were exaggerated by a tale of woe that figured heavily in the ginned-up reaction to Wheeler’s plan. A little more than a year ago, Netflix users in many parts of the country experienced a service meltdown. Instead of seeing their movies and TV reruns flow seamlessly across their screens as they always had, they were jarred by frozen images, erratic delays, and disconnections. Netflix had created this problem itself by poor planning and was ultimately bailed out by the large Internet service providers, but not without cash payments and a great deal of complaining.
The underlying cause of the Netflix catastrophe was a hasty attempt to realign the middlemen of a complex entertainment delivery process. Netflix delivers video content to ISPs in “Internet peering centers,” gigantic buildings in major cities where dozens of Internet companies rent floor space in order to exchange packets according to bilateral agreements. These deals are lifesavers, but critics are more likely to describe them as extortion because they’re not aware of the flow of payments between Internet firms. The story is hard to tease out of popular narrative because it centers on a portion of the Internet few understand, even within the FCC.
Netflix was able to overcome the problem by negotiating special deals for its massive traffic load — a third of all traffic on the Internet comes from Netflix — with AT&T, Comcast, and Verizon. But the deals weren’t free, just as the middlemen Netflix eliminated by going direct to the large Internet providers weren’t free, either. The financial details are shrouded by non-disclosure agreements, but most experts estimate that Netflix’s communication costs are now lower than they’ve ever been because the company was able to bypass the middlemen and negotiate its own deals at the source. This is a typical trajectory for successful Internet content merchants.
Apple is undergoing the same transformation today to meet demand for its immensely popular iTunes service, but without any of the drama that colored the Netflix transition. Amazon and YouTube built their own acceleration facilities a decade ago, again without fanfare. These other companies didn’t wait until their traffic loads outstripped the Internet’s capacity before changing their delivery method as Netflix did. The lesson here is to plan for growth instead of lurching from crisis to crisis.
But the Netflix story combined with media distortion of the third way created the appearance that Netflix was being held hostage to the kind of bargaining contemplated by Wheeler’s notice.
The November shellacking
Despite voters' sharp rebuke to President Obama and the Democrats in the 2014 midterms, the president went on the offensive with a slew of orders only a few hours after a chastened talk about bipartisanship. Obama almost appeared to be trolling Republicans in hopes of provoking an over-the-top backlash that would undermine the “party of grownups” image that had been key to the midterm success.
This campaign of provocation culminated in a YouTube video encouraging Wheeler to impose the “strongest possible rules” on the Internet instead of the third way “light touch” he was refining. Wheeler appeared blind-sided by the president’s message, which amounted to an order, and he buckled. Forced to explain his change of position, Wheeler offered two ambiguous stories. The first, presented as an epiphany, stressed the success of mobile networks under Title II and glossed over the fact that mobile data service was classified under Title I by the 1996 Telecommunications Act.
The second story was even more puzzling. Wheeler had been hired by a personal computer company called NABU after IBM had released its PC and Apple had introduced the Macintosh. NABU's PC used an older-generation central processing unit, the eight-bit Z80, and an obsolete operating system that had been the precursor to Microsoft’s PC-DOS. The NABU machine attached to a TV set and to a one-way cable network that continually downloaded an uncontrollable stream of applications and other content. Wheeler negotiated deals with 40 cable networks to support the NABU machine, but none of these placements spurred significant sales because the overall value proposition for the NABU PC was too weak. Like so many others in the early personal computer era, the company went bust. In an op-ed in Wired, Wheeler blamed NABU’s failure on too little network openness rather than too little computer performance.
So, with these flimsy justifications, the US has arrived at a turning point in the development of the web. The present direction of Internet regulation threatens the ability of networks to advance. Supporters of Title II regulation may be well-schooled in telecom law, but few grasp the strengths and weaknesses of Internet technology. The engineering community is concerned about improper regulation, hence luminaries such as the Internet’s lead inventor Bob Kahn and networking pioneer Dave Farber have spoken out on the risk of regulatory overreach. IEEE Spectrum, a publication of the largest professional organization of computer engineers, has just published a detailed critique of the chilling effects of neutrality regulations on the growth of the Internet, “Net Neutrality’s Technical Troubles.” Ironically, the first victims of the White House’s “strongest possible rules” imported from the annals of telephone regulation will be the Internet phone calls carried by Skype and Vonage.
The Internet as we know it today is a distillation of ideas developed around the world by both public and private sector researchers supported by investments from both public and private sources. Most of the magic comes from the profit-driven private sector. As amazing as the Internet is today, it still falls short of its potential.
It’s a certainty that the rules imposed on Wheeler will slow the Internet’s rate of progress and lock in current applications. Congress can take steps to undo the damage about to be inflicted — and it should.Richard Bennett is a co-inventor of wi-fi and a visiting fellow at the American Enterprise Institute.