What the December 14th Net Neutrality Vote really means: answering your questions. All of them.
Network neutrality is often portrayed as being about packets — whether the ISP can deliver some via fast lanes, drop some to block web sites or charge more for some traffic. But the planned reversal of the FCC 2015 Open Internet Report and Order is much more, as the FCC 2017 Restoring Internet Freedom order removes essentially all regulatory oversight by the FCC from broadband Internet access service and leaves only the general consumer protections offered by the FTC Act as well as possible antitrust enforcement. Thus, a better framing is whether consumer internet access service should be subject to any industry-specific regulation at all, for any issue.
This summary tries to capture the major questions related to the 2017 effort, in a “frequently discussed issue” format, including questions beyond the core concerns of network neutrality. Much has been written in the past ten years about this topic, so my effort below attempts to summarize the main issues, with links to some relevant other writing, but omitting a rich body of scholarly work and voluminous FCC filings. I was working at the FCC during both the 2010 and 2015 rule makings and participated in the discussions, but all of the below relies only on public documents.
Is there a definition of network neutrality?
You know it when you don’t see it.
Wu roughly defines it as: “absent evidence of harm to the local network or the interests of other users, broadband carriers should not discriminate in how they treat traffic on their broadband network on the basis of inter-network criteria.”
The “four Internet freedoms” from the 2005 FCC policy statement provide another operational definition:
access the lawful Internet content of their choice;
run applications and use services of their choice, subject to the needs of law enforcement.
connect their choice of legal devices that do not harm the network.
competition among network providers, application and service providers, and content providers.
There is some dispute among network neutrality advocates as to whether all network discrimination should be banned. Wu and others argue for a more differentiated set of criteria that mirror the common carrier traditions in pre-internet telecommunications and other domains such as postal services.
Incentives for provider actions are economic, so any definition has to take both network engineering and economic discrimination into consideration. The initial discussion by Wu (2003) focused on packet-level blocking, but economic discrimination has been seen as a more useful framing, as, among others, Vishal Misra explains.
A “plain vanilla” network that delivers packets best effort and generally avoids artificial bottlenecks, except limited by last-mile capacity and bandwidth tiers, is assumed to be neutral, even though TCP may favor short-haul connections and some applications are more sensitive to certain impairments like packet loss and delay jitter.
While the exceptions and weasel words like “reasonable network management” would seem to make for regulatory uncertainty, in practice issues only arise if a provider explicitly favors or degrades particular applications or advantages their own content or services. FCC Commissioner Clyburn has challenged predictions made by opponents of the 2015 Open Internet order.
Are there hard corner cases?
We’re currently not debating alternatives or “friendly” amendments to network neutrality rules, but it is worth pointing out that there have been ISP practices that are, strictly speaking, non-neutral, but where proponents of network neutrality have been hesitant to squawk too loudly: tethering and commercial use of home internet service. With increasing bandwidth, “unlimited” 4G LTE services could substitute at least for home DSL service, but tethering speed and volume restrictions prevent this. Similarly, fiber connections could easily support a commercial web site, school or a small business, but such use is prohibited even by, say, Google Fiber. This isn’t completely intellectually consistent, but given the content neutrality of these policies, probably a reasonable trade-off. It is unlikely, for example, that the current unlimited plans, generally seen as consumer-friendly, could be offered without tethering restrictions.
Similarly, some port blocking is seen as acceptable if disclosed, with blocking of port 25 (SMTP for email) helpful for spam prevention, but also interfering with running a home email server.
What did the 2015 Open Internet order do?
The 2015 Open Internet order banned blocking of content, services and device attachments, throttling, and paid prioritization:
A person engaged in the provision of broadband Internet access service, insofar as such person is so engaged, shall not block lawful content, applications, services, or nonharmful devices, subject to reasonable network management.
…, shall not impair or degrade lawful Internet traffic on the basis of Internet content, application, or service, or use of a non-harmful device, subject to reasonable network management.
…, shall not engage in paid prioritization.
“Paid prioritization” refers to the management of a broadband provider’s network to directly or indirectly favor some traffic over other traffic, including through use of techniques such as traffic shaping, prioritization, resource reservation, or other forms of preferential traffic management, either (a) in exchange for consideration (monetary or otherwise) from a third party, or (b) to benefit an affiliated entity.
…, shall not unreasonably interfere with or unreasonably disadvantage (i) end users’ ability to select, access, and use broadband Internet access service or the lawful Internet content, applications, services, or devices of their choice, or (ii) edge providers’ ability to make lawful content, applications, services, or devices available to end users. Reasonable network management shall not be considered a violation of this rule.
Was this just about network neutrality?
No, at least not in the narrow sense. The 2015 order provided a framework for resolving future industry disputes and consumer complaints related to internet access services.
Section 208 gave consumers a way of raising complaints with the internet service providers.
The FCC also used the 2015 rules to impose privacy obligations on carriers; these rules were, in early 2017, overturned by Congress. With the 2017 Order, even the generic obligations left in Section 222 will no longer apply. Thus, as long as providers disclose what data they sell to advertisers or data brokers, consumers have no recourse.
Was the first network neutrality attempt in 2015?
No, the FCC has, in fits and starts, made a number of attempts to ensure open access, either by unbundling regulation in the early days of consumer internet services (from 1996 until roughly 2002) or through speeches (Powell, a Republican FCC chair in 2004), policy statements (2005), enforcement actions against Madison River, on VoIP, in 2005, and Comcast, on peer-to-peer traffic in 2007, the 2010 Open Internet rules or the 2015 Open Internet rules. Wu provides an extensive history. Moreover, one can argue that concerns about interconnection and “paid priority” go back to the earliest days of phone service and the emergence of long-distance voice calls, with the 1913 Kingsbury Commitment a well-known example. The Open Internet device attachment rules have a similarly long-ago historical precedent: Connecting non-harmful devices to the telephone network has been allowed by FCC rulings going back to the Hush-A-Phone (1956) and Carterfone (1968) cases, even if it sometimes took court action to get the FCC there.
What does the 2017 Restoring Internet Freedom order do?
In short, it undoes almost all of the 2015 and 2010 orders and re-re-classifies broadband internet access service as an information (“Title I”) service, leaving only some parts of the 2010 transparency rules in place. It also suggests that these non-rules preempt any state-level consumer protection activity related to broadband services. The 2017 rules modifies the 2010 rules slightly to read:
Any person providing broadband Internet access service shall publicly disclose accurate information regarding the network management practices, performance, and commercial terms of its broadband Internet access services sufficient to enable consumers to make informed choices regarding the purchase and use of such services and entrepreneurs and other small businesses to develop, market, and maintain Internet offerings. Such disclosure shall be made via a publicly available, easily accessible website or through transmittal to the Commission.
However, it removes the enhanced 2015 requirements that provided more granular performance measurements.
How does this compare internationally?
As best as I can tell, the United States will be the only OECD country without any real sector-specific telecom regulation and, at the same time, without forced competition through unbundling. Elsewhere, countries that have no strong network neutrality rules typically have unbundling or structural separation rules.
How will competition constrain ISPs?
It would be too simple to claim that US ISPs are completely shielded from competition. The re-introduction of unlimited 4G plans by all four national carriers clearly was caused by competitive pressures, introduced by T-Mobile and Sprint. Similarly, cable operators have continued to increase speeds or introduced, like FiOS, symmetric upload-download speeds in areas with significant telco-cable competition. The rates charged for competitive (telco) gigabit service are significantly lower in areas where Google Fiber is present or building out.
Most residential and business customers have one or maybe two choices for the most popular broadband access speeds. For example, for the cable entry-level speed of 25 megabits per second download, 37% have only zero or one choice (as of June 2016). The provider at that speed is typically the local cable company or, in fewer cases, the successor to Ma Bell. 41% have two choices in their census block and 22% supposedly three or more. (I say “supposedly” since it is unclear who these third providers could be. RCN is one of the larger overbuilders that offers a third option in some urban areas, but it only passes 3.9 million households. In some cases, business-only providers may get counted as offering service to consumers, even though they do not market to consumers and their offerings are priced for enterprise use.) Switching providers requires an excruciating call with the retention department, an early termination fee, a technician visit and possibly running new wiring. In urban areas, even though the census block is nominally served by two or more providers, many apartment and condo buildings typically only offer cable for anything above DSL speeds, so the actual consumer choice is likely even lower.
The contiguous continental areas of the United States are served by Viasat, but it offers only modest competition. The latency constraints of geostationary satellites are well known and make, for example, voice calls between two satellite subscribers painful. Subscribers value high-latency services significantly less than wireline options. However, capacity constraints are likely more serious impediments preventing the large-scale adoption of satellite systems. Viasat is currently capacity-limited to about 500,000 subscribers, and beams in many rural areas are currently not accepting additional subscribers. Prices are high and likely targeted towards areas with no wireline competition. For example, a $75 plan offers “up to” 12 Mb/s download speed and 25 GB a month of data. All available consumer behavior data show that satellite services do not convince customers from cable or better DSL options to switch; it is usually the current carrier of last resort where the traditional regulatory COLR is not offering usable DSL service. Thus, satellite serves a vital role of delivering basic internet connectivity to about 0.5% households, those not served by wireline and WISP options, as well as offering maritime and in-flight connectivity, but is hardly a competitive threat to ISPs offering cable or fiber. Planned satellite options, such as the OneWeb LEO constellation, are also likely to be able to serve only a similarly small number of households with streaming video requirements.
Competition is stronger for mobile services with four large national providers, but the 7 to 15 GB or so of tethered data on “unlimited” plans is no replacement for wired services where the median usage is around 100 GB per month. Two of the largest mobile providers, AT&T and Verizon, are also among the largest residential ISPs, with limited incentive to compete within their own company.
In addition, since the 2010 Open Internet rules, service providers have merged at a furious pace. The top 14 providers in the US now have 95% market share, the top six around 80%.
With unlimited LTE 4G services, households that don’t consume much video may well be able to use only a mobile service with tethering, thus adding competitive options with wireline carriers at the low end. Indeed, even in 2015, 13% of adults had only a smartphone and no home internet.
But, generally, competition for residential US broadband service is modest, particularly compared to other countries that implement some kind of mandatory unbundling.
The presence of competition does not guarantee consumer-friendly outcomes. With four major US airlines, air travel is still miserable, particularly for anybody of above-average height. Despite competition, many smaller airports have lost service. Extrapolated to the network realm, a consumer may well face a situation similar to the current cable TV retransmission disputes, where now parts of the internet fade in and out as conflicts are messily resolved, or not. Switching is largely pointless since one would only trade one unpredictable set of disruptions for another, just like ditching the cable company in favor of Dish or DirecTV does not restore all channels or prevent temporary loss of some TV channels. (The limited windows for movies on streaming video services is also somewhat similar, but users at least have a reasonable option of subscribing to multiple services at the same time— or pirate sites, if so inclined.)
The competition argument is a probabilistic argument and largely addresses only popular content that already has an audience. It is hard for consumers to choose providers based on unseen and unknown future services. Also, the mere presence of competition does not guarantee that a consumer can access any (legal) information — for any number of reasons, given the limited number of providers in a region, they could all decide not to offer access to certain services or content or all charge certain content providers extra, just like almost all airlines charge for checked baggage.
Why the concern about paid prioritization and interconnection?
Traditionally, telco internet service providers offered only limited video services (e.g., for FiOS) and did not own any content of their own. Cable companies have owned some media properties, such as regional sports networks and relatively niche cable networks, but only the NBC Universal acquisition by Comcast in 2009 turned Comcast also into a major content owner. In 2015, the acquisition of DirecTV by AT&T created another major content distributor.
Thus, today five of the top six ISPs either own content companies or are major distributors of cable TV programming. As cord cutting starts to significantly curtail traditional cable and satellite TV subscriptions, these companies have the means, motive and opportunity to try to use their gatekeeper position to extract more revenue from consumers, whether directly through bandwidth caps or indirectly through interconnection fees charged to CDNs or transit providers.
What are the likely consequences of abandoning internet access regulation?
“Follow the money” seems like good advice in predicting what carriers might do. Thus, changes are more likely if they either generate significant new revenue, primarily from third parties, create the ability to reach new customers, monetize consumer information, or protect the (video) services of the carrier against over-the-top competition.
Won’t the ISPs be regulated by the FTC and constrained by anti-trust law?
Others have addressed this topic in depth, but in short: neither anti-trust nor general consumer protection rules are likely to be a replacement for ex-ante rules. The FTC has no rule-making authority and largely restricts itself to pursuing large-scale fraud, given the limited resources it has available. There is also some doubt as to whether the FTC can investigate complaints against any company that still provides some telecommunications services. Thus, unless a carrier employs lawyers incapable of writing terms of service that essentially promise nothing, they are unlikely to be in violation of the FTC Act against unfair and deceptive acts in commerce. For example, blocking a service or content is seen as no different than not stocking a particular item in a store — inconvenient, maybe, but a consumer can’t go and complain to the FTC, as the FTC is generally set up to deal with fraudulent behavior in competitive markets where consumers can be assumed to have other choices. Beyond the local must-carry TV stations, cable subscribers have no regulatory remedy if their cable system does not carry their favorite cable programming or places it into a higher-cost tier.
Anti-trust law is likely (Hal Singer) also a largely impotent remedy except in the most egregious cases of foreclosure, e.g., were a cable company to block Netflix to protect its own video service. Proving consumer harm is difficult in many cases, particularly where the harm is foregone innovation. You cannot see and count the start-ups that never got founded and funded since the investors considered risking the wrath of major ISPs, although Joshua Wright’s article cites some examples. Waiting for the resolution of an anti-trust case is unlikely to be an option for a startup or any small company, even if the new entrant prevails, as the Microsoft — Netscape case showed. (Unless, of course, your business model is to be a not-for-profit afterwards.) Unlike in the merger case, where the merger parties have every interest in moving things along as quickly as possible, here the incentives are reversed: The company accused of untoward behavior has every incentive to drag out the process and can appeal any adverse FTC action through the legal process all the way to the Supreme Court.
The gateway function that Microsoft occupied in the 1990s offers a reasonable parallel, where start-ups generally tried to stay out of any area either already occupied by Microsoft or requiring the cooperation of Microsoft through its desktop operating system monopoly, except that it is much less likely that some other technology will replace wireline or mobile broadband. As noted by Singer, antitrust also relies on the FTC to take the initiative. With the FCC Open Internet framework, there is both an informal and formal complaint process where the FCC Enforcement Bureau has to take action within a reasonably short amount of time, even if this is sometimes not observed.
Thus, advocates of using anti-trust law ask others to rely on a mechanism that hasn’t worked particularly well in the technology domain, has high uncertainty, little predictability and clearly favors the incumbent.
As Harold Feld points out, advocates of dismantling the rules also generally do not consider any of the commonly-cited ISP non-neutral behavior to be problematic, beyond possible lack of disclosure by Comcast in the peer-to-peer blocking case, so it is unclear what kind of ISP behavior should, in their view, trigger FTC or anti-trust scrutiny.
Is anything drastic going to change on December 15?
It is unlikely that ISPs will dramatically change their business practices immediately after the FCC vote. They will not introduce practices that are widely seen as anti-consumer or as blatantly blocking competitors. Among other reasons, the reversal of the 2015 Open Internet rules will likely immediately be contested in federal court, so offering I-told-you-so arguments to the plaintiff and various friends-of-the-court seems unwise. It is also plausible that any change in business practices may be delayed until after the November 2018 election, just in case the House or Senate flips. There is precedent: after cable rate deregulation, some cable companies got greedy, leading to swift re-imposition of some rules.
Did internet investment decrease after the Open Internet rules?
You are likely to get different answers depending on how the question is phrased:
- Do you count investments only by internet access providers, by other internet providers such as long-haul fiber providers, or other companies that strongly depend on internet services, such as the various platform companies?
- Do you count investment in content, given that services like Netflix invest in original content partially because more people have high-speed internet access?
- Do you count only hardware purchases, or also investment in R&D, as many of the platform companies do not purchase much hardware, with R&D having a similar investment role?
- How do you account for recessions and other business cycles that affect the overall economy?
- Do you count investment since the “modern” era of Open Internet rules (2010) or just after the 2015 rules took effect?
- What kind of investments count? For example, cable companies spend a large fraction of the capital investment on video set-top boxes for video services not directly affected by broadband regulation.
- Do purchases of spectrum for mobile services count as capital investments?
- Generally, routers, customer premises equipment and radio access network equipment has decreased in cost over time, at least for the same performance, so do you count absolute dollar values or discount for inflation for the sector?
Also, it has been argued that carrier investment changes quite significantly year-over-year, for any number of reasons, with relatively modest regulatory changes as one of the less-important factors. Decisions by a single company can skew the overall total investment, given that a handful of companies dominate the industry. Companies change investments because they build out new technologies, such as 4G/LTE or DOCSIS 3, in bursts, or they invest by purchasing spectrum, with FCC auctions held only every few years and uneven participation. Companies may hold off on capital investment if they hope to be acquired or hope to merge, or if their balance sheets is loaded down with too much debt already.
In short, the FCC claims that the 2015 rules have harmed investment by internet service providers, while others either dispute the claim, refer to industry statements that do not make the connection to any regulatory changes, opine that short-term variations in investment are meaningless or that investment across a broader section of the ecosystem should be considered.
Will investment increase after the FCC decision?
Similarly unlikely, but there’s a wrinkle due to the timing of technology generations. While the LTE/4G build-out was largely completed in 2015, some carriers will start to roll out 5G (or something their marketing department can label as 5G) in 2018 and 2019. Similarly, AT&T and Charter are obligated by prior merger conditions to build out fiber. Finally, Sprint, after failing to come to a merger agreement with T-Mobile, has recognized that they either need to invest in their network or are likely to continue to lose customers, and have announced a significant increase in capital expenditures. FCC auctions for millimeter-wave spectrum may also occur in 2018 or 2019. Thus, for reasons objectively unrelated to any change in the rules, ISPs may well invest more, but it will provide a convenient “I told you so” narrative.
In the past, some carrier behavior treated as potential neutrality violations could often also be seen as a way to avoid or delay investment. For example, Comcast blocking BitTorrent, AT&T blocking FaceTime or major carriers not offering sufficient interconnect capacity to transit providers can be partially explained by not wanting to spend money on upgrading services. Thus, if selective treatment of high-bandwidth applications becomes allowable, carriers may decide to invest less.
Will the abandonment of regulation foster rural broadband deployment?
There could be two possible impacts on small rural carriers: decreased compliance cost, primarily by reducing the need to disclose network performance, and the possibility for additional differentiated services. Small and mid-sized providers are already large exempted from detailed network performance measurement obligations, so reducing that burden further seems to provide limited upside.
Small rural providers have traditionally relied on intercarrier compensation for voice calls terminated into their network, where large long-distance carriers would pay per-minute charges to the rural carrier. This model is fading, but some rural carriers as well as the European Telecommunication Network Operators Association have proposed that content providers pay volume-based charges to access ISPs. Without a regulatory mandate, it seems unlikely that individual small carriers would have the negotiating leverage to make Netflix pay. (Any collective action of the carriers as a cartel would likely trigger anti-trust concerns.) The per-subscriber profit of providers such as Netflix is also modest — for example, for 3Q2017, the global profit per user was approximately $0.39/month. Thus, even in the unlikely event that Netflix, with roughly 40% of downstream traffic, were to give up all of its profits to the ISP, the contribution to the ISP’s monthly revenue would be modest. Naturally, no ISP would refuse the “free” money, either.
Will we see peering and interconnection disputes again?
Possibly — this is likely one of the least predictable outcomes, as such disputes often only affect a subset of carriers. The transit market has been decreasing in importance, as most traffic is now delivered directly to the “eyeball” ISP via CDNs. Some CDNs have been paying access charges to at least some ISPs for a while.
If large providers were to try to reinstate or increase termination charges, this would actually be a reversal of their position in the voice telephony world. For example, Verizon has, in the intercarrier compensation reform debate, championed bill-and-keep, i.e., zero intercarrier compensation.
Will competition increase or decrease?
It is hard to imagine a fundamental shift in the intensity of competition for broadband access, as competition in that market is driven by the fundamental difficulty of overbuilding existing providers and the barriers they can erect to new entrants. There might be one somewhat unexpected outcome for providers that are not also MVPDs (“cable” video), as they may lose their right to access poles as telecommunication providers. Since many smaller providers, including Google Fiber, are now abandoning trying to deliver “cable TV”, due to the small margins for smaller providers, such providers may no longer be able to obtain the statutory rates guaranteed to telecom providers. (I am no expert in pole attachment law, so ISPs may lose their Title II status and still retain those rights somehow. This seems more likely for providers that also offer business telecommunication services.)
Ben Thompson, Vishal Misra and other have made the point, with some justification, that we should strive for more competition in residential broadband, which would either discourage bad behavior or at least give consumers a choice to go elsewhere. However, these voices rarely lay out a realistic path to get there. While there have been some new entrants in wireline broadband, such as Ting, Sonic and Google Fiber, there have probably been more articles in the tech press about them than they have customers. Google Fiber, in particular, has found that overbuilding is expensive and can be made much more expensive by the incumbents, by refusing to grant pole access on reasonable terms or by discouraging landlords from allowing in-building competition. Since any new fiber-based entrant is unlikely to attempt deployment in more than a handful of locations at a time, incumbents can simply offer attractive long-term contracts to their existing customers. The pattern is well known to start-up airlines: If they can get gates, the incumbent will temporarily lower the fares to the few destinations the new arrival flies to, just long enough until the fledgling start-up runs out of cash.
Curiously, as mentioned by Mike Masnick, the Title II approach disliked by network neutrality opponents actually offers a legal basis, Section 251(3), that was designed by Congress to make local exchange carriers, i.e., the local phone company, make their copper loops available to new entrants, as one of the unbundled network elements (UNE). Indeed, early on, there were a number of providers, Earthlink most prominent among them, that used that route to offer Internet access services. Megapath still offers DSL for business customers on that basis. Requiring offering wholesale copper loops, at regulated prices, is probably the main reason European residential internet access is much cheaper than in the US. (There are trade-offs that are beyond the scope of this discussion.) However, the 2015 Open Internet order explicitly forebore from that obligation. How likely would incumbent carriers that object to the 2015 Open Internet obligations willingly consent to much more economically salient infrastructure sharing?
What about telemedicine?
It has been claimed that network neutrality prevents the delivery of guaranteed-quality services that might be desirable for, say, video-based telemedicine to consumers. (Business services, such as video delivery from a rural health clinic to a regional hospital, are not subject to the Open Internet rules.) First, given the absence of technical means of service differentiation both within and particularly across networks, video telemedicine is still being deployed, albeit hampered by lack of broadband availability in rural areas. It remains unclear how the current rules would prevent such services, as they would likely be specialized services rather than generic internet access services, as pointed out in Footnote 315 in the 2015 Order.
As discussed elsewhere, the rules do not seem to prevent different service quality subscriptions, since the “no paid prioritization” rule only applies to third parties. Thus, if a QoS-enabled delivery were necessary and technically available, the end user could, possibly supported by insurance reimbursement, order such as service. Footnote 22 also envisions a waiver process. (In my personal opinion, a neutral service available only to external parties and on clear conditions does not strike me as fundamentally unsound, but nobody, either in the US or elsewhere, has offered such a consumer service, so models are lacking. Part of the reason is that QoS is difficult to market — most modern wireline networks that are well-run have near-zero queuing delay and near-zero packet loss nearly all the time, so there’s an incentive to only pay the extra-QoS fee under unusual circumstances, making providing the service uneconomical. In LTE networks, the QCI mechanism governs the trade-off between bit errors and delay and thus is more generally useful, but no mobile operator anywhere seems to offer this service to third parties.)
Are ISPs likely to block web sites?
No. Blocking individual sites is unlikely to produce any additional revenue or strengthen the competitive position of ISPs, but will create significant public blowback. Major ISPs have publicly committed to not blocking content, but have been more circumspect about paid prioritization.
However, there is a, as of now, somewhat far-fetched scenario that could occur: An ISP may feel pressured or want to curry favor with the current administration to block “fake news”. Under the 2015 rules, the ISP could respond with “we’d love to help you, Mr. President, but we are common carriers and this would be against the law.” The abolishment of the Open Internet rules will remove any such excuse. Similar “we are a common carrier and we cannot do this” retorts will also no longer hold water if an entity alleges copyright violations and requests that certain IP addresses be blocked. Instead of going after domain names, as for Sci-Hub, publishers may ask ISPs to directly block the IP addresses that they believe hosts copyright-violating content.
Will there be slow lanes?
Not directly, most likely. Based on the interconnection disputes in 2013 and 2014, not upgrading interconnection links unless the CDN or transit provider pays the ISP is much more likely. Such payments, albeit in relatively small amounts, are already taking place. In some such disputes, it’s hard to attribute blame, as who-should-pay-whom peering disputes have occurred for decades and sudden shifts in traffic from one provider to another can tax even the most neutral ISP. However, the abandonment of rules makes it impossible for the FCC to step in and investigate, potentially leading to the same kind of stalemates we see for retransmission disputes, except that they are likely also to affect services that just happen to share the same transit connection or CDN.
What are the most likely new services to be offered?
Cable and fiber ISPs have the problem that most users are not willing to pay much more for gigabit or even 100 Mb/s bandwidth compared to 25/3 Mb/s. Lower bandwidths than 25/3 make for unhappy customers — kind of like the equivalent of the no-frills airfares that provide middle seats and no carry-ons. Also, broadband prices have probably reached a level that they are unaffordable to a significant fraction of the population. (The fraction of households subscribing to wireline Internet service is plateauing at around 70% of adults or 85% of households.) Thus, the classic model of service differentiation by speed tiers is likely to be more difficult to implement. This leaves a number of options:
- Tier by bandwidth cap, but exempt services offered by the provider (zero-rating). To a limited extent, this model is being implemented, but tends to affect only a small percentage of users. If usage volumes cluster around a median, it is hard to pick a bandwidth cap that doesn’t just raise the price for everyone. Since the current price is likely close to the monopoly price in many places, this would be sub-optimal. Comcast has exempted ISP-owned services if streamed to an Xbox, by having these services fall into the exempt category of “specialized services”. Almost all mobile providers have implemented a variety of zero-rating services, with different degrees of neutrality. Evidence shows that participating in zero-rating can significantly boost viewership, although the advent of “unlimited” plans has made zero-rating less salient, with a transition to more indirect attempts to limit video traffic by capping video stream bandwidth.
- Offer additional services, such as home security. Comcast and others have offered such services. However, this would not have been affected by the Open Internet rules — either these are over-the-top services or specialized services.
- Try to slow cord cutting. The two most promising approaches would be to limit performance of CDN interconnections so that HD (4K or 8K) becomes infeasible, streaming becomes unreliable (“buffering”) or to make third party streaming services economically unattractive by charging hefty interconnect fees. This is not without reputational risk and would repeat the 2013/2014 interconnection disputes between the largest ISPs and Netflix.
- Emulate mobile plans, where the basic “unlimited” plan caps video resolution. This would likely require changes in how services are advertised. Mobile services do not advertise raw speeds, so mobile providers can more readily implement service-differentiated speeds.
- Pay for privacy. Earlier, AT&T and Verizon (“super cookies”) had implemented different ways to deliver anonymized traffic data to third parties. AT&T’s initially allowed subscribers to opt out for a fee, but later abandoned the practice.
- Offer higher-quality (“QoS”) services. Although some network neutrality advocates seem opposed to any service differentiation, QoS differentiation would most likely have been acceptable under the 2010 or 2015 rules, as long as these services were application-neutral or otherwise available to providers without discrimination. For example, allowing consumers to purchase a certain number of voice-quality or low-latency-gaming GB of traffic does not seem to violate many notions of network neutrality as long as any service provider could avail themselves of that quota with user permission. For example, §201 of the Communications Act explicitly mentions “different charges for different classes of communications.” (Some would object that such higher-priced services, even if available on non-discriminatory terms, would disadvantage new entrants.) However, the absence of rules would allow special arrangements or, absent vigorous anti-trust enforcement, rates that implicitly favor provision to services operated by the ISP. The classical problem, known as margin squeeze, with charges in vertically integrated providers is that the charge can be sufficiently high to deter external use, but since it is just shuffling money around within the vertically-integrated company, does not harm the provider itself. Given that video and VoIP work pretty well on higher-bandwidth best effort services, low-latency gaming might offer a niche service, but seems unlikely to be a major source of additional revenue.
But the Frightful Five are worse!
A more recent argument, fueled by increasing concerns about the power of the five largest internet platforms, Alphabet (Google), Amazon, Apple, Facebook and Microsoft, and concerns about the ability of other important platforms such as Twitter and Cloudflare to favor or suppress controversial speech, is a new variation of the “don’t regulate me, regulate the one behind the tree” approach. (This has a long tradition — whenever the TV station owners are accused of abusing their retransmission consent market power, they immediately point to the unpopular cable companies and suggest that they should fix their customer service first.)
But abandoning network neutrality is most likely to strengthen, not weaken, the position of these platform companies. They are the only ones likely to negotiate favorable deals with ISPs as they can credibly threaten to withhold content and applications that consumers know and value. We are seeing variations of this script play out in all platform markets, where there are intermediaries that depend on others to deliver services. For example, Internet service providers such as AT&T and Comcast buy up or attempt to buy up content companies or large content-distribution companies such as DirecTV so that they can negotiate lower prices for network TV retransmission fees. Small, rural cable companies pay far more for accessing network TV channels than the largest cable operators. In turn, TV stations that were largely locally owned become part of large national chains like the Sinclair Broadcast Group. In health care, local hospitals become part of chains to improve their negotiation position against a decreasing number of insurance companies. In all of these markets, it is hard to argue that consumers are the winners. If carriers freed from regulation can charge access fees, smaller companies will likely have to go through one of the large service companies to obtain access, ensure quality and get favorable rates. The “facilitators” will naturally collect a fee, particularly if the smaller company might potentially compete with their own current or future offerings. Thus, even if one believes that the platform companies should be subject to tighter regulation or anti-trust remedies, abandoning network neutrality is likely to further strengthen their position.
Internet access is not a telecommunication service!
Opponents of internet access regulation sometimes use engineering arguments to make the point that internet access is an information service, not a telecommunication service. According to 47 U.S.C. § 153(50), a telecommunication service “is the transmission, between and among points specified by the user, of information of the user’s choosing, without change in the form or content of the information as sent and received.” This pretty much defines the Internet Protocol (IP) — the end system provides the destination address drawn from the set of global IPv4 or IPv6 addresses and a well-functioning ISP will deliver the packet to that address. Any modification of the content will lead the receiver to discard the packet. Indeed, this delivery is typically much more faithful than the classical telecommunications networks that delivered analog and digital voice services. Digital voice services would sometimes twiddle bits to embed signaling information (“robbed bit signaling”), would translate audio encodings between different national systems (A-law in Europe and Japan, µ-law in North America) or between the compression used by mobile telephony systems and landline systems. Analog systems modified content even more, whether inadvertently due to transmission noise or to make voice sound better (AT&T TrueVoice in 1994).
Information services, from a user perspective, generally create or host content. Indeed, in the early days of home internet services, AOL and CompuServe offered both local services such as newspaper content or discussion forums and access to the fledgling internet, and services like @ home combined cable access with search engines and portals.
This classification is not really contentious anywhere outside the US, even if the terminology differs somewhat. While there is some discussion as to whether certain messaging services should be communication services, no regulatory agency else treats IP packet delivery as the same regulatory class as search engines, e-commerce sites or newspaper web sites.
One can discuss whether pure store-and-forward services like email, SMS or messaging services should be considered telecommunication or information services, as they are meant to deliver email messages unchanged to the destination chosen by the user, but this discussion can safely be left to a telecommunication law article.
Filing by a group of engineers and Scott Jordan make this point in more depth, with a reply comment by Richard Bennett asserting that the internet is more similar to information services than telecommunication services.
What about DNS?
The Domain Name System (DNS) translates domain names such as example.com to IP addresses. The actual translation is performed by the server operated on behalf of the domain name owner. For example, if you buy a domain name at Godaddy, that company will offer to host the translation service for you and the translation will be specified by the domain owner, not the ISP or even the domain registrar. The ISP operates “caching resolvers” that first check whether somebody on their network has already recently asked for this information. If not, they make the query on behalf of the program needing the information. A well-functioning ISP will simply relay the information in the origin server, without any content alteration. It is not providing information on its own, except for the domains, such as comcast.com, that it operates itself.
Internet service providers deliver packets based on their IP address, not the domain name, even if users or application internally may use such names. Mapping domain names to IP addresses is a completely separate service, and the provider generally doesn’t even know that a particular DNS request later leads to a particular traffic flow. (In that, the model differs from the telephony databases, where database lookups are an integral part of the call setup.)
DNS is functionally similar to directory and database services found in classical phone networks. The closest analogue is probably the number translation offered for toll-free (800) numbers, where the caller provides the number, and the carrier then consults a database to translate the number to a regular phone number. The mapping was, as for DNS, provided by a third-party, namely the carrier serving that toll-free number, typically through a third-party database provider.
The classical digital voice network incorporates numerous such number translation services, with databases for local number portability, operated by Neustar or iconnectiv, being consulted for nearly every call.
In numerous rulings, the FCC has declared such services to be “adjunct services” “if that option or feature is clearly basic in purpose and use and brings maximum benefit to the public through its incorporation in the network.” If provided by the ISP, operating a DNS resolver clearly fits that definition.
Unlike in the classical voice network, where the subscriber had no choice whether to use the toll-free database chosen by the carrier, users can readily reconfigure their end system to use a third-party DNS service, such as the well-known 126.96.36.199 DNS resolution service operated by Google. It would seem odd if an ISP that operates a DNS resolver is an information service, but one that tells their subscribers to use a third-party service is a telecommunication service.
But caching is an information service!
Caching temporarily stores web content on a server closer to the requesting party, so that the content is delivered faster and without consuming long-haul network bandwidth. Caching services, often now labeled as content delivery networks (CDNs), are operated by third-party service providers such as Amazon AWS, Akamai or CloudFlare or by content providers on their own behalf, such as Apple, Google or Netflix. Some ISPs host CDN servers in their network, but they are then operated by the CDN provider. When retrieving a video segment, DNS will point the video player to the IP address of the closest local CDN server, rather than the master (“origin”) server.
If an ISP were to offer such CDN services for sale to content customers, this would be a separate (information) service, just like a cable company can offer web hosting or their own content.
Mostly, when caching is discussed in the context of classification, it is likely that so-called transparent caches are meant. These caches are not explicitly addressed by either the content provider or the user, but rather silently insert themselves into the HTTP setup. They became popular for enterprise networks when network traffic was dominated by images embedded in web pages. But they have generally fallen out of favor since they turned out not be quite as transparent as promised: they do not work with encrypted connections, web cookies or copy-or-access-protected content and may interfere with delivering targeted ads or making sure that ad impressions or page views are counted properly. (Indeed, it is likely that content providers have migrated to HTTPS in part to avoid these problems.)
Regardless of the mechanism, caching servers are meant to be invisible to the user and just optimize the operation of the telecommunication network, either for the service provider, by reducing transit costs, or the end user, by improving performance.
If the act of storing information itself turns a service into an information service, no packet service could ever be a telecommunication service as every router stores packets in its memory, whether for nanoseconds or, for slow networks, for half a second or more. After all, the Internet is a store-and-forward network. However, business packet services such as MPLS are non-controversially telecommunication services. From a technical perspective, IP can run over MPLS, adding no substantial functionality beyond a global address space and checksums.
Again, it would seem odd that if a service provider were to change classification simply by optimizing network performance or removing user-invisible caching services.
The Communication Act is outdated and doesn’t fit the internet!
Clearly, the drafters of the Telecommunications Act, whether in 1934 or 1996, could not anticipate how the internet would be organized or operate, although this was more of a lack of attention than lack of reasonable predictions in 1996. But in the US legal tradition, we routinely use old laws for technologies that did not exist when the law was created, using courts and regulatory agencies to color in the details until the law can be updated to better match current reality. It would be nice if Congress could update the Communication Act to remove references to telegraphs and pay phones and maybe focus on key bottleneck facilities, but there is no requirement to abandon enforcement of the basic sentiments of open access and interconnection, for example, until Congress gets its new Act together. The House voted 414 to 16, the Senate 91 to 5 on what seems now quite “heavy handed” regulation; Republicans, in particular, are unlikely to want to restrain carriers at all. (After all, if they did, they could do so today, either at the FCC or in Congress, even without any Democratic votes.)
The incoherence of being pro-network neutrality, but against any realistic path to get it
Libertarians often simply consider network neutrality a bad idea, e.g., because broadband providers should just be allowed to do whatever maximizes their profit. Some believe that the First Amendment gives ISPs an unfettered right to determine what is carried on their networks. Since they have few data points to offer where this has worked out well for consumers, this is more of a legalistic, philosophical or faith-based argument, but it’s at least true to a set of definable principles and deserves credit for honesty.
But there are also those who profess to believe in at least some of the principles of network neutrality and just believe that the Title II approach is too heavy-handed. Unfortunately, we seem to have tried every other approach within the current Communication Act framework — policy statement, Section 706, Title II, and all but the last have been rejected by the courts as insufficient ground. Clearly, rules other than those in the 2015 Order are possible within the existing Title II framework, but the current real options are variations of the 2015 Order or hoping that providers self-interest, the FTC and antitrust law will protect consumers. Unless the court overturns the 2017 Order, US consumers will be the, largely unenthusiastic, subjects of this experiment.
Anything I can do about this?
Since you obviously care enough to read a few thousand words on communication policy, you might want to do something to limit the damage. First, the usual advice of submitting short comments to the FCC is unfortunately not at all helpful. Such comments, for any rulemaking, are only counted if they support the chairman’s position. The current chairman has indicated that only substantive comments count, but it’s too late for those, as the comment period ended a long time ago.
Your best option is to contact your local representative or senator, particularly if they are Republican. If you want to go pro, Indivisible has a handbook. Beyond registering your opinion by phone or email, consider whether your local representative or senator particularly cares about an issue that might be affected by the abandonment of internet rules. Is there local industry that may suffer? Does the member of Congress care about competition and small businesses? Do you belong to a professional society or trade association that can state its opinion? Do you special expertise, e.g., as an economist or engineer, that makes your opinion stand out?
As the 2018 election approaches, find out how your local candidates think about telecom policy in general. Do they support pro-competitive telecom policies such as network neutrality or networks built by non-traditional providers? Ask about these issues at community forums, as that indicates to the candidate that voters care.
While December 14 is a milestone, it’s not the end of the road.