>>> 2021-01-29 broadband-ish competition-ish (PDF)

There are a number of technical and regulatory aspects which make the discussion around broadband availability in the United States rather confusing. The real problem is that this confusion of conversation leads to confusion of public policy---a result that is easily exploited by the industry to avoid serious and effective policy action to improve the situation.

Well, maybe that should be the end of my opinions on telecom regulation for the day, but alas, it's not.

One of the really strange parts of the broadband internet debate is the huge inconsistency in policy applied to different types of internet carriers. The fundamental reason for this is simple: some people obtain their internet from a telephone companies. Telephone companies, which are more or less all descended from AT&T[1], have directly inherited the legacy of AT&T's longstanding monopoly. An important aspect of this legacy is a strong system of government oversight of many aspects of telephone carrier operations, from tariffs[2] to technical architecture.

Some people, on the other hand, obtain their internet service from the cable company. Cable companies have historically been a far more competitive industry and so, while they do face some regulation, that regulation is far more lax than that faced by telephone companies.

The core of this distinction today is the concept of a "common carrier," which often came up in net neutrality debates. Put very simply, common carriers are required to offer service to the public on an open basis and subject to a tariff. Telephone companies are common carriers, but with limited exceptions, cable companies are not.

Further complicating the whole thing, internet service providers are or are not common carriers depending on which year you are thinking about. A big part of the ongoing net neutrality debate has been reclassification of ISPs in and out of common carrier status. See how this is all a bit confusing? But that's not even what I meant to talk about here, so let's get back to the history part.

Because telephone companies are common carriers that exist under the weight of AT&T's history, they are subject to many provisions of the Telecommunications Act of 1996. Among the contents of that Act is a significant effort to improve competition in the telecom industry by lowering the barrier to entry for new competitors.

This seems like a noble goal, as we know that one of the major problems with broadband internet availability is the very high cost involved in establishing an ISP, due to the need to build extensive outside plant (wiring, amplifiers, etc). So how does the Act attempt to resolve this issue? by introducing a mandate for local loop unbundling.

Local loop unbundling is not necessarily intended as internet policy, but rather as telephone policy. After the breakup of the Bell System, the long-distance industry had become far more competitive due to the mandate that local telephone carriers (typically the regional Bell operating company or RBOC) allow consumers equal access to non-AT&T interexchange carriers such as MCI and Verizon, typically through the use of "feature group D" capabilities such as the "dial-around code." Local loop unbundling was an effort to do the same thing for the local carrier: allow a consumer to (relatively) easily choose their local carrier from one of many options.

The problem, of course, is that there's only one local carrier with actual wires connecting to that consumer, typically the incumbent local exchange carrier (ILEC), which is typically an RBOC. The solution: require that all telephone carriers with outside plant allow other carriers to make use of their outside plant at a wholesale rate.

This is very similar to the setup used for electrical service in some parts of the US and, to my understanding, all of the UK. There is a company which operates the actual physical infrastructure, and there is a company that you buy your electricity from. They are not necessarily the same company; if they are not, you pay the infrastructure operator for only the cost of maintaining the infrastructure.

Similarly: CenturyLink owns the physical outside plant. You can use CenturyLink as your phone carrier, but you can also select a different carrier. If you do, you will have to pay CenturyLink for just the cost of their physical plant. This is called local loop unbundling, because it... unbundles the local loops? I'll be honest, the term has never completely made sense to me, but it might help to know that "local loop" is the telephone equivalent of "last mile," roughly meaning the wiring from the exchange to your house.

So this is a telephone issue, and to be honest it was never especially successful in the telephone world as the carriers with the plant were usually much larger and better resourced, and even though their wholesale rates for use of their plant were ostensibly regulated they tended to be high enough that alternate carriers could not be cost-competitive. So where does the internet come in?

Because the policy also applies to internet. The DSLAM (DSL access multiplexer, the thing your DSL modem talks to) is considered part of the local loop, and is similarly unbundled. This means that it is possible to obtain DSL internet from a company other than your telephone carrier. If you do, in addition to paying them for their service, you will also need to pay your telephone carrier for the use of their physical plant. On the backend, the telephone company delivers your packets (well, ATM cells) to your selected ISP instead of to their own infrastructure, so their involvement more or less ends at the DSLAM, although typically, as I understand it, they are also providing wholesale ATM or IP transit to get your internet traffic to a more central and convenient location for your ISP (e.g. an internet exchange). In smaller markets, though, your ISP may be leasing colo space directly in the telephone exchange in order to receive your traffic.

Doesn't this just sound great? It's almost exactly what many fiber-to-the-home advocates have called for, a network that is provided on an open-access basis to an ecosystem of competitive ISPs. In practice, that network is even mostly fiber. What a world!

Obviously, it does not actually work.

There are a few problems. One of them is the same that we saw with unbundled telephone service: in reality, it is usually not possible for a competitive carrier to actually compete with the incumbent one. Imagine the proposition to consumers for those competitive ISPs: Instead of paying CenturyLink for your internet, you could pay CenturyLink and also us! The wholesale rates set by the telephone carrier combined with the smaller size of the competitive ISPs meant that rates were virtually always higher, and it's very hard for ISPs to offer differentiated services in a way that consumers really care about.

In the small town that I used to live in, Socorro, a local competitive ISP called SDC provided DSL on exactly this basis. A quick review shows that it is still true today that SDC costs at least $10 more for service that is at least 50% slower than CenturyLink. For example, $28 to CenturyLink and $25 to SDC for 1.5mbps downstream[3]. I don't necessarily intend to damn the whole industry based on the example of SDC which seems to be a very small operation that has done very little to modernize their operation in at least a decade, but they are surprisingly typical of competitive unbundled DSL carriers.

Another major issue, perhaps the biggest, is that DSL is simply increasingly uncompetitive. DOCSIS (cable internet) is capable of far greater bandwidth than DSL, so it's uncommon for DSL to be a seriously competitive option outside of low price brackets (Major telephone carriers are good at offering rock-bottom rates, competitive DSL ISPs are not) and rural areas.

These two combined basically eliminate local loop unbundling as a serious factor in the broadband industry, leaving is in the same position we were in before: you can only choose from one of the very small number of ISPs that have physical plant in your area. Typically, two.

This problem is exacerbated significantly by the fact that there is no local loop unbundling mandate for cable carriers, so this kind of competitive access is limited to the less capable DSL.

Could the situation be improved by mandating competitive wholesale access to cable infrastructure? Maybe, but I'm skeptical. First, the cable industry has its own interesting and complex history that means that this policy would not be simple to implement. Second, unbundled cable service would likely have the exact same problem of the incumbent cable provider having a far greater ability to attract customers and compete on price than any upstart competitive carriers.

I am further skeptical of the concept of open-access fiber networks for similar reasons. Public ownership of the means of production, err, physical plant would reduce perverse incentives to keep wholesale costs high, but would not eliminate the basic problem of it being generally more efficient and less expensive to provide physical plant and internet service as a vertically integrated operation. I'm not at all opposed to the concept of municipal internet infrastructure, but I think that it needs to be offered on the same vertically-integrated basis as all successful commercial internet service.

Postscript: I am starting to prepare a video about environmental remediation sites at Kirtland Air Force Base. If you find military and atomic history interesting and can stand repeated mention of the vadose zone, you might enjoy it. I'll post it here when it's done. I am hoping this will be the start of a series on the history of Kirtland AFB which is both a place I've spent a lot of time and a parcel of land with a surprisingly long and interesting history, with military R&D going back to the turn of the 20th century.

[1] More accurately, most telephone companies today are either part of a remarkably overgrown competitive (non-AT&T) carrier once called Oak Ridge Telephone and now called CenturyLink, or, confusingly, regional bell operating company Southwestern Bell, which later bought AT&T and renamed itself to AT&T. The fact that one of the companies that AT&T was forced to divest in order to reduce its monopoly later outgrew it, bought it, and now wears its skin is one of the many surprisingly amusing aspects of telecom regulation.

[2] I am not entirely sure of the etymology here, but this is a less common use of the word "tariff" to refer to the regulator-approved price list that utilities use as the basis of their billing. In highly regulated utilities like telephone service, all customer bills must be calculated based on the tariff, and any changes to the tariff require approval of the regulator.

[3] For laughs, the same company also offers wireless internet service at similarly uncompetitive rates. Their website is confusing on the technical points, it seems they finally upgraded from 802.11b but parts of their website still list early-'00s vintage Cisco 802.11b equipment as standard CPE.