A Disturbing View of Future Cable Broadband

There was a recent article in FierceTelecom that quotes a leading cable company consultant as saying that cable companies are not likely to universally upgrade broadband networks in the future. The consultant is Sean McDevitt, a partner at Arthur D. Little, a consulting firm that largely works for the giant ISPs.

In the past, when a cable company migrated from DOCSIS 1.0, to 2.0, and to 3.0 everybody in a community was upgraded to the latest technology. He says going forward that it’s almost certain that there will not be across-the-board upgrades. He says there will be neighborhoods upgraded to fiber, neighborhoods migrated to the next-generation DOCSIS 4.0, and neighborhoods that will see no upgrades at all.

I label this as disturbing because it raises the possibility of digital redlining in communities if cable companies start upgrading selectively. The upgrades to fiber or DOCSIS 4.0 will bring faster upload speeds to match fast download speeds. Any neighborhood left on the current DOCSIS 3.1 platform will be stuck with inadequate upload speeds, which was the predominant problem with cable broadband during the pandemic and continues to be a problem for the millions who want to work from home.

I suspect his statement was made to point out that companies will be prudent with capital spending – something that sounds like smart business. Both technology upgrades will be expensive, and a cable company could save a lot of capital spending by not upgrading everywhere. Perhaps he didn’t realize that he was outlining a plan for future discrimination against lower-income neighborhoods. Selective upgrading is also a scary possibility for towns outside the major metropolitan areas, and markets like county seats might get left behind.

In the past, all discussions of digital redlining were leveled against the big telcos, which often clearly upgraded to faster DSL in neighborhoods with higher household incomes. McDevitt outlines a future in which the cable company would likely be making those same kinds of selective upgrade decisions.

To be fair, cable company broadband networks are rarely the same today across larger communities. When we’ve helped communities take speed tests, we’ve often seen that performance on cable networks varies from neighborhood to neighborhood. This seems to mostly be due to varying conditions of the [hiscal plant. Some of the neighborhoods that got cable in the 1970s have much older coaxial cables than neighborhoods built more recently. Many cable companies have been loath to rip and replace wires for whole neighborhoods. It’s also likely that cable companies have taken shortcuts during upgrades, which require replacing all repeaters and power taps if done right. We also see cable markets are configured in what is called cascading, where neighborhoods are daisy-chained together – something that was probably done to save money in the past but which contributes to degraded broadband performance today.

It will be ironic if a decade from now, we’ll be talking about digital redlining from the cable companies while the telcos will have upgraded to symmetrical fiber. That would be a complete reversal of past history.

The Regulatory Struggle to Maintain Copper Networks

The California Public Utilities Commission has been investigating the quality of service performance on the telco networks operated by AT&T and Frontier. The agency hired the consulting firm Economics and Technology, Inc. to investigate numerous consumer complaints made against the two telcos. Thanks go to Steve Blum for following this issue in his blog.

Anybody who still has service on the two carriers will not be surprised by the findings. The full study findings have not yet been released by the CPUC, but the portions that have been made public are mostly what would be expected.

For example, the report shows a correlation between household incomes in neighborhoods and the quality of service. As an example, the average household incomes are higher in neighborhoods where AT&T has replaced copper with fiber. More striking is a correlation between service calls and household income. The annual frequency of repair calls is double for neighborhoods where the average household income is $42,000 per year or less compared to neighborhoods with household incomes of $88,000 or more.

Part of that difference is likely because more high-income neighborhoods have fiber, which has fewer problems and generally requires less maintenance. But there are also hints in the report that this might be due to economic redlining where higher-income neighborhoods get a higher priority from AT&T.

This is not the first time that AT&T has been accused of redlining. I wrote a blog a few years ago about a detailed study made in Dallas, Texas that showed a direct correlation between the technology being delivered and household incomes. That study followed up on a similar report from Cleveland, Ohio, and the same things could likely be said for the older telco networks in almost every big city.

The big telcos are in a rough spot. The older copper networks have largely outlived their economic lives and are full of problems. Over the years copper pairs of wire in the outdoor cables have gone bad and the remaining number of working copper pairs decreases each year. The electronics used to deliver older versions of DSL are long out of production by the telco vendors.

I’m not defending the big telcos, because the telcos caused a lot of their own problems. The telcos have deemphasized copper maintenance for decades. The copper networks would be in bad shape today even had they been maintained perfectly. But purposefully neglected maintenance has hastened the deterioration of copper networks. Additionally, the big telcos have also been laying off copper-based technicians over the last decade and the folks who knew how to best diagnose problems on copper networks are long gone from the companies. Consumers have painfully learned that the most important factor in getting a repair made for DSL or copper is the knowledge of the technician that shows up to investigate an issue.

The California Commission is likely at some point to threaten the big telcos with penalties or sanctions, as been done in the past and also by regulators in other states. But the regulators have little power to effect improvements in the situation. Regulators can’t force the telcos to upgrade to fiber. And no amount of documentation and complaining is going to make the obsolete copper networks function any better. AT&T just announced that on October 1 that it is not longer going to add new customers to the DSL network – that’s likely to really rile the California Commission.

I’m not sure exactly how it will happen, but the day is going to come, likely during the coming decade when telcos will just throw up their hands and declare they are walking away from copper, with zero pretenses that they are going to replace it with something else.  Regulators will rant and rave, but I can’t see any ways that they can stop the inevitable – copper networks at some point won’t work well enough to be worth pretending otherwise.

The Digital Redlining of Dallas

In 2018 Dr Brian Whitacre, an economist from Oklahoma State University looked in detail at the broadband offered by AT&T in Dallas County, Texas. It’s an interesting county in that it includes all of the City of Dallas as well as wealthy suburban areas. Dr. Whitaker concluded that AT&T has engaged for years in digital redlining – in providing faster broadband only in the more affluent parts of the area.

Dr. Whitaker looked in detail at AT&T’s 477 data at the end of 2017 provided to the FCC. AT&T reports the technology used in each census blocks as well as the ‘up-to’ maximum speed offered in each census block.

AT&T offers three technologies in Dallas county:

  • Fiber-to-the-home with markets speeds up to 1 Gbps download. AT&T offers fiber in 6,287 out of 23,463 census blocks (26.8% of the county). The average maximum speed offered in these census blocks in late 2017 according to the 477 data was 300 Mbps.
  • VDSL, which brings fiber deep into neighborhoods, and which in Dallas offers speeds as fast as 75 Mbps download. AT&T offers this in 10,399 census blocks in Dallas (44.3% of the county). AT&T list census blocks with maximum speeds of 18, 24, 45, and 75 Mbps. The average maximum speed listed in the 477 data is 56 Mbps.
  • ADSL2 or ADSL2+, which is one of the earliest forms of DSL and is mostly deployed from central offices. The technology theoretically delivers speeds up to 24 Mbps but decreases rapidly for customers more than a mile from a central office. AT&T still uses ADSL2 in 6,777 census blocks (28.9% of the county). They list the maximum speeds of various census blocks at 3, 6, 12, and 18 Mbps. The average speed of all ADSL2 census blocks is 7.26 Mbps.

It’s worth noting before going further that the above speed differences, while dramatic, doesn’t tell the whole story. The older ADSL technology has a dramatic drop in customer speeds with distances and speeds are also influenced by the quality of the copper wires. Dr. Whitaker noted that he had anecdotal evidence that some of the homes that were listed as having 3 Mbps of 6 Mbps might have speeds under 1 Mbps.

Dr. Whitaker then overlaid the broadband availability against poverty levels in the county. His analysis started by looking at Census blocks have at least 35% of households below the poverty level. In Dallas County, 6,777 census blocks have poverty rates of 35% or higher.

The findings were as follows:

  • Areas with high poverty were twice as likely to be served by ADSL – 56% of high-poverty areas versus 24% of other parts of the city.
  • VDSL coverage was also roughly 2:1 with 25% of areas with high poverty served by VDSL while 48% of the rest of the city had VDSL.
  • Surprisingly, 19% of census blocks with high poverty were served with fiber. I’m going to conjecture that this might include large apartment complexes where AT&T delivers one fiber to the whole complex – which is not the same product as fiber-to-the-home.

It’s worth noting that the findings are somewhat dated and rely upon 477 data from November 2017. AT&T has not likely upgraded any DSL since then, but they have been installing fiber in more neighborhoods over the last two years in a construction effort that recently concluded. It would be interesting to see if the newer fiber also went to more affluent neighborhoods.

I don’t know that I can write a better conclusion of the findings than the one written by Dr. Whitacre: “The analysis for Dallas demonstrates that AT&T has withheld fiber-enhanced broadband improvements from most Dallas neighborhoods with high poverty rates, relegating them to Internet access services which are vastly inferior to the services enjoyed by their counterparts nearby in the higher-income Dallas suburbs…”

This study was done as a follow-up to work done earlier in Cleveland, Ohio and this same situation can likely be found in almost every large city in the country. It’s not hard to understand why ISPs like AT&T do this – they want to maximize the return on their investment. But this kind of redlining is not in the public interest and is possibly the best argument that can be made for regulating broadband networks. We regulated telephone companies since 1932, and that regulation resulted in the US having the best telephone networks in the world. But we’ve decided to not regulate broadband in the same way, and until we change that decision we’re going to have patchwork networks that create side-by-side haves and have-nots.

Broadband Redlining

The National Digital Inclusion Alliance (NDIA) recently asked the FCC to investigate the practice of digital redlining, where big ISPs only bring the best technology to more affluent neighborhoods while ignoring poor ones.

The NDIA has statistics to back up it’s claims. They used the FCC’s data in 2017 to look in detail at how AT&T had deployed DSL in Cleveland. Years ago, AT&T deployed the first generation of DSL almost everywhere in the market. However, the company became far more selective in where they upgraded to faster DSL.

NDIA mapped where AT&T had deployed VDSL and later DSL technologies in Cleveland and found that the company had deployed faster DSL mostly in affluent neighborhoods and in the suburbs while leaving older downtown neighborhoods with the older DSL. VDSL offers speeds of at least 18 Mbps, up to nearly 50 Mbps when deployed using two copper pairs. NDIA found that the 55% of the census blocks in downtown Cleveland still had DSL speeds of 6 Mbps or less while 22% had speeds below 3 Mbps, with some as slow as 768 Kbps. It’s likely that AT&T marketed all versions of DSL the same, advertising ‘up-to’ speeds that described the fastest product in the market.

The AT&T deployment in Cleveland is not an isolated incident and the same is true in communities across the country. It’s not just AT&T that’s done this and Verizon deployed its fiber FiOS product in a similar manner and largely ignored northeast downtowns in favor of serving suburbs. We also tend to think of cable company networks being deployed ubiquitously in cities, but there are pockets in every major city that don’t have cable broadband.

In the industry this practice is generally referred as cherry-picking. It means deploying a new network in the places where the costs are lower or the expected penetration rates are higher – and ignoring the parts of a market that don’t fit a desired financial profile.

Historically the big telcos weren’t allowed to cherry-pick or redline. AT&T was still largely a regulated company when the first DSL was deployed. But the trend over time to deregulate telephone providers has led to laxer regulation, and obviously in Ohio and many other states the telcos were not required to build later generations of DSL everywhere.

One of the reasons we see so much cherry-picking is that many states have adopted statewide cable franchising. Cable franchises were historically negotiated in each community, and cities insisted that a cable provider build to the whole community as a condition for getting a franchise. However, AT&T and other broadband companies lobbied hard for statewide franchising rules, using the storyline that they wanted to deploy fast DSL to bring cable service. The statewide franchises generally give a cable provide the ability to build anywhere in a state. The telcos argued that the cost of negotiating with every community was killing innovation and deployment of faster broadband. What these companies really wanted was the ability to cherry-pick with no obligation to serve whole communities.

The practice of cherry-picking is still common today and most commercial fiber overbuilders engage in it to some degree. Most overbuilders have limited financial resources and they deploy fiber or other broadband technologies in those places where they get the best return for their investment. Many communities have seen fiber built to businesses and to new subdivisions while ignoring the rest of the town.

It’s hard to fault s smaller fiber overbuilder for maximizing the return on their fiber investment. On the flip side, there are few communities that don’t want fiber everywhere. However, most communities are realistic and know that if they always insist on getting fiber everywhere they might not get it anywhere.

Communities that really care about good broadband everywhere are the ones that are building fiber themselves or trying to attract a partner that will build the whole community. However, there are numerous states that hinder or prohibit communities from building broadband networks, and many other cities find the costs to build new networks to be prohibitive. The majority of communities must rely on the good behavior of the incumbents, and unfortunately they don’t always do the right thing.

The Downside of Big Data

DARPA_Big_DataBig tech companies have been crowing about some of the amazing things that can be done using big data. For example, in the area of interacting with people, retailers are working hard to create personalized shopping experiences aimed at individual shoppers. Specials will pop up on cell phones as someone walks by a display that are aimed at them specifically. While many will feel this is an invasion of privacy, others are looking forward to an enhanced shopping experience. Big data promises to also personalize things like health care so that every doctor you ever see will truly understand your health history and they can guard against conflicting medicines and other things detrimental to your health.

But there are already downsides to big data. Big data is being used to put together a detailed portrait of everybody. And that leads to various degrees of profiling. The very same data that can be used to make your shopping experience better can also be used for many negative purposes. Consider some of the following examples:

  • The Chicago Police department apparently used big data to create a list of the 400 people in the community that they think are most likely to commit a murder. But then they went so far as to contact these people to tell them they were watching them. If anybody remembers the movie Minority Report, this feels like we are already reaching that time where the police convict people for crimes they are going to commit in the future.
  • Big data contains a lot of information about us – our age, race, sexual orientation, religion, weight, general health, number of kids or pets, state of our finances, etc. That kind of data can be easily used to discriminate against people in a variety of settings. We start entering a scary societal place when we use this kind of data to profile people for consideration for housing, employment, etc. There is already an industry of firms who sell this kind of profiling data to anybody for a fee. Where a prospective landlord used to check your credit report they can now find out everything about you. Let’s face it – people are bigoted or just biased and the availability of this kind of data makes it easy to redline or discriminate.
  • There is a big uptick in scams against the elderly who are being found through big data. The scams themselves are as old as the hills, but it’s the use of big data to identify the most vulnerable among us that is disturbing.
  • It was reported in 2012 that Staples displays different on-line prices to different customers based upon where they live. For example, customers who live close to a competitor might get cheaper prices than somebody who does not. But this same ability makes it easy to price differently based upon other factors and again can lead to redlining.
  • I have read where it is fairly easy to buy databases of people who have something in common – such as having diabetes, having tried to quit smoking, or nameless other traits. These lists can be used to market products specific to an ailment, but they also have been used for scams, blackmail and other nefarious purposes. It’s not hard to picture being able to take advantage of people with a gambling addiction or some other such problem.
  • The FAA’s Do Not Fly list is another result of big data and is notorious for containing names of toddlers and others who are obviously not a threat to national security. The list even ended up including several US Congressmen.

This all points to the need for some sort of legal protection of people from the misuse of big data. This is a hot topic in Europe right now but is not yet commonly debated here. Several civil rights groups have identified big data as a big threat and a new source for discrimination. But misuse of big data can go far beyond discrimination based upon race, religion or sexual orientation. Unfortunately it’s now possible to discriminate based upon a whole lot of other reasons as well.