I am the last person to deny anyone a good snarky gloat. So while I don’t agree entirely with AT&T’s policy blog post taking a jab at reports of Google Fiber stumbling in deployment, I don’t deny they’re entitled to a good snarky blog post. (Google, I point out, denies any disappointment or plans to slow down.) “Broadband investment is not for the feint hearted,”
But the irony faeries love to make sport. The following week National Digital Inclusion Alliance (NDIA) had a blog post of their own. Using the publicly available data from the FCC’s Form 477 Report, NDIA showed that in Cleveland’s poorest neighborhoods (which are also predominantly African American), AT&T does not offer wireline broadband better than 1.5 mbps DSL – about the same speed and quality since they first deployed DSL in the neighborhood. This contrasts with AT&T’s announcement last month that it will now make its gigabit broadband service available in downtown Cleveland and certain other neighborhoods.
Put more clearly, if you live in the right neighborhood in Cleveland, AT&T will offer you broadband access literally 1,000 times faster than what is available in other neighborhoods in Cleveland. Unsurprisingly for anyone familiar with the history of redlining, the neighborhoods with crappy broadband availability are primarily poor and primarily African American. Mind you, I don’t think AT&T is deliberately trying to be racist about this. They are participating in the HUD program to bring broadband to low-income housing, for example.
There are two important, but rather different issues here — one immediate to AT&T, one much more broadly with regard to policy. NDIA created the maps to demonstrate that a significant number of people who qualify for the $5 broadband for those on SNAP support that AT&T committed to provide as a condition of its acquisition of DIRECTV can’t get it because the advertised broadband in their neighborhood is soooo crappy that they fall outside the merger condition (the merger requires AT&T to make it available in areas where they advertise availability of 3 mbps). Based on this article from CNN Money, it looks like AT&T is doing the smart thing and voluntarily offering the discount to those on SNAP who don’t have access to even 3 mbps AT&T DSL.
The more important issue is the return of redlining on a massive scale. Thanks to improvements the FCC has made over the years in the annual mandatory broadband provider reporting form (Form 477), we can now construct maps like this for neighborhoods all over the country, and not just from AT&T. As I argued repeatedly when telcos, cable cos and Silicon Valley joined forces to enact “franchise reform” deregulation in 2005-07 that eliminated pre-existing anti-redlining requirements – profit maximizing firms are gonna act to maximize profit. They are not going to spend money upgrading facilities if they don’t consider it a good investment.
Again, I want to make clear that there is nothing intrinsically bad or good about AT&T. Getting mad at companies for behaving in highly predictable ways based on market incentives is like getting mad at cats for eating birds in your backyard. And while I have no doubt we will see the usual deflections that range from “but Google-“ to “mobile gives these neighborhoods what they need” (although has anyone done any actual, systemic surveys of whether we have sufficient towers and backhaul in these neighborhoods to provide speed and quality comparable to VDSL or cable?) to “just wait for 5G,” the digital inequality continues. I humbly suggest that, after 10 years of waiting and blaming others, perhaps we need a new policy approach.
More below . . .
Our story begins with the FCC’s Order approving the AT&T/DIRECTV deal. If you flip to the actual merger conditions in Appendix B, one of the public interest sweeteners was expanding the availability of affordable broadband service. AT&T agreed to provide low-cost stand alone broadband access to “qualifying” households (“qualifying”=at least one person in household on Supplemental Nutrition Assistance Program, aka SNAP aka food stamps.) Under the merger condition:
- Where AT&T offers top speed of 10 mbps or better, AT&T must offer a 10 mbps for $10/month, where technically available.
- Where AT&T offers top speed between 5 mbps and less than 10 mbps, AT&T must offer 5 mbps down for $10/month, where technically available.
- Where AT&T offers top speed of less than 5 mbps, AT&T must offer 3 mbps down, where technically available.
What You Need To Know About How DSL Works To Understand What Comes Next.
Why do it this way? AT&T offers a lot of different DSL speeds, depending on the technology used and distance from something called the “central office” (CO). For those not familiar with telco jargon or DSL technology, the “central office” is the place where the phone system delivers phone calls or data destined for a particular neighborhood or other geographic area. If you had a dial tone back in the old days, you might have noticed that the second three digits of your ten digit phone number (e.g., “xxx-555-1212”) were the same as your neighbors’. That’s because that three digit code directs the phone call to the right central office. The last 4 digits then go to the specific address.
For DSL, particularly older DSL, how near or far you are from the central office makes a huge difference in how fast and reliable your DSL is. If you are reasonably close to the CO, you get better speed. The further you get from the CO, the slower and less reliable the DSL. Additionally, people have worked a lot to improve DSL since phone companies started deploying. We now have several flavors of “VDSL.” VDSL is even faster than DSL, but the same problem exists. The further away from the CO, the slower and less reliable DSL, including VDSL, gets.
Additionally, how the Internet packets actually get delivered to the CO matters. “U-verse” was a “fiber to the node” approach where AT&T ran fiber to the CO. DSL lines connected to nodes with U-verse are faster, but the speed still gets slower the further you get from the CO.
Fascinating, But Why Do I Need To Know This?
As a result of these tech issues, the DSL speed AT&T offers varies dramatically across it’s wireline footprint and even by neighborhood. I have a friend Dave in Sacramento who can get at most 3 mbps DSL while the people literally across the street in the same housing development can get 6 mbps. Why? Because AT&T just never upgraded the lines from his CO. ‘Sorry, Dude. Unregulated IP-based service and all that. We couldn’t get to your CO the day we were in your development, and it wasn’t worth paying overtime or delaying our roll-out schedule to finish the job, so we just moved on to the next block. So it’s overpriced bundled cable or 3 mbps. Nothing personal, just sucks to be you. Love & smoochies, AT&T.”
So the FCC wrote the low-cost alternative merger condition to recognize the following realities. First, the FCC wanted AT&T to offer a decent speed. Since Comcast offered crap speed under the NBCU merger condition, the FCC wanted to make sure that AT&T offered speeds the FCC considers actually useful (minimum of 10 mbps down). But they also didn’t want to force AT&T to upgrade its footprint just to provide better low-cost Internet to poor people. (Besides, AT&T promised to upgrade its entire wireline footprint.)
So the FCC set this up so that, if AT&T doesn’t offer 10 mbps, they will make the next best decent speed available to low income households. But even here, the FCC had to recognize that customers served from the same CO might get different speeds. So they included the phrase “where technically available.” So if it turned out you were like Dave of Sacramento in an area where AT&T generally offered 6 mbps, but you can only get 3 mbps on your block, AT&T doesn’t have to upgrade their network. They can sell you 3 mbps.
Where the FCC apparently screwed up was in not recognizing that there would be significant areas (particularly outside rural areas, which were addressed in a separate network upgrade condition) where AT&T doesn’t make available DSL speed of at least 3 mbps. In theory, even totally not upgraded since it got deployed in the early/mid-00s DSL supports 3 mbps down. What the FCC missed is that, even in urban areas, there are neighborhoods where the crappy totally ancient not upgraded 3 mbps DSL gets even slower (advertised speed of 1.5 mbps).
What?! How Did AT&T Get Away With Not Upgrading The Lines In These Neighborhoods For Ten Years?
Once upon a time, states had laws about something called “redlining.” The name comes from the practice of service providers outlining “bad” neighborhoods in red ink and refusing to serve anyone on the wrong side of the red line. As a general matter, redlining targeted neighborhoods predominantly African American or other people of color (POC), as well as particularly low-income communities (often, but not always, a convergent set).
During the 1960s and 1970s, various level of government decided to deal with this by requiring service to the full community. If you wanted to offer a service that needed a franchise from state or local authority – like cable or telecom – you had to agree to serve all parts of the franchise area equally. As a result, cable companies built out cable systems through their entire franchise area, and phone companies built out phone systems throughout their entire service area. When it came time to convert these systems to broadband in the early and mid-00s, these companies were required to upgrade their entire footprint, so everyone in the franchise area got broadband.
In the mid-00s, the telcos started offering video as well as DSL to compete with cable – which was now offering voice service in addition to video and cable modem service. Everyone rejoiced at the prospect of competition. Yay! As usual, however, there was a catch. The telcos said that they could not enter the video and broadband market quickly enough to compete with cable if they needed to get local franchises from local cable authorities on top of their existing state franchises for telephone service (DSL, being reclassified as an information service in ’05], required no state franchise). So they lobbied the states (and the FCC) to eliminate local franchising and replace it with a considerably deregulated state franchising system. In addition to striping out the power of local franchise authorities to negotiate terms, it also eliminated the requirement to serve the entire franchise area – thus eliminating the specific anti-Redlining provisions.
As someone who lived through this and opposed it at the time, let me be clear that this was not some secret fringe thing. Lots of people in both political parties supported it. Cable companies and tech companies joined with the telcos in getting the legislation passed on a state-by-state basis as a means of “encouraging broadband investment.” Heck, groups like the National Council on Aging and Rainbow/PUSH coalition joined with the incumbents and Silicon Valley in urging passage of franchise reform.
What! Rainbow/PUSH Supported Repeal of Anti-Redlining Rules?
In fairness, it wasn’t exactly sold that way. As always, “franchise reform” was sold as a deregulatory move to get supposedly greedy local franchise authorities who were purportedly demanding too much from telcos and other potential new entrants out of the way of new investment in broadband and video competition. Also in fairness, these laws tended to have a general anti-discrimination provision, so the incumbents could swear up and down that the law still prevented discrimination. Unfortunately, while the law prevented overt discrimination, it did not mandate that providers build out and/or upgrade throughout the franchise area – which was the heart of the anti-redlining laws.
Here is the page from the Ohio PUC explaining the relevant Ohio law passed in 2007 (and signed by then-Governor Ted Strickland, now running as the Democratic candidate for Senate). As you can see, it contains a general prohibition on discrimination. But it does not contain a requirement to build out systems equally throughout the franchise area. So any provider sells their service without discriminating on the basis of race. However, if you have the misfortune to live in an area the provider does not find profitable to upgrade, you have no recourse.
As a result, AT&T broadband in Cleveland varies from gigabit speed in those neighborhoods where AT7t just launched its Gigapower service to less than 3 mbps down in these neighborhoods identified by NDIA. I am not a huge expert on the geography of Cleveland, but I suspect that the areas with Gigapower are a lot richer and whiter than the areas with less than 3 mbps down.
How Did NDIA Document The State of Broadband In Cleveland.
By law, the FCC needs to do an annual report on the deployment of broadband “to all Americans.” Under Chairman Tom wheeler, the FCC has increasingly improved the granularity of the report. (As usual, these improvements in the reporting requirements were fiercely resisted by industry and harshly criticized by Republicans.) When the FCC started doing the annual broadband report (also called the “Section 706 Report” for Section 706 of the Telecommunications Act of 1996), it did not require providers to break out speeds and considered an area “served” by a provider if the provider offered any service in the zip code. As of last year, the FCC required that broadband providers list what speed they actually offer by census block.
For those unfamiliar with this terminology, a “census block” is the smallest, most granular division used by the Census Bureau. Census blocks are the smallest unit of a community identifiable by existing boundaries such as streets. In an urban area, a block of apartment buildings with several hundred residents makes up a census block. In suburban areas, it is generally larger, but still tries to capture the same idea of a logical community boundary.
AT&T therefore properly reported the broadband speed actually available on a neighborhood by neighborhood basis. Remember the background above about DSL and the whole “central office” thing. As AT&T reported, buildings too far from the central office to receive the expected 3 mbps only get 1.5 mbps. When NDIA tried to get $5/month broadband for qualifying homes in these neighborhoods, AT&T reported that they could not get the Access from AT&T discount rate because they did not have access to DSL with even 3 mbps download speed.
NDIA then took the existing 2015 FCC Broadband Report data and overlaid it with the existing census map that lists census blocks. Using this rather simple method, NDIA produced a map of the neighborhoods in Cleveland that could not get Access AT&T. No surprise, it turns out that the inner city neighborhoods of Cleveland have the worst broadband in the city, with zero upgrades apparently done since AT&T first brought DSL to these communities.
Or, in other words, contrary to the promises from AT&T, every other broadband provider, Silicon Valley, and everyone else who supported “franchise reform,” AT&T acted in precisely the manner predicted by opponents of franchise reform. They did not invest in neighborhoods they did not deem “profitable,” recreating the redlining that had existed prior to repeal of the anti-redlining laws.
This Isn’t Just About AT&T – Or Just About Cleveland.
As always, it would be nice to believe that the problem is solved because AT&T has announced it will make Access AT&T available to low-speed homes in Cleveland. But while that solves the immediate problem, it does nothing about the return of redlining and the ever increasing digital divide problem and the broadband homework gap for children in these neighborhoods.
We have a lot of anecdotal evidence that this problem extends all over the United States. Remember my friend Dave in Sacramento? Additionally, we know that Verizon stopped deploying its FIOS upgrade in 2010, and while Verizon may now be starting to deploy again in the urban core, a lot of inner city areas remain outside Verizon’s existing FIOS footprint. I don’t have the capacity to do this myself, but I am willing to bet that if someone with mapping software wants to replicate what NDIA did for other cities, I’m willing to bet they will find a similar pattern of broadband providers not upgrading in neighborhoods they find less desirable, i.e., poor and/or predominantly POC.
We need to recognize that profit maximizing firms will behave like profit maximizing firms. If we want companies to deploy where the combination of poor customers and persistent negative stereotypes scares companies off, we need to DO SOMETHING to change that dynamic other than deregulate and hope for the best.
Every city ought to replicate what NDIA did to find holes in broadband coverage. First, everyone ought to thank Chairman Wheeler and the other Democrats on the FCC for resisting the usual resistance from industry and partisan pushback from Republicans on enhancing the detail and granularity the FCC requires for its annual broadband report. That said, nothing prevents any other organization – or any local government – from doing what NDIA did for Cleveland. Take the available FCC data for 2015, put it over a census block map, and figure out which neighborhoods have crap broadband and/or more than one broadband provider. (I expect Charter or whoever offers broadband in the relevant Cleveland neighborhoods, for example, but I have no idea what price and what speed.)
Don’t keep waiting for someone else to come in and save us. For the last ten years, we keep getting explanations for how someone else will come along and solve the problem. First we got the “wireless 4G is just as good as wireline.” Then we got “Google Fiber will come and save us.” But wireless quality varies around these neighborhoods based on things like tower density and thickness of walls, and availability of fiber for backhaul. We have no way of knowing service quality for 4G in any of these buildings. Meanwhile, as AT&T itself gloated 2 weeks ago, Google Fiber isn’t likely to get deployed in new locations any time soon.
Recognize we need a plan – at the federal, state and local level. As I keep stressing, this isn’t about a single company or even a single local. To paraphrase Anna Karenina: ‘Neighborhoods with lots of high-speed broadband choices are all alike. Every neighborhoods with lousy broadband choices is screwed up in its own way.’ As we look to the incoming administration, we need a national plan to figure out how to address the lack of affordable high-speed broadband in urban neighborhoods as well as the problem of deployment to rural areas. But a federal plan won’t be enough. At the end of the day, local communities know their own neighborhoods and know their own problems. We will need local plans and strategies and cooperation at the state level, as well as a federal government that makes affordable broadband for ALL Americans a priority.
Stay tuned . . .
Harold, thanks! Great explanation.
For interested Wetmachine readers, there’s a somewhat more interactive version of the Cleveland map you point to, along with a comparable map of Detroit, at Connect Your Community’s blog. (CYC is a local NDIA affiliate working in Cleveland and Detroit.)