Joe Biden’s campaign agreed to a skeletal broadband policy in what amounts to a peace treaty with Bernie Sanders and his supporters. The “unity task force recommendations” published on Wednesday amount to little more than a declaration that broadband is good, but it’s the first time that Biden has explicitly signed on to any conventional democratic party positions on telecommunications policy.
Network neutrality gets a mention. The “recommendations” propose to…
Restore the FCC’s clear authority to take strong enforcement action against broadband providers who violate net neutrality principles through blocking, throttling, paid prioritization, or other measures that create artificial scarcity and raise consumer prices for this vital service.
Presumably, what they really mean is that Biden will appoint FCC commissioners who will use the “clear authority” that already exists. Which means policy ping pong will continue. During the Obama administration, the democratic majority on the Federal Communications Commission classified broadband as a regulated common carrier service. The Trump administration’s republican led FCC said it wasn’t.
The recommendations don’t explicitly promise to restore broadband’s common carrier status. There are other ways of imposing net neutrality obligations on Internet service providers, as California discovered. Initially, the Obama era FCC proposed a no lobbyist left behind approach that would have allowed monopoly model incumbents to negotiate permission for decidedly non-neutral network management practices.
Direct intervention by the Obama white house put an end to that pretence. Whether a Biden administration would do the same is an open question. Biden has enjoyed a long and comfortable political career in a system fuelled by, and largely subservient to, cash from corporations and labor unions. The peace treaty with Sanders doesn’t change that reality.
It is not unusual, of course, for internet traffic to grow…What is unusual in the Covid–19 environment is the suddenness of the traffic growth. Rather than growing 30% in a year, traffic grew about that much in a month. Sandvine reports a “staggering increase in volume for network operators to cope with and absorb.” During March, according to Sandvine, global traffic grew 28.69% with an additional 9.28% during April, for a total of 38% over the two months. Upstream traffic growth was even more stunning, up 123.18% in March before leveling off…
The U.S. networks’ fixed-broadband speed bottomed out within three weeks, as did the global index, while the speeds of the EU, EU–4, and OECD continued to decline for another three weeks.
Kovacs’ conclusions – the apparent superior performance of U.S. networks is due to the beneficence of telecoms companies and the somnambulance of the Federal Communications Commission – are unsupported. Her top level observations might correlate to Ookla’s network traffic data, but she offers no analytical rigor or evidence of causation. The data sets she relies on aren’t necessarily globally consistent. For example, Ookla’s Speedtest.net is based in Seattle and relies on crowdsourced data. It’s a mistake to assume that the crowd generating the data in the U.S. is largely identical to the crowd in the E.U. It might or might not be.
Comparisons of the same population over a few weeks time are valid, though. The top line conclusion stands: globally, networks withstood the initial covid–19 induced surge, and adapted to higher traffic levels within a few weeks.
The Internet was originally designed to ride out a nuclear war. It works just fine in a pandemic, too.
AT&T’s and Verizon’s troubles weren’t exactly a shock, either. Some business lines, like video and copper-based broadband service, have been fading for some time. The covid–19 emergency accelerated that trend. In the first three months of 2020, AT&T lost 897,000 video subscribers and nearly 300,000 DSL customers. Even though its broadband business added 209,000 fiber subs, it still saw a net loss of 73,000 broadband accounts overall. Verizon lost 84,000 FiOS video subs, while gaining 59,000 fiber broadband customers.
AT&T gained wireless subscribers in the first quarter, while Verizon lost some, blaming the store closures forced by the covid–19 lockdown. The real numbers to watch, though, will be the results of the now big three mobile operators in the second quarter. By July, we’ll know if the shift to in-home mobile network-enabled hotspots is significant.
Both companies “withdrew financial guidance”, which means they’re not willing to make any predictions about how shareholders will fare over the next few months.
AT&T’s captain is jumping ship. In a move that’s been long expected, CEO Randall Stephenson will hand off to COO John Stankey in July. Stankey has been working for AT&T for 35 years. He’s been running Warner Media since AT&T took it over, and is in charge of launching HBO Max, which is a streaming video service that’s supposed to compete with the likes of Netflix and Disney. That would be difficult for any executive, but for someone with no history and no apparent friends in the entertainment business, and who spends a lot of time talking about things like “headcount rationalization” – AKA firing people – it would be a miracle.
Most broadband networks in urban California managed the load well as covid–19 shelter in place orders began taking effect, according to speed data analysed by BroadbandNow, a broadband sales website. The underlying data came from Measurement Lab, an open source Internet monitoring project that includes universities, foundations and major tech corporations among its members.
BroadbandNow compared the median download speed in the 200 largest U.S. cities (34 of which are in California) during the week of 15 to 21 March 2020 to the median speed range for the prior weeks of 2020. That’s the week that began with Bay Area counties imposing lockdowns and ended with a statewide quarantine order from governor Gavin Newsom.
Broadband speeds in 21 cities, mostly in southern and central California, stayed within 2020 ranges. The 13 that saw decreases were evenly split between the north and south.
San Jose, which along with Fremont had the slowest median speed, ran 38% under its prior 2020 range, the second worst in the state. Oxnard was hit the hardest in California, with a 42% drop in median Internet service speeds, which was also the second worst in the U.S.
I’m not reading too much into these numbers yet. The week of 15–21 March was an uneven scramble of lockdown orders and efforts to comply. The following week’s data, when all Californians were under more or less the same regime, will be more telling.
The worst broadband infrastructure in California is, not surprisingly, found in mountain counties at the north end of the state. Trinity and Siskiyou counties both get “F” grades for broadband infrastructure, with a numerical score of dead zero. Sierra County likewise gets an “F”, with a numerical score of 0.03 that’s effectively zero. It is also the county with the highest percentage of population – 88% – without any access to wireline broadband service. It’s a serious problem for rural residents as business, education, health care and education move almost exclusively online during the covid–19 lockdown.
As the map above and the tables below show, the best broadband infrastructure in California can be found in San Francisco and Alameda County, and in Los Angeles and Orange counties, and it generally gets worse the farther you get from those two hubs.
Broadband Report Card Grade
Percent of Population at Zero service
Broadband Report Card Grade
Percent of Population at Zero service
The grades are based on a methodology I first developed in 2013 for the East Bay Broadband Consortium, and have since been updating on a statewide basis for the Central Coast Broadband Consortium and other such regional organisations. It uses the broadband speed and technology claims that Internet service providers submit to the California Public Utilities Commission and Federal Communications Commission every year – the most recent data set is current as of 31 December 2018. The underlying data are not very accurate, but it is precise in the sense that companies tend to be consistent about the way they exaggerate coverage. That makes it useful for comparative purposes, which is what the California Broadband Infrastructure Report Card is all about.
The best broadband infrastructure in California is found in the Bay Area and Sacramento, and the Los Angeles/Orange County core. Those two counties straddle California’s “C” average. The large share of the state’s population those counties represent means what’s available to people there is going to look a lot like what’s available to the average Californian.
San Joaquin and Stanislaus counties also made the top 10 due to upgrades by AT&T, which claims to have built fiber to the premise systems in about 16% of the census blocks (but not necessarily throughout whole census blocks) in those two counties, and Vast Network’s middle mile fiber infrastructure, which runs through a lot of census blocks but isn’t available to homes or most businesses. It’s similar story in Fresno County. The grades reflect infrastructure deployments rather than service availability, so for a lot of Central Valley communities it’s water, water everywhere but not a drop to drink.
At least 1.5 million Californians – 4% of the state’s population – cannot get wireline broadband in their homes, as the second week of the corona virus lock down begins. That’s what the most recently published broadband availability reports filed with the California Public Utilities Commission show. Nearly twice that many – 2.8 million people, 8% of the population – don’t have access to primary wireline service that delivers 100 Mbps download/20 Mbps upload speeds, the minimum service level needed for in-home work, education, health care and entertainment. The dead spots are disproportionately found in rural communities.
That’s the top line result from the number crunching I’ve been doing, as part of the Central Coast Broadband Consortium’s (CCBC) efforts over the past few days. Like other regional broadband consortia around California, the CCBC is working to identify broadband access gaps and resources to fill them for the 40 million Californians ordered to stay at home. The raw data are the broadband service claims filed by primary wireline Internet service providers – the companies that own the copper and fiber – as of 31 December 2018, which is the most recent data set available. I ran the data for the 58 counties in California, with breakouts for cities, unincorporated communities (AKA census designated places) and tribal areas – it’s pretty much as easy to do it for all 58 counties as it is to do it for one (or the CCBC’s three – San Benito, Monterey and Santa Cruz counties).
The analysis includes:
Infrastructure grade (A, B, C, D, F). The grading methodology is here.
Median household income estimates.
Population, percentage of population, number of housing units and percentage of housing units, as of 2019 estimates, with zero access to wireline broadband service.
Population, percentage of population, number of housing units and percentage of housing units able to access primary wireline broadband service at 6 Mbps download/1 Mbps upload, 10 Mbps/1 Mbps, 25 Mbps/3 Mbps, 100 Mbps/20 Mbps, 1,000 Mbps/500 Mbps.
The data is not particularly accurate, as is widely acknowledged, but it is precise in the sense that the inaccuracies tend to be consistent across a provider’s data set. It is sufficiently precise for the purpose of comparing one place to another, and for generally assessing broadband availability in a given county, city or other area. I’ll be writing more about it in the coming days, and welcome any ideas or critiques.
More and more people around the world will have access to faster and faster broadband connections, with speeds for fixed and mobile service doubling and tripling by 2023, due in large part to increased global deployment of fiber to the premise and 5G technology, according to a white paper recently published by Cisco. Although North America will continue to beat world broadband speed averages, the U.S. will not be among the leaders in advanced infrastructure deployment.
Cisco’s research indicates that the average North American mobile broadband connection in 2018 ran at 22 Mbps, and that will nearly triple to 58 Mbps by 2023. 5G networks will own a large share of that increase, but the U.S. won’t earn a podium spot…
The top three 5G countries in terms of percent of devices and connections share on 5G will be China (20.7%), Japan (20.6%), and United Kingdom (19.5%), by 2023.
Globally, fixed broadband speeds will also jump, particularly in countries that, unlike the U.S., are focused on making fiber infrastructure ubiquitous…
The global average broadband speed continues to grow and will more than double from 2018 to 2023, from 45.9 Mbps to 110.4 Mbps…Several factors influence the fixed broadband-speed forecast, including the deployment and adoption of Fiber-To-The-Home (FTTH), high-speed DSL, and cable broadband adoption, as well as overall broadband penetration. Among the countries covered by this study, Japan, South Korea, and Sweden lead in terms of broadband speed largely because of their wide deployment of FTTH.
North America’s average fixed broadband download speed will be comfortably above the global average at 142 Mbps by 2023, according to the white paper. That only earns us second place on the world league table, though. Asia will still be tops at 157 Mbps, and higher growth rates – 30% annually or better – in Latin America, the Middle East and Africa mean that the gap between U.S. broadband speeds and those in the developing world will continue to close.
Fears that Internet routers and switches will melt under an onslaught of 8K-enabled cord cutters can be put aside for a few years, according to projections released by the Consumer Technology Association (CTA). But the number of U.S. households with 4K screens will continue to grow rapidly, and that will be problematic enough for broadband service providers: 25 Mbps download speeds will be the minimum needed to serve the typical U.S. home.
8K is a big screen technology. According to Steve Koenig, CTA’s vice president of research, those sets are in the 70-inch and up range, which is more limited market – not everybody wants something that big in the living room (although the same could have been said about 40-inch to 50-inch screens a few years ago). He projects half a million 8K screen shipped into the U.S. market in 2020, which is a drop in the bucket, particularly considering that many, if not most, of those will be used for commercial and industrial applications.
On the other hand, CTA’s projections show 4K sets hitting a steady state shipment rate in the upper 20 million annual unit range for the next three years, climbing to 32 million by 2023. Using the same, back of the envelope calculations that I used a couple of years ago, that means that about half of U.S. homes have 4K screens now, and that share will climb to at least three quarters by the end of 2021. We’re to the point where 4K resolution is the market standard, and it’ll soon be almost as hard to buy anything less as it is to find a black and white TV set now.
4K video needs a steady 15 Mbps stream to function in real time, and to have a fighting chance of getting that bit rate consistently requires service specced at 25 Mbps download speeds or better. That’s the minimum set for rural homes by the federal agriculture department, and it’s the FCC’s de facto minimum as well.
California’s department of transportation, AKA Caltrans, is a step closer to actively collaborating with broadband service providers and local governments to put more conduit in California’s thousands of miles of state highways and make it available. It published a Dig Smart white paper that summarises “dig once” policies that have already been adopted by cities and other states. Those policies are intended to ease the way for telecoms companies to install conduit when road construction or utility excavation projects happen, and to encourage them to take advantage of the opportunity.
The white paper doesn’t offer any hard recommendations, but it does outline policies “which may be investigated” by Caltrans as it moves ahead with implementing a law – assembly bill 1549 – which was passed four years ago…
State [departments of transportation] make agreements with service providers to exchange the use of right-of-way or existing conduit infrastructure for the use of fiber optic services
Require providers of broadband services and other utilities to install infrastructure at the same time, in the same trench, or in the same conduit, and share the cost of installing the infrastructure
Moratorium on excavation to preserve new roadway construction and encourage utility coordination planning
Encourage the use of trenchless technologies, such as:
Horizontal Directional Drilling: A trenchless method of installing underground pipes, conduits and cables along a prescribed bore path by using a surface-launched drilling rig
Micro-Trenching: Digging a small trench just inches under the road surface along the curb line to install fiber optic lines
Provide access to fiber, conduit and projects maps
Notify telecommunication companies of projects where broadband infrastructure can be installed
Reduce permitting costs and wait time for projects which implement coordinated utility planning
If you’re interested in the topic, the white paper provides an interesting overview of dig once programs at the state, local and federal levels. There’s no timeline for adopting a comprehensive dig once, or dig smart, policy, but the agency is moving in that direction. Caltrans is often criticised for being slow to change, but the flip side is that once its gets moving, it moves relentlessly.
I serve on Caltrans’ conduit task force, and I advocated for and helped to draft AB 1549. I’m involved and proud of it. Take it for what it’s worth.
Technological tipping points are easy see in the rearview mirror – do you remember what the world was like before the iPhone? – but hard to spot in advance. One might be on the way. A well respected analyst, Ming-Chi Kuo, who works for TF Securities, predicts that Apple will start using ARM-based chips it designs and makes itself in Macintosh computers.
Kuo forecasts that Apple will be using a 5-nanometer process at the core of its new products 12 to 18 months’ time. As part of this, Kuo believes there will be a “new 1H21 Mac equipped with the own-design processor”…
Shifting over to an ARM-based chip would also give some context to Apple’s decision to move away from supporting 32-bit apps in macOS Catalina, as well as Apple’s work on Catalyst. In theory, this could allow Apple to use the same chips in the Mac as it does in iPhones and iPads, reducing its overall costs and enabling apps to be more usable throughout the entire Apple ecosystem.
It would be the first time a major personal computer maker abandons processors built on Intel’s 40 year old x86 architecture, and switches to chips based on ARM designs. Other hardware companies have dabbled with ARM-based servers and PCs – like Blackberry and Palm dipped their toes into the pre-iPhone smartphone market – but if he’s right, Apple will be the first to 1. shift an entire line of mainstream computers off of Intel and onto ARM, and 2. build a complete smartphone-tablet-computer-consumer device ecosystem around a single chip architecture.
There’s a broadband angle to this move, too. If the Apple device universe collapses into a single, integrated hardware and operating system platform, with the only distinction between devices being form factor and peripheral functions like sensors and telephone network access, then its value will be maximised by giving those devices seamless access to a common set of data, content, applications and services via persistent and ubiquitous connectivity.
It’s one thing to rely on file and data syncing across a family of products, as Apple does now, but it’s quite another to build a costly lineup of hardware, software and content on top of the assumption that connectivity can be taken for granted anywhere you go.