Opinion: U.S. broadband holds up under virus pressure

Richard Bennett

To be sure, the coronavirus pandemic has laid bare some the most pressing short- and long-term challenges in our public health infrastructure. But it’s also tested another infrastructure — our broadband networks — that serves as a lifeline for offices, schools, places of worship and even much of our health care during the crisis. Fortunately, this infrastructure has weathered a 20-35% spike in demand with scarcely a hiccup. 

This successful stress-test of U.S. broadband networks stands in stark contrast to the more heavily, government-regulated, underfunded networks in Europe which haven’t been as able to handle the surges. There, the commissioner for internet markets, Thierry Breton, had to ask Netflix, YouTube, Facebook, and the newly launched Disney+, to degrade the picture quality of their streams.

So far, nothing like that has occurred here in the U.S.

Ashley Bell, left, 13, and Natalie Bell, 14, daughters of Kyle Bell of Negaunee Township, surf the internet at their home. The U.S.’ internet infrastructure has weathered a 20-35% spike in demand during the pandemic, Bennett writes.

Policy makers can learn many lessons from the U.S. success. For most of the last twenty years, U.S. broadband was regulated lightly by the federal government where it was necessary to protect consumers. This produced nearly $2 trillion in private investment — nearly $80 billion a year —and built some of the fastest and most affordable networks in the world. Ninety-five percent of American households now have broadband available, and 80% have access to gigabit speeds. 

Europe, by contrast, regulates networks more heavily with utility-like rules. Speeds lagged, investors held back, and European consumers are now paying the price. Yet, this unhappy experience abroad hasn’t stopped some of the more extreme voices from urging that we follow Europe’s model.  

The widely acknowledged gold-standard of broadband performance measurement is SamKnows. It reports a dip of less than 1% on download speeds in most places since the stay-at-home orders. Other world-leading measures like Ookla’s speedtest.net, finds that while download speeds dropped initially in some places by 5% or less, they have ticked back up to pre-COVID levels. Speedtest reports U.S. broadband speeds are even better than they were in December. 

But every crisis breeds its own cottage industry of opportunists claiming the sky is falling, and COVID-19 is no exception. BroadbandNow — a new kid on the internet performance block — recently published a tabloid-like report claiming that half of our largest 200 cities were seeing broadband performance downgrades — of sometimes 20% or more — since we started sheltering. These reports have about as much credibility as bigfoot sightings. They are the equivalent of “junk science".

These sensationalized reports rely on M-Lab tests, which are notorious among engineers for producing false negatives.

M-Labs design flaws are widely known and mostly technical: They test one TCP connection at a time when most users have anywhere from 30-100 open at one time, all optimizing speeds by exchanging data over the ISPs’ network in parallel. That’s like measuring the throughput of a firehose through a thin cocktail straw.

But even though inaccurate, this kind of tabloid sensationalism is a fix for cortisol-junkies who thirst for conflict — knowing that it helps build brands and Twitter followers. Sadly, the easily debunked shoddy research doesn’t stop others — sometimes even credible outlets — from parroting them.

By any credible measure, U.S. broadband is performing better than nearly anywhere else and, if anything, is even making our internet stronger according to many experts.

This is not to say that U.S. broadband doesn’t have its challenges — we need to get every rural and poor community connected and online. But shooting at decoy issues distracts from where we need to focus.

Richard Bennett is the founder of High Tech Forum, a network engineer, and an internet policy expert.