FCC Chairman Jessica Rosenworcel has circulated a draft Notice of Inquiry inside the FCC to kick off the required annual report to Congress on the state of U.S. broadband. As part of preparing that report, she is recommending that the FCC adopt a new definition of broadband of 100/20 Mbps and establish gigabit broadband as a longer-term goal. I have a lot of different reactions to the idea.

First, the FCC is late to the game since Congress has already set a speed of 100/20 Mbps for the BEAD and other federal grant programs. This is entirely due to the way that the FCC has become totally partisan. Past FCC Chairman Ajit Pai was never going to entertain any discussion of increasing the definition of broadband since he was clearly in the pocket of the big ISPs. The FCC is currently split between two democrats and two republicans, and I find it doubtful that there can be any significant progress at the FCC on anything related to broadband in the current configuration. I have to wonder if the Senate is ever going to confirm a fifth commissioner—and if not, can this idea go anywhere?

Another thought that keeps running through my mind is that picking any speed as a definition of broadband is completely arbitrary. We know in real life that the broadband speed to a home changes every millisecond, and speed tests only take an average of the network chaos. One of the things we found out during the pandemic is that jitter might matter more than speed. Jitter measures the variability of the broadband signal, and a customer can lose connectivity on a network with high jitter if the speed drops too low, even for a few milliseconds.

I also wonder about the practical impact of picking a definition of speed. Many of the current federal grants define a served customer as having an upload speed of at least 20 Mbps. It’s clear that a huge number of cable customers do not see 20 Mbps upload speeds, and I have to wonder if any State broadband offices will be brave enough to suggest using federal grant funding to overbuild a cable company. If not, then a definition of broadband as 20 Mbps upload is more of a suggestion than a rule.

Another issue with setting definitions of speed is that any definition of speed will define some technologies as not being broadband. That brings a lot of pressure from ISPs and manufacturers of these technologies. This was the biggest problem with the 25/3 Mbps and DSL. While it is theoretically possible to deliver 25/3 Mbps broadband on a single copper wire, the big telcos spent more than a decade claiming to meet speeds that they clearly didn’t and couldn’t deliver. We’re seeing the same technology fights now happening with a 100/20 Mbps definition of broadband. Can fixed wireless or low orbit satellite technology really achieve 100/20 Mbps?

Another issue that has always bothered me about picking a definition of broadband is that the demand for speed has continued to grow. If you define broadband by the speeds that are needed today, then that definition will soon be obsolete. The last definition of broadband speed was set in 2015. Are we going to wait another seven years if we change to 100/20 Mbps this year? If so, the 100/20 Mbps definition will quickly become as practically obsolete as happened with 25/3.

Finally, a 100/20 Mbps speed is already far behind the market. Most of the big cable companies have recently declared their basic broadband download speed to be 200 Mbps. How can you set a definition of broadband that has a slower download speed than what is being offered to at least 65% of the households in the country? One of the mandates given to the FCC in the Telecommunications Act of 1996 was that rural broadband ought to be in parity with urban broadband. Setting a definition of broadband only matters for customers who don’t have access to good broadband. Do we really want to use federal money in 2022 to build 100 Mbps download broadband when a large majority of the market is already double that speed today?

Trying to define broadband by a single speed is a classical Gordian knot—a problem that can’t be reasonably solved. We can pick a number, but by definition, any number we choose will fail some of the tests I’ve described above. I guess we have to do it, but I wish there was another way.