US broadband speeds keep getting faster. Over the last year, Speedtest by Ookla says the average fixed location download speed increased by 37%, from 68 to 93 Mbps. Mobile rose at a slower, but still respectable rate of 22%, from 23 to 28 Mbps. In the previous year, fixed had risen at a higher rate, 48%, and mobile at a slightly lower one, 21%.

These figures reflect the consumer’s point of view as they’re taken from tests consumers choose to run that aren’t protected from effects such as WiFi speeds or computer speeds. While they’re influenced by network performance, the stronger factor is the pricing and positioning of ISP broadband plans.

This becomes clear when they’re broken down by ISP, as PC Magazine does in its annual report on the fastest ISPs. This report declares Verizon FiOS the fastest major ISP (139 Mbps) and Longmont, Colorado’s NextLight the fastest minor at 278.

What the Data Mean

Verizon and NextLight offer multiple service tiers, so their average speeds tell us which plans people are buying. NextLight offers two tiers, 1 Gbps and 25 Mbps, so their average speed indicates that four consumers choose the 25 Mbps tier for each one that goes for the higher speed.

Verizon is a bit more complicated because it offers three tiers: 100, 500, and 940 Mbps. Its average speed suggests that 16 customers are choosing 100 Mbps for each one that opts for one of the higher speed plans (assuming that 500 and 940 are equally popular.)

Fiber has an advantage over asymmetric systems such as cable and DSL in that fiber systems tend to offer the same speeds for upstream and downstream use. But consumers are not convinced that the higher speeds – especially upstream – are valuable enough to warrant higher prices.

A Litany of Errors

PC Mag’s write up of their speed test makes a number of errors. It starts off by claiming the Akamai data showed an average Internet speed of 7.2 Mbps in the first quarter of 2017 (the last time Akamai reported Internet download speeds.)

But Akamai didn’t say that. The writer confuses Akamai’s report on average TCP stream speeds with “Internet speed”, a grievous error. As Akamai has said, “Internet speed” is best measured by its Average Peak Connection Speed because average connection speed reflects the division of bandwidth across multiple measurements.

PC Mag is also confused about the role of competition in raising speeds. We mostly get higher speeds for fixed broadband from ISPs discontinuing old low-speed plans and upgrading consumers to higher speeds involuntarily. You’ll note that you can’t buy a tier from Verizon that provides less than 100 Mbps any more.

While that’s a competitive response to the fact that Comcast is an option in most Verizon markets, competition doesn’t explain why fixed speeds rise faster than mobile speeds to. Mobile is a much more competitive market where speeds increase at a lower rate than they do in the fixed market. This is a key point.

Investment in Networks Drives Speed Increases

Network performance is determined by the technology embedded in network devices. Many analysts – most of them non-technical – have long opined that wire is the most important such technology; they want optical fiber everywhere.

But we’re now learning that several media are capable of handling the gigabit speed that will be the next plateau. Coaxial (TV) cable and high frequency wireless can get the job done. Coax does it over miles, while wireless is limited to a thousand feet or so.

The equipment to drive high speeds over less-desireable media costs more than comparable gear for fiber, but not that much more. Network equipment has a short lifetime, but its upgrade costs are built into your monthly bill. As long as carriers can make reasonable profits, we can expect speeds to increase year after year.

Recent Increases are Greater than the Norm

Over the last ten years, fixed broadband speeds in the US have increased by an average of 35% per year. The increases we’ve seen in the last two years – 42.5% per year – are well above that norm. It’s reasonable to believe that this higher rate of improvement is driven (at least in part) by expectations created by recent regulatory changes.

Fixed broadband also sees some competition from wireless. This happens despite the fact that fixed has higher top-end speeds because consumers realize they can have a perfectly acceptable broadband experience at 25 Mbps or even less.

Operators realize that 5G will bring additional options to the residential market. T-Mobile has shared a desire to become the nation’s fourth largest ISP by 2024, based solely on the 100 Mbps that its 5G network will enable. Given that 100 Mbps is now the US average for fixed networks, this is a very doable plan.

Reacting to Potential Competition

Technology markets are different from conventional ones. Tech firms are building products today to stave off competition in the future. So fixed network are raising their speeds to create some daylight between advanced wireless networks.

5G isn’t here yet, but it’s already stimulating a competitive response from incumbents. Better broadband plans and prices than we might see without this looming threat are entering the market.

That’s a wonderful thing, but we also need to see some apps that can exploit higher network speeds. There’s a lot that can be done on a 100 Mbps network that can’t be done at 10 Mbps. So let’s get with it.

Some of these new apps will use bi-direcational video very heavily. In addition to needing a lot of bandwidth, these applications will require more stringent control of network latency and jitter. This is an avenue that appropriate regulation can also enable, so there’s work for both the public and private sectors.