Does Your Broadband Internet Service Suck?

Young man using smartphone iStock_000037506278_XXXLargeMuch of the debate over Internet policy in the US presumes that our broadband services aren’t very good. Would there be a need for the FCC to pass the most prescriptive regulations for broadband ISPs in its history if there weren’t something wrong with the services we get, the prices we pay, or the ability of these services to support the applications of yesterday, today, and tomorrow? Probably not.

It’s more or less conventional wisdom that the US has fallen behind the rest of the developed world in broadband. If you read the popular press or the leading technology blogs (not to mention blog comments) you’re subjected to a constant stream of complaints to the effect that the US has slower broadband speeds at higher prices than dynamic tech powerhouses Korea, Japan, Sweden, Hong Kong, and Singapore, and even sleepy backwaters like Moldova and Romania.

As with most popular myths, there’s a kernel of truth in this allegation. According to the Akamai State of the Internet report, the “Average Connection Speed” (which doesn’t mean what you think it means) in the US is a mere 11.5 Mbps, ranking 12th in the world in the third quarter of 2014, the most recent reporting period. According to the Ookla Netindex survey (AKA “Speedtest”), the average download speed in the US is a much more robust 33.6 Mbps, but that’s only 27th in the world, behind the Asian Tigers, Scandinavia, the Baltics, and several former Soviet satellites.

Before we leap to the conclusion that these figures are cause for alarm, it’s worthwhile to understand what they mean and to resolve their inconsistency.

First, there’s the question of speed. Is the average speed of American broadband connections closer to the Akamai figure, 11.5 Mbps, or the Speedtest figure, 33.5 Mbps? A variation of 300% doesn’t come down to measurement anomaly; it’s an indicator that the tests are measuring fundamentally different things.

In the case of Akamai, “Average Connection Speed” doesn’t assess the speed of our broadband connection to the rest of the Internet, it’s a measurement of an internal speed of the software connections between browsers and parts of web pages. Akamai is in the business of speeding up web browsing – it’s not unreasonable to say the company is in the “fast lane” business – so they analyze web browsing performance in detail. A web page consists of several elements – each graphic is an element – and several load at the same time.

The typical web browser tries to load four to eight elements at the same time, so it’s impossible for each one to load as fast as the broadband network will allow. If the capacity of our networks is really 33.5 Mbps as Speedtest claims, and page elements load at an average speed of 11.5 Mbps as Akamai says, that simply means that web browsers are loading an average of three elements at a time: 11.5 times 3 is 34.5, pretty close to the Speedtest average. There are some nuances in TCP that easily account for the missing megabits, such as the “Slow Start” congestion management technique. So it’s reasonable to say that US broadband networks are as fast as Speedtest says they are.

But what about the question of ranking, is the US 12th in the world or are we 27th, and what difference does it make?

I address that question in the follow-up, “Where Does US Broadband Speed Rank?” I also provide a more in-depth analysis of these questions in my paper, “G7 broadband dynamics: How policy affects broadband quality in powerhouse nations”.