Broadband Truth in Advertising – Regulation?

There needs to be a standard for truth in broadband advertising. Since the operation of the marketplace hasn’t resulted in a standard, a standard should be imposed. Note that standards are often helpful to markets AND that government often has a constructive role both in imposing and enforcing the standard. Enforcement of weights and measures was good for honest grocers who otherwise had a hard time competing with dishonest ones who could charge less per “pound”. Fuel and energy efficiency standards, even if not strictly predictive of real-world performance, are useful for comparing different models.

There are at least three dangers in a government imposed standard: it could and easily might be bent to favor powerful incumbents, compliance might be too costly for all but the largest of companies, and it could be so proscriptive that it discourages innovation. Nevertheless, in this case, especially with the LACK of a competitive market for broadband services in the US, I think we need to take the risk.

This is an example of a standard we might impose without, I think, unduly burdening honest ISPs (note that lost of my numbers are arbitrary but there need to be some numbers):

1. If an ISP advertises a speed in any direction, the speed in the other direction must be given as well and be advertised as prominently.
2. 90% of the speed advertised in each direction must be physically achievable by all customers of the service between their premises and a dedicated test point in the ISP network at least 10% of the time (not very onerous but meant to put some constraints on what maximum speed means).
3. The ISP must maintain connections to the rest of the Internet of sufficient capacity that the mean speed experienced by customers on a monthly basis in each direction is no less than 75% of the speed advertised and that speeds less than 40% of the advertised speed are not experienced more than 10% of the time averaged over all active customers on a monthly basis.

The final point is the crux of the matter and measuring it without being intrusive is not necessarily easy. Users do not always send at the maximum rate available to them and ISPs shouldn’t be penalized for this. Moreover, whatever server or other subscriber the user is communicating with may also not respond quickly enough to fill the pipe to the user even if that pipe is a s broad as advertised. This latter point MIGHT disappear in any large sample but I don’t know that.

It may perhaps be necessary to allow compliance monitoring of the third point not by actual user experience but by certified proxies (like test weights) deployed on the ISP’s network. It’s not a good idea to have users running speed tests just when they think things are slow (although I must admit I do that) because the speed tests themselves add volume to what is actually probably already a saturated network and don’t deliver a valuable payload.

There also needs to be a suitable escape hatch for startups who have no user experience to go on and for very small ISPs who won’t have a meaningful statistical base for measurement.

It is essential that the regulator neither tell ISPs what speeds to deliver nor how to deliver those speeds; either is sure to stifle innovation. Even imposing methods of measuring is dangerous since the methods may preclude or make it expensive to deploy new technologies.

Complicated once you start to regulate.

This post is about how one ISP manages his network to try to assure that users have an acceptable experience related to what is advertised.