Understanding the Metrics of Internet Broadband Access: How Much Is Enough?

Abstract

Typical offerings of wireline broadband access are tiered according to the maximum speed the user will be able to achieve. Speed tiers have emerged as an approach to segment the market, but leave consumers with a puzzle. What speed tier will be sufficient for their needs?

A recent law now requires that providers of broadband access describe their product offerings in a uniform way, using a format that is modeled on a nutrition label. The labeling includes a specification of pricing, and also several performance parameters, in particular downstream and upstream speeds, latency and packet loss rates. The implication of giving this information to consumers is that the consumer will be able to make sense of it. However, in our view these numbers will just increase the degree of puzzlement, because there is no context to interpret them. Measures of technical performance can only be understood in the context of their impact of the quality of the user experience.

One element of a real nutrition label that is missing from the proposed broadband nutrition label is an analog to the Daily Value (DV), which allows the purchaser of a food product to roughly compare the nutritional value of the product to their overall daily requirements. Without some companion to the broadband nutrition label that gives consumers some baseline against which to compare the numbers, we think the reporting of performance numbers may do more harm than good. We believe that the FCC, which is responsible for the design of the broadband nutrition label, must augment the broadband label with some analog to the DV, which we call the Satisfactory Service Level, or SSL. The reported values, to be effective, should be simple enough for consumers to understand, meaningful, and independently verifiable.

In this paper, we draw on data about application performance, including models of application requirements and actual measurements of performance, to shed light on the question of “how much is enough”. Our goal is to give some insight, in a form useful to non-technical readers, of how to interpret metrics such as latency, and suggest how an SSL might be structured. We explain why the answer to the question of “how much is enough” is not always simple, and suggest how the FCC might balance the objectives of simplicity, providing meaningful information, and allow independent verification.

Keywords: Internet performance, broadband, internet measurement, network neutrality, latency, congestion