Your Mileage May Vary or Performance You Can Count On: What Should Broadband Consumer Labels Measure?

25 Pages Posted: 2 Aug 2022 Last revised: 14 Dec 2022

See all articles by Henning Schulzrinne

Henning Schulzrinne

Columbia University - Department of Computer Science

Walter Johnston

Danu Consulting, LLC

Andreas Carlos Freund

Columbia University - Department of Computer Science

Date Written: August 1, 2022

Abstract

The FCC-proposed broadband consumer labels are designed to enable quality and price transparency, and thus competition, for residential broadband services. Key quality indicators for residential networks are download and upload speeds, typically the only quality-of-service differentiator disclosed at the time of purchase. Even as speed remains the most-commonly accepted quality-of-service indicator, its use in advertisements has been problematic. Often, the fine print indicates “up to” speeds, and the terms of service often disclaim any warranties on the speed advertised. Actual achieved speeds can vary significantly depending on the measurement methodology and on how the statistical variation in speeds across time and space is represented. We have compared several different ways of measuring speed and performance using the most recent FCC Measuring Broadband America (MBA) data, from 'classical' measures such as mean and median, to percentiles and the MBA 80/80 consistency metric. Our results show that well-performing networks can offer speeds close to or above their advertised number almost all the time and across all of their measurement locations, with 5th percentile measurements differing only marginally from the median or 80/80 speeds. Therefore, we argue that leaving the measurement method of speed to providers will likely lead to the use of the most favorable speed metric possible, particularly since the format will confer a notion of government-approved legitimacy on the numbers presented. Low-percentile cut-off metrics reward high-quality networks, without having to disclose hard-to-understand metrics such as packet loss, and makes it far less likely that a new customer finds themselves residing in a service territory with unreliable service.

Keywords: Broadband Consumer Labels, Broadband Internet Access Service (BIAS), Open Internet, Internet Service Provider (ISP), Network Measurements, Network Reliability, User Experience, Telecommunications Regulation

JEL Classification: D18, D82

Suggested Citation

Schulzrinne, Henning and Johnston, Walter and Freund, Andreas Carlos, Your Mileage May Vary or Performance You Can Count On: What Should Broadband Consumer Labels Measure? (August 1, 2022). Available at SSRN: https://ssrn.com/abstract=4178758 or http://dx.doi.org/10.2139/ssrn.4178758

Henning Schulzrinne

Columbia University - Department of Computer Science ( email )

New York, NY 10027
United States
2129397042 (Phone)

Walter Johnston

Danu Consulting, LLC ( email )

10013 Dianella Ln
Austin, TX 78759
United States

Andreas Carlos Freund (Contact Author)

Columbia University - Department of Computer Science ( email )

New York, NY 10027
United States
+14154252491 (Phone)

Do you have a job opening that you would like to promote on SSRN?

Paper statistics

Downloads
106
Abstract Views
757
Rank
479,942
PlumX Metrics