Improve the median, not the mean

I like seeing Google Fiber prompt incumbent dinosaurs into boosting speeds. It’s great.

But: it doesn’t matter much. I have 10x the bandwidth at work (gigabit) than I do at home. I barely notice it, however, for both technical and perceptual reasons.

In fact, my home internet recently improved 3x from ~30mbps to its current ~100mbps. Maybe videos achieve HD a little more quickly? Maybe?

Point being, I had decent bandwidth for my use cases, and the new bandwidth — which I am intellectually thrilled about — offers diminishing returns of actual utility.

Because I already had above-average bandwidth, these improvements move the mean more than they move the median. This disparity offers a clue about utility.

There’s more bang for the buck in improving the lower percentiles of performance. Moving a substantial number of users from 1mbps to 10mbps is a greater increase in utility than my recent improvements.

Google has the right idea here with QUIC. It offers a substantial improvement for the worst 1% of latency, which nudges the median upward. Let’s focus on that.

Published by Matt Sherman on April 20, 2015