This strikes me as burying the lede -- if TMobile is traffic shaping based on what looks like the speedtest folder, are they also QoS-prioritizing that traffic to get better speedtest results?
That is a great point, but kind of outside the scope of his article. I think using the /speedtest/ folder as a test across the same resource would provide some clues.
s3/asset.jpg
s3/speedtest/asset.jpg
Of course, they might whitelist using speedtest but give priority based on originating domain (which would obviously make a billion times more sense).
Maybe. Louis Rossmann demonstrates a definite difference between speedtest and youtube[1]. The comments suggest it might be "binge-on" throttling, not favoring speedtest per say.
s3/asset.jpg s3/speedtest/asset.jpg
Of course, they might whitelist using speedtest but give priority based on originating domain (which would obviously make a billion times more sense).