Because the are unrepresentative at best and in many cases inaccurate. For context see Bauer, S.; Clark, D. & Lehr, W. "Understanding Broadband Speed Measurements," Massachusetts Institute of Technology, 2009. Part of the larger problem is that most people don't understand that you can't measure the "speed" of a network connection as simply as one measures average velocity of a physical object. It is much more complicated and abstract.
As for your comment on "HARDWARE," it isn't purpose built hardware but it is just a modified OS running on commodity hardware. It is a workaround to avoid inaccuracies introduced by poorly setup home networks, poorly maintained PCs and it also allows for consistent tests to be run over a period of time. Maybe you have not noticed that you Internet connection tends to be a little faster at 2am than it is at 7pm.
Slashdot Mirror
Because the are unrepresentative at best and in many cases inaccurate. For context see Bauer, S.; Clark, D. & Lehr, W. "Understanding Broadband Speed Measurements," Massachusetts Institute of Technology, 2009. Part of the larger problem is that most people don't understand that you can't measure the "speed" of a network connection as simply as one measures average velocity of a physical object. It is much more complicated and abstract. As for your comment on "HARDWARE," it isn't purpose built hardware but it is just a modified OS running on commodity hardware. It is a workaround to avoid inaccuracies introduced by poorly setup home networks, poorly maintained PCs and it also allows for consistent tests to be run over a period of time. Maybe you have not noticed that you Internet connection tends to be a little faster at 2am than it is at 7pm.