Bias in speed tests

I almost forgot one funny little discovery I made about speedtests. Before I get to that though, let me explain a little abotu speedtests. As the name implies, they are designed to determine the upload and download speeds of whatever internet connection you happen to be using at the moment, usually in the form of some little web applet. I tried them all when testing my new FiOS line out. I'm paying for 15 Mbs down and 2 Mbs up, and I'm told I share my line with no one, so any decent speedtest should reliably show that these are the speeds at which I am connecting. The same logic is true for dial-up and DSL connections, but not cable connections, which are usually shared among many users and line speeds drop as more users logon. Anyway, back to the subject. As I said, I tried many speedtests; Speakeasy's, the dslreports tests, Visualware's, and of course, Verizon's.

So here's the deal, most of the speedtests out there aren't designed to give accurate readings for internet speeds much above what you get with a cable modem (~5 Mbs). This is in fact what I discovered with most of the speedtest sites except for the one from Visualware, which pegged me at 15.1 Mbs down and 1.8 Mbs up, a fair margin of error. The most interesting result came from my FiOS provider themselves, Verizon. According to them, I'm getting a blazing 49 Mbs down and 6.1 Mbs up! That's nearly 35 Mbs extra down and 5 Mbs extra up speed they are giving me for free apparently. That or they are just lying to try and make me feel that much better about this FiOS service? hard to say, but if you don't believe me, I took a screen shot of the output page (but unfortunately, the graphic didn't come out).



I just tried this test again now, and its telling me I'm maxed out on both down and uploads. Methinks a bug lies within.