[LINK] Why is European broadband faster and cheaper?

[Ash Nallawalla]

> From: Paul Brooks

> This tells me:
> a) Speedtest.net doesn't measure latency very well, yet these were the
> figures used in the Berkman study
> b) There must have been a huge number of tests from Australia that did not
> use servers in Australia, or were sitting on the end of satellite links,
logging
> extremely high latencies, to average out the measurements with a more
> reasonable 20 - 40 ms to 'fairly local' servers to arrive at an average
latency of
> 160ms for AU
> c) All those Speedtest.net measurements that logged latencies of 160ms+
> are likely to have polluted the average download speed measurements,
> dragging the average lower than reality.
> 
> It would be fascinating to get the Speedtest.net data for AU, with all the
> tests with latency > 100ms cut out, and see what the averages come to
then.

Speedtest.net needs some improvement. I am in Melbourne, but am unable to
refine my location to Melbourne, since the IP address (iiNet) resolves to
Sydney, at least according to the tool. I am with Netspace, although it was
acquired by iiNet. I used the Melbourne-based Micro21 server but the
"distance" for my test was shown as 700 km. Hence it could be making
assumptions when it gave me an A+ (97%) score.

Ash