I have gone through the report uploaded by baronic, and it's something I bet the people at Star Tech knew long ago. It's just very basic info of how the line might get affected (eg: the copper line thing) and tests not detailed enough.

In my humble opinion, I suggest:

Get a control test. Round up /k/ armies according to their area, and do the speed test again, but this time include more detailed information instead of just the end result. Like individual IP number with date/time of test, targeting screamyx specific routing server, or host.

Here's a random set of hosts I identified, which were visitors to my blog.
67.101.49.60.klj04-home.tm.net.my
106.68.49.60.klj04-home.tm.net.my
235.168.48.60.cbj06-home.tm.net.my
156.189.48.60.cbj01-home.tm.net.my
203-59-190-136.dyn.iinet.net.au (not screamyx lulz)

That will narrow down how the line was distributed in an area. No matter how dynamic your IP can be, it will still go through the same host before you can connect to other things.

Other info like weather condition, router/modem model, could help paint the bigger picture better. Like a group test of Cheras during rain versus Banting sunny day. Something like that.

After that, do the same control test with P1, as a comparison.

To demand better service, concrete evidence of the current sucky service must be shown. Get the raw data ready and cook up some real info, and screamyx will shit brix once you bring it up to the press (provided the press actually does press it).

Else, be an internet hero and blog about it.

p/s: I might be wrong with the tech-info, you might want to double check on it.

because it's normal.

bring up an abnormal one, and they might just do something about it.