The idea that the real reason they download all media queries is because many CHANGE devices have their responses to those queries after they boot.
The iPhone5 image that is in the portrait when the page loads (the message is “width” like 640px, but not “portrait”, unfortunately, iSeries does not support these requests) ... then you decide to turn the iPhone sideways, and the browser now activates pseudo-landscape mode (again, it fires from width @ 1126, not "landscape").
Most likely, responsive web design was designed to feed various stylesheets into a browser displayed at 640 (rather narrow, probably a phone / tablet) than a browser displayed at 1126 (more likely on a laptop).
If he had not bothered to download additional media request sheets, he would suddenly have to stop, remove the HTTP request, wait for the sheet to load, and then analyze it for display. This can lead to a rather ugly delay.
Since most browsers follow a code reuse pattern, and, for example, Webkit or Gecko core boxes may not know if they are on a laptop or tablet (as if these lines do not start to blur in any case), it simply loads each multimedia request regardless whether he chooses to display it.
While this saves every browser from the bad, it generally breaks up a good piece of the utility for media queries.
A cell phone or a cheap Android tablet should not download additional files (especially in limited data plans) that it simply does not need.
My projects currently use media queries, but I use them sparingly. Most of the requests to my sites are implemented by downloading the necessary files to remove this waste. The remaining requests are used in cases when javascript is disabled or for sheets that need to be loaded “just in case” (for example, my 640px layout usually loads, since most devices can display it in a particular situation).
If anyone has a better, cleaner way of handling this, please let me know.
In the meantime, if you can just come up with functionality that can get around this (maybe android-style manifests built into browsers?), You might need to drop the line into Mozilla or Chromium commands ... it seems like they could use their hand on this one.