Using a tripod would have made this article irrelevant, nobody ever actually uses a mini-tripod when shooting with their phone.
If the cameras are subject a random motion which causes blur greater than the blur caused by low-light conditions then the results are meaningless as the random motion is independent of the variable being tested.
This post wasn't about the quality of the sensor it was about the quality of the photograph.
My point is that regardless of if a 4S shot better photos, the photographer could hiccup and get a blurrier photo than they might have taken last with a 3GS. In a less extreme circumstance, they could have just exhaled instead of hiccuping, in which case no one would have noticed and the error could be incorrectly attributed to the camera.
A great photograph with a camera phone requires a certain set of parameters that can almost entirely be individually measured outside of human interaction (and randomness). If you want good handheld performance, you need a fast exposure time, which can be measured from the EXIF data of each photo. If you want high dynamic range, measure the ratio of pure white and black pixels to the rest of the image (white and black is basically over/underexposed). The only thing that would be effected by motion that I can't think of how else to measure beyond being handheld is time to good focus (outside of crazy contraptions that result in more repeatable motion, like sticking your phone in a centrifuge), but there probably exists a way to do so while avoiding the inherent non-repeatability of humans behavior.
With a tripod there would be no blur, but the picture quality would still shine through. The amount of noise difference in each phone's photos is very noticeable.
In an experiment all the conditions should be the same except for the variable, which in this case is the phone.
The motion blur is significant because it speaks to the increased aperture/ decreased exposure time in the 4S, which are both part of the phone wouldn't you agree? On a tripod these aren't crippling to the quality of the photo like they actually are.