The only valid criticism to emerge is that measurement error might be greater than I allow for. The solution is to repeat the experiment on control cases, and I've said I will do that.
Ace, some criticism of you paper has been unfair, and I believe your method is roughly correct. However your stabilisation effort is very poor, and your tracking of the plane is very poor. Please take this from an expert in stabilisation and tracking. As I said, it's my job.
Get uncompressed video. This is very important. Compressed video is just a computer generated reconstruction of the original, with most of the original data lost.
Do not enlarge the footage by 200%. This merely adds useless data which will distort the results.
Edit: Also, do not tweak the contrast or brightness, or sharpen or otherwise 'enhance' the picture. All of these things destroy information, not recover it. There is NOTHING you can do to an image to make it more accurate.
Try different tracking strategies. I've done a fair amount of manual tracking and I can tell you simply stepping through the frames and placing the mark where it looks like it should go does not give a good track - not even good enough to fool a casual observer. As you're not a pro, get ten volunteers to track and stabilise the video. Eliminate all subjectivity. See if their results match, or if you get any correlation at all. Try averaging the tracks.
I did earlier suggest using an automatic tracking program, but having seen the footage I doubt it will provide any benefit. To get a good accurate track you need to be tracking something with a 2 dimensional shape, roughly an x shape (this applies to manual tracking too). A plane is perfect because the wings form an x with the fuselage. However, this x shape is only visible in about 8 frames of your video, thanks to the logo that's in the way. The nose of the plane, is almost 1 dimesional. It's easy to track it up and down, but left and right tracking will be innaccurate. Secondly there is a lot of jitter along the edges of the building. Just look at the black edge of the frame. It's supposed to be perfectly straight. To me some of those pixels look to be 4 or 5 pixels astray.
In conclusion Ace, the image itself is distorting, up to 5 pixels either way. Your stabilisation is easily 2 pixels out on some frames. Your track of the plane is easily two pixels out on good frames, and hard to judge on bad frames. In total that's a nine-pixel potential measurement error. Yet, you're convicting based on a claimed 4 pixel discrepancy.
I hope that helps.
Shrinks,
VFX Supervisors for Truth
PS, I've put many CGI planes into filmed shots. Evening winning awards for such acts. If you'd like to learn the truth about these processes fire away. If you'd prefer to continue guessing, then please carry on with watching those DVD extras.