Comment 283 for bug 532633

As far as I know, there have been no papers or research published on window button placement, or the advantages/disadvantages of any of the suggested positions and combinations.

What we do know already is that it appears that MS arbitrarily chose the top-right position, and Apple chose the left because they didn't want to follow MS. As with many decisions taken in the infancy of UI design, it was probably a "well that seems alright" or a "can't think of anything better right now" type of decision. These decisions have stuck with us for many years now, so there is much to be said for keeping it that way. It may just happen to be that the initial top-right position is in fact the best or close to the best position possible, and that the initial designers just got it right first time.

There is a natural ease of movement of the arm and wrist of right-handed mouse users in an arc roughly centred on the elbow/wrist area. A natural pivot-point.
This means that the top-right position for most-used buttons is an easily reached one. Top-left buttons would appear to be reached via a non-natural wrist movement, or a movement requiring small movement of the whole arm which entails more effort from the user. There are of course a series of mouse-movement enhancers (like acceleration) in the software which are designed to mitigate these problems.
There is a natural tendency to "throw" the mouse pointer towards the known position of a button, and anything which makes this easier or harder will be embraced or rejected by users.

There is also the principle of positional constancy to be considered.
If there is no conclusive evidence to show that repositioning buttons will be advantageous, then it is better to keep the status quo. There is a saying "if it ain't broke, don't fix it". People often quote this for good reason: they don't like change.
Even if something is "broken", there is often such widespread acceptance of the "broken" item, that changing it causes more alienation and or user rejection than the efficiencies a correction or fix would create.
A good example is the 'qwerty' keyboard.
The layout was set when manual typewriters could not keep up with the speed of key strikes by operators. The 'qwerty' layout is designed to slow down the user. This obviously does not now apply in the digital age, yet changing the vast majority of the populace away from a known bad design, to a proven better layout (e.g: dvorak) would be extremely difficult if not impossible.
Any hardware manufacturer insisting on this would most likely see a large drop in sales.

Keeping it all in perspective, as window buttons are probably not on the same level of user-interaction importance as the main interface device (keyboard), I would be all for change if there was good evidence showing improved window management with button repositioning, with only relatively small retraining or disruption involved. However, as I mentioned before, there does not appear to be any research to support such a change.
I would suggest there is a whole PhD in such a research topic!

If a window button layout, which is different from the accepted status-quo, is imposed upon users, then it is my considered belief that this will be detrimental.