Time has passed, and hopefully some passions have eased. So perhaps we can re-analyze the question of bending the bars 2"?

Let me confirm that I haven't bent a set of bars, or a stem, in such a fashion. I have no intention of starting now - aluminum isn't good at fatigue resistance, and I don't provoke it just for grins. My original post should have said that one can't bend the bars 2" without causing permanent harm. It's pretty tough to prove a negative condition - for example that one can't bend the bars 2". But that really doesn't matter.

What matters is that it’s destructive to bend the bars 2". And, that there is no need to do so while riding. If someone is experiencing breaking stems or bars, that someone is doing something wrong.

Looking at Danno's pictures, it ought to be clear how hard it is for him to stabilize the front wheel enough to try and bend the bars. If one is trying to use upper body strength to help provide a base against which to pedal, then the correct form is to push the bars from side to side, not to rotate them around the axis of the stem's extension. Having said that, people will often ride as they do and blame their equipment for failures.

Having said this, a larger diameter extension will better resist motion, hence fatigue, better than an otherwise equal skinny one. Therefore it will better resist fatigue failure, up to its structural limits. But, this is only important if during proper use, the narrower stem can't adequately resist motion and fatigue.

Now, reviewing the pictures provided, I continue to challenge the assertion that they demonstrate 2" of motion by the handlebars. Before going off, consider the following:
1) The location of the camera has moved (look at the box or crate on the floor behind Danno on the left (his right).
2) The direction the bike is pointing is relative to the camera has changed between the pictures.
3) The frame is leaning to the left in one picture, and to the right in another picture.
4) The fork is turned relative to the frame; in one picture to the left, in the other to the right.
5) Mysteriously, where the top-tube exits the bottom of the picture, it is neatly centered in both pictures. Given all the other motion, this looks to be the result of cropping - to be polite.

No these pictures prove nothing because we can't see the effects of foreshortening. That the visual distance is different from the two ends of the handlebars to the fork crown is beyond question. Does this translate into actual distance? It’s not clear; we don't understand the relationship of the camera angle to each of the pictures. It seems likely that the pictures, as presented exaggerate the differences in real distance. How much does this exaggerate? We don't know, but likely a good bit.

Consider a couple of further points:
1) The computer, mounted on the stem, seems to move relative to the bars in the two pictures. Surely the bars aren't moving in the clamp? Is it fully tightened?
2) The torsional resistance of the stem's extension should be twice its bending resistance. There is no sign of bending. Similarly, the handlebars show no sign of bending. Is it possible to get all of this perceived motion simply from twisting the extension, without part visibly bending? I don't think so.

So, to summarize:
1) Don't bend the ends of your bar's 2" - you're liable to become a statistic.
2) Don't assume that a picture, just because it’s worth a thousand words, proves anything geometric until you've made sure of what you're seeing.
3) Doubt that stems have the twisting flexibility to allow the ends of the handlebars to be moved 2" up or down.

What's the real point of all this? There has been a good deal of poor logic wrapped around emotion in many of the point raised in this thread. The marketers have most of us in their grips, more than we like to acknowledge. One's objective statements have to be taken in context to determine their truth value. That the typical threadless stem is stiffer than typical quill statement is true. But, if we instead said: "The typical quill stem better absorbs shock than the typical threadless stem" we would also be telling the truth. Both statements convey non-objective opinion. The first suggests that greater stiffness is desirable. The second suggests that more shock absorption is desirable. And, there are no published data regarding the stiffness or shock absorption of stems, much less a clear body of information of how much of which benefits riders most. So adding the subjective content to a true statement diminishes the truth-value of that statement.

Enjoy what you have. Be cautious of experts (including myself). Ignore the marketeers.