It's nice looking, but I'd like to see some actual user testing on whether this kind of progress bar performs its most important task.

"What?" I hear you asking, "Of course it does! It shows how much longer you'll have to wait."

In which case I must inform you that giving the user information is not the most important function of a progress bar. The most important function of a progress bar is Time Travel.

That is to say, subjective time travel. Try this fiddle: http://jsfiddle.net/gUkgX/1/embedded/result/

First, click "No feedback". Wait for it. Feel how long it takes to run. Try not to count the seconds (you wouldn't usually do that on a real website) - just see how it feels. Next, click "Spinner", and then finally try "Progress bar".

Which one felt shortest?

Of course, they are all the same ten seconds. Yet the spinner feels faster than no feedback, and the progress bar feels faster still. Watching that bar fill up makes the time seem to pass faster.

And the progress bar is totally fake. In fact, it doesn't even take 10 seconds to fill. It fills in 9.2 seconds, starts out fast and slows down, and none of this has anything to do with what is really happening, which is that you're waiting for a 10 second timeout to fire.

My point with all of this is that in order for a progress bar to fulfill its duties as a time condenser, I suspect that it needs to be prominent. I get the idea of keeping the indicator out of the way and minimal, but I think this is misguided in this case.