Comment by chrisweekly

Ok, I think we're actually in agreement -- given your all-important qualification "for the range 200ms-1s". Yes, ofc, that first part of the curve above the drop is quite flat; there's hardly time for the user to get impatient and bounce.

My point about the shape of the curve stands. 100ms can matter more on the steepest part of the slope than 2s does further to the right.