This is very hard to explain without diagrams, but I don't think the fanciful explanations we're getting ourselves into are really necessary. Consider an analogue in translational motion first:
A ball on a string, at rest, will accelerate toward you if you grab the end of the string and pull on it, making it taut.
Now, what about a ball with some initial translational momentum? You pull on the string in a direction perpendicular to the ball's initial velocity, and voila! It suddenly begins to move in a circular loop about your hand. But it's distance from your hand does not change! You are pulling it toward you...why is it not responding as in the first case? Is it defying the tension force? Obviously not...we know that the tension force acts as a centripetal force. A simple analysis of the vectors reveals why the object with some initial translational momentum "behaves" differently under some applied force than an object at rest.
The same is true in the rotational case. Consider a rotating bicycle wheel. with a string attache to the centre. When it has no angular momentum (it is not spinning), and you try to hold in in the vertical orientation, it will fall back down:
|
| <--- string
|
| / <--- wheel
/
side view of the wheel, hanging from a string, almost vertical. Why does it flop back down? In the position shown above, the centre of mass is no longer directly below the pivot point (where the string is attached), so the gravitational force exerts a torque that rotates the wheel back down to this position:
_______|
_______| <----string
_______|
_______|
_____------ <---- wheel
Now, what happens when the wheel is spinning (ie it has some inital angular momentum pointing roughly to the right, for example. The very same torque (which points into the page), has a very different effect...that angular momentum vector does not decrease (because the torque is perp. to it), and it changes direction...so the wheel remains vertical even after you let go! But it precesses (rotates about an axis // to the string in this case):
|
| <--- string
|
| / <--- wheel (now spinning, so it remains in this orientation)
/
So, in a direct analogy to the previous example, an object with some initial angular momentum reacts differently (and in a non-intuitive way) to an applied torque than an object initally non-rotating does. For those of you (probably everyone) who couldn't make heads or tails of my stupid diagrams, please check out the video here, and read the article...it explains gyroscopes quite well:
http://science.howstuffworks.com/gyroscope1.htm