The problem is at the philosophical level you will find discussions about the observer observing a system in a sense unpredictably jolting it and leading to an imprecision in the measurement. This is highly vivid, pictorial and played a role in the early development of QM.
The thing though is it is wrong. It really is a deduction from the fundamental postulates of QM for certain observable's such as position and momentum. You see in QM you generally can not predict the outcome of an observation but only probabilities. What the uncertainty principle is about is the spread of those outcomes. If you know the position of a particle with high accuracy (ie the spread of possible outcomes is small) then the spread of the possible outcomes of a measurement of momentum is large, and conversely. The philosophical issue is the principles of QM imply you can not know certain observable's with a high degree of precision at the same time - reality is unknowable exactly.
In fact even those conversant with the machinery of QM can fail to understand this being imbued with the early discussions based on vivid visualizations. Check out:
http://io9.com/5942921/scientists-now-uncertain-about-heisenbergs-uncertainty-principle
Captainmuon gave the correct analysis:
The uncertainty principle doesn't say that you always disturb a system when you measure it. That is a common misconception. It says that you cannot know, for example, momentum and position completely precise at the same time. Not because you have fat fingers, but because a particle just doesn't have those properties at the same time. *)
There have been a couple of nice experiments in the last years that go against "quantum intuition" and perform non-interacting measurements. Famous is the quantum bomb detector, that can tell (theoretically) if a single-photon triggered bomb will explode or not, without actually setting it off. Crazy stuff.
(*: Think of a perfect, infinite wave shape (°º¤ø,¸¸,ø¤º°`°º¤ø,¸,ø¤°º¤ø,¸¸,ø¤º°`°º¤ø,¸¸,ø¤º°`°º¤ø,¸,ø¤°º¤ø,¸¸,ø¤º°`°º¤ø,¸ ). This has a perfect, known wavelength, but no real position (because it goes infinitely in both directions). Contrast this with a localized wave, like on an ocean (,¸.•´¯`•.¸¸). You can now say where it is, and measure the wavelength, but not with so much accuracy as it doesn't repeat itself. That's basically the core of the uncertainty principle. If you want to know the details, all you have to understand is the Fourier transform. It really all makes sense and is not magic at all once you know the math.)
Thanks
Bill