I think david927's intuition is more correct here. The uncertainty in the position and momentum is intrinsic to quantum mechanics - it's built into the 'wave function'.
The suggestion that if one pushes away something by throwing something else builds on a purely classical intuition and wouldn't require quantum mechanics to explain if this was all we observed. The uncertainty in quantum mechanics is fundamental (to quantum mechanics) and emerges through a different, as yet unknown, mechanism.
3Blue1Brown has an extremely good explanation[1] of the intrinsic uncertainty, and why it's separate from measurement uncertainty. (the previous episode[2] is a recommended prerequisite for background on how the Fourier Transform works)
> emerges through a different, as yet unknown, mechanism.
In 3Blue1Bron's explanation[1], he shows how the intrinsic uncertainty is an inherent trade-off of trying to measure both position and frequency. A short wave packet only a few wavelengths long correlates with a narrow (precise) range of positions, but also correlates well with a very wide range of frequencies due. A Heisenberg-like uncertainty exists any time you are working with weave packets with length near the wavelength. 3Blue1Brown gives a very good example using Doppler radar.
Yes, I like these sources too. Good for building intuition. I would just add that Heisenberg-like here means that both systems share features of wave mechanics. Doppler type effects aren't quantum mechanical though.
When I suggest the mechanism is unknown, I mean that Heisenberg uncertainty is a postulate of quantum mechanics. In other words the fundamental reason that quantum mechanics should appeal to wave mechanics isn't really established - we don't really know yet the fundamental objects and interactions that lead to quantum mechanics (despite much effort).
I'm not sure about the historical part, but now the uncertainty principle is not an independent postulate. It's deduced form the non commutation of the operations to measure the position and the momentum of a particle. This can be done in the wave representation or in the matrix representation.
Moreover, similar calculations can be done with other measurements that don't conmute. One that is very important is the spin of a particle in the x, y, and z axis.
Another is the polarization of a photon in directions that are at 45°. For example, most of (all?) the experiments of the EPR paradox are done with polarization instead of position-momentum, because polarization is much easier to measure. https://en.wikipedia.org/wiki/EPR_paradox
It's a good point that uncertainty relations exist for all kinds of physical observables. But whether they're expressed as commutation relations or as in Heisenberg's original formulation, or whatever formulation you choose (wave mechanics, matrix mechanics, dirac representation, qft, or anything else one can think of) it's still asserted, rather than derived from an underlying set of fundamental physical objects and interactions.
Why unknown? Heisenberg's uncertainty principle can be derived mathematically, using a property of the Fourier transform. It has nothing to do with disturbing the system during measurement.
I'd say that's more a mathematical statement than physical derivation. The effort of subjects like string theory is to lay down fundamental objects and interactions from which other theories (quantum mechanics, gravity) emerge. But I don't think there is a final word at the moment of what the fundamental theories than result in quantum mechanics should look like.
The suggestion that if one pushes away something by throwing something else builds on a purely classical intuition and wouldn't require quantum mechanics to explain if this was all we observed. The uncertainty in quantum mechanics is fundamental (to quantum mechanics) and emerges through a different, as yet unknown, mechanism.