> as far as I know, the uncertainty principle means that as the signal in the time domain gets more localized, the signal in the frequency domain gets more spread out (and vica versa).

That is exactly right.

And you can also tell a similar story for the position-velocity relationship: if you make a wave by throwing a rock into a pond the resulting wave front will be circular, i.e. the position of origin of the wave is well defined (it's where the rock landed), but its direction of travel is not.

By way of contrast, if you want to make a wave with a well-defined direction of travel, the only way to do it is to make a very wide wavefront, i.e. one whose position is not well determined.

A simple example of a "measurement effect" is the quantum zeno effect [1]. You can steer a system by making a series of measurements and gradually transitioning the measurement axis from aligning with the initial state to aligning with the target state [2].

For example, placing a diagonal polarizer between a horizontal and vertical polarizer will allow light to pass through [3]. As you use more and more polarizers that gradually transition from horizontal to vertical, you converge on a mechanism for rotating horizontally polarized photons into vertically polarized photons without loss.

1: https://en.wikipedia.org/wiki/Quantum_Zeno_effect

2: http://algorithmicassertions.com/quantum/2015/10/26/Quantum-...

3: https://youtu.be/PJHCADY-Bio?t=6m45s

Coming from an audio background, this explanation clicked immediately. You can't determine anything about frequency from a single audio sample. And conversely, you can't determine anything about signal phase from a frequency spectrum.

Is this just an apt analogy or is it basically the same concept at work in QM?

Observer effect.

Thanks :) That gives me something concrete to google