The earth is a rotating object.

The earth's total angular momentum is fixed, unless there is a significant interaction with some external system.

There are no external systems close enough to produce the kinds of forces necessary for changes of the kind proposed.

The earth has a fairly solid mantle, but under that there is a liquid layer, apparently mostly iron. There is also a pretty trivial gas layer outside, called the atmosphere.

Under the liquid layer, there's another core that's pretty much the same stuff as the liquid layer, but acts like a solid due to the extreme compression due to gravity (and the weight of all the crap on top of it).

The earth would have a fixed rotational speed if all the parts rotated together, but the liquid layer can have "eddy's" and whirlpools and the like in it. If the liquid changes direction, the "solid" parts must make an opposite change in order to maintain the fixed total angular momentum. If the liquid speeds up (unlikely, except locally) the rest of the system must slow down. If the liquid slows down, the rest speeds up.

If the liquid layer changes direction, as a uniform change or by "prevailing flows" in deviant currents, it's angular momentum in the direction that everything else is going is changed, so everything else has to change.

The magnetic field is produced by the motion of "conducting parts" of the system, and if everything moved together, the magnetic poles would be exactly coincident with the geometric poles of the axis of rotation. The "declination" (the distance between the pole of rotation and the magnetic poles) indicates that the net rotation of "something inside" (assumed to be the liquid layer) is not precisely in the same circles as the rest of the globs.

If part of the system doesn't rotate in perfect alignment with the net axis of the total angular momentum, then the other parts of the system must rotate on an axis that also is not quite coincident with the axis of the total angular momentum. Mr. Coriolis has explained how this will probably cause a "nutation" of the axis of rotation of whichever part you happen to be attached to. This nutation of the geometrical axis of rotation of the solid crust of the earth has been fairly accurately measured and future squirrely behavior can be predicted with "approximate accuracy."

The parts not going with the flow can be disturbed and can change direction, which causes the declination to change - and the magnetic poles to move. Causes for such changes are only vaguely known, but you can visualize them as "somebody dropped a turd in the tub."

A reasonable understanding of how all the various wiggles play together requires combinations of Newtonian Mechanics, fluid dynamics, and magnetohydrodynamics, along with a fair bit of conventional electrodynamics with only a small touch of quantum electrodynamics and plasma physics. The expert on the subject is (was?) perhaps the Indian Astrodynamacist Subrahmanyan Chandrasekhar, who made it all seem quite simple. Most of his work, however, was about 50 years ago and I don't know if he's still accepting questions from the audience.

John