The term pole shift, while used often by lay people and geologist alike, is not a precise term. Magnetic Reversal is better. The earth's magnetic field is created (as best we can deduce)by a dynamo resulting from the complex flow of the Earth's outer core, which is a high pressure iron "fluid". As the various vortices change over time, the magnetic field changes. And while Lindy is correct in asserting that there has never been a reversal in recorded human history, we have a pretty good idea of the changes in this magnetic field preserved in rocks around the world for much of its long history. We can collect oriented samples of igneous and some sedimentary rocks, put them into a machine called a spinning magnetometer*, and read not only the direction of the north pole at the time of the formation of the rock sample, but the magnetic inclination, magnetic strength, and the signal carried by several different magnetic minerals that capture the signal at different temperatures. The inclination is important because we can deduce the latitude that the rock formed.

This then leads to the reconstruction of what are called polar wander paths. When combined with other geologic evidence (Mantle plume locations, fossils, structural reconstructions, etc.) we can deduce the positions of the various tectonic plates over time.

That's not to say that these investigations are not complex, nor scientifically debated, but Paleomagnetism is an important part of geologists' tool box.

Our earlier discussion on pole shifts may be found here, which includes links to pages of greater detail.

* A device I have spent countless hours operating, as my Master's Thesis used Paleomagnetism to determine the rotation (or lack thereof) of faulted Precambrian granites.