Previous estimates on the time it takes for a flip to happen are in the 1000-30000 year range[1]. That paper gives a 7000 yr average but the variance on the tim

Previous estimates on the time it takes for a flip to happen are in the 1000-30000 year range[1]. That paper gives a 7000 yr average but the variance on the time it takes is likely large. The Berkeley article you linked to is on this paper[2] which makes a claim of that short flipping time. People in the field think it is quite an unlikely short period of time for a flip and we would need more evidence to firmly believe they can happen this fast. Sediments are notoriously complicated on how they record the magnetic field and the rates at which they are deposited at are difficult to determine. Source: I did my PhD work on the Steens Mountain reversal. It is the Earth's reversal with the most transitional directions recorded in much more reliable volcanic rocks. This was possible because of the huge volume of the Columbia River Basalts that were erupting at that time around 16.7 million years ago. Unfortunately it happened so long ago that direct radioactive dating of its length is impossible with current methods. [1]https://doi.org/10.1038/nature02459 https://doi.org/10.1038/nature02459 [2]https://doi.org/10.1093/gji/ggu287 https://doi.org/10.1093/gji/ggu287