If anyone's interested, here's the napkin math: [assuming a round earth, SI units used because it makes conversions easier] Earth weights 6E24 kg with a radius of 6.4E6 m, with an angular frequency of 2π/(24*60*60) = 7.3E-5 rad/s giving it an approximate angular momentum of (2/5)MR²ω = 7.2E33 kg·m²/s The actual value would be somewhat smaller than this, since this assumes a uniform density sphere, while the earth has most of its mass closer to its core, but a slightly wider-than-spherical shape. To swing the axis of rotation 180° would require double this input of angular momentum. The change described in the paper is ~0.08 meters per year, while the part of that change attributed to groundwater movement is only ~0.04 meters per year. In angular terms, that's 6.2E-9 rad/year (2E-16 rad/s) attributed to groundwater. How much angular momentum would that shift require? At that rate, the earths rotation would take π/6.2E-9 = 500 million years to completely flip earth's rotational axis, so that's an annual change of 2·7.2E33/5E8 = 2.9E25 kg·m²/s/year (9E17 kg·m²/s²) Assuming all the water pumped by humans is pumped from the equator in the optimal direction to shift the axis of rotation, what flow-rate of water would this correspond to? 9E17/6.4E6² = 22000 kg/s, or about 22 tons of water per second (about 6x the flow rate over Niagara falls). Considering total human freshwater use is close to 120000 tons per second, 22 tons per second doesn't seem unreasonable as the net effect of all human water pumping. Main take-away is that *if all this math is correct*, and **nothing changes for the next hundred million years**, humans will have fucked up the rotation of the Earth.