Originally Posted by svHyLyte
Consider a rotating wheel. The angular speed is constant. Measuring the linear speed of the flow tangent to the circumference at any radius from the "axel", the speed must increase as the radius increases (preservation of angular momentum).
True, but a river is not very much like a wheel. The latter is rigid while the former flows. With rigid rotation you will get tangential speed proportional to radius, but with a liquid, not necessarily.
A naive application of the Bernoulli principle puts the fast water inside, since the inside of turns is associated with low pressure, which is associated with high speed.
Erosion could indicate fast water on the outside, or it could indicate that water is better at eroding soil that it crashes into than soil that it rushes past, ie. On the high pressure side of the curve.