The first is that bias orientations are dependent on the location of adjacent CPs. This means that the gamma angles of those corner CPs require different  values and so, are inconsistent from one CP to the another.

Beause of this, two opposing CPs of a corner will generally require different Gamma values.

The other disadvantage is that because the bias orientation is dependent on the location of adjacent CPs, any change in position of any adjacent CP will modify the bias orientation and thus, the gamma value will need to be readjusted.

This makes it very difficult to build a simple basic primitive and stretch it to obtain different parts of a model.

Since scaling a model in order to obtain smaller bevels will also modify the reference line orientaiton, this also will produce models which will need readjustments of the gamma values.

When building complex beveled models, this readjustment of the gammas can be time consuming.

To better understand the mechanics of Gammas, please read the Animation Master Gammas.