Comments on an approximation coefficient proposed in 'stochastic model simplification

It is pointed out that the approximation coefficient defined in the above paper does not always satisfy the claimed properties. An alternative definition is suggested.