1: \begin{abstract}

2: 

3: We discuss the role of random basis function approximators in

4: modeling and control.  We analyze the published work on random basis

5: function approximators and demonstrate that their favorable error

6: rate of convergence $O(1/n)$ is guaranteed only with very

7: substantial computational resources.  We also discuss implications

8: of our analysis for applications of neural networks in modeling and

9: control.

10: 

11: \end{abstract}

12: