At first I wondered if there was going to be some interesting use of the blocked space, where parts overlapped, or would occupy the same space. After the authors point it out, it was obvious this could be used to limit the configuration search space. When the fact that much of the work done is parametric design it is a wonder I didn't think of blocked space as a way to eliminate possible parameter values before the kindly authors state the fact.

      One of the limitations of the system is the reliance on mechanisms being planar assemblies. That is, it does not handle curved surfaces. They mention the utility of this when identifying contact points. One possible solution would be to model curved surfaces as many, very small flat surfaces. Of course computing technology would have to be adequate at doing this in a reasonable time. Considering the improvements made over the last four years in real time graphic rendering it seems possible to consider even today.

      For the first eleven pages I was unsure what was meant by dynamical simulation. I was surprised that is wasn't explained earlier since it is mentioned often. I was very impressed by the graphical output of the system. Maybe I had become used to the quality of the graphics prevalent before the 1990's. This is certainly easier to look at than plain text traces of code execution.