Some thoughts, page-54

Bothers me too. It will be embarrassing if the virtual prototype (or modelling using the methods of artificial intelligence as they're now talking) shows up that the cart was always put before the horse by the revelation of reasons why the physically implemented prototypes were infeasible from the start and could never have worked. The converse would be that the computer modelling demonstrates the feasibility of what they were hoping to achieve with the physical prototyping and why it failed. The pathway to successful implementation of a physical prototype should then be illuminated.