When John Walker chose LISP as the core extensible language for AutoCAD, he did so on the basis of its inherent dynamic polymorphic nature. Recursion and chameleon-like characteristics made it as fluid and flexible as a the T2 walking through the mental hospital metal bar gate (without the pistol, of course).
What Autodesk is ignoring is potential. There is and always has been potential within the Visual LISP world to grow the language as a standalone platform. It could be used for so much more than CAD purposes. Even DCL could join in on the ride beyond the walls of Fort AutoCAD.
Once unfamiliar programmers got used to working with lists and functions like mapcar, apply and lambda, who knows where it could lead?