After some dispassionate reflection, I believe my proposal to rewrite the entire system in Python was an ill-considered, unscholarly, and unsound overreaction to a poorly functioning product.  In plain English -- it was a bad idea.

I've spent a few days experimenting with Python/QT and Java/Swing as possible alternatives to the current Java/GWT implementation.  Neither of the alternative platforms is a panacea for the problems with the current system.

I believe that the approach most likely to produce a usable product is to focus clearly on fixing the problems in the existing code, fully test the entire system, and release as soon as possible for client review.  We will discuss this approach on Monday.