its nice to see that my point has been well debated and I don't want to contribute any more noise. thanks for all the vigorous feedback.
One thought I did want to make clear is my assumption that field entry was handled at a fairly low level by a subroutine, probably as part of the framework used to build the interface, and rather than recoding for each control one would only need to change the default behavior (if not hard-wired)
Another anecdote...
I build complex engineering interfaces for R&D software that is heavily GUI oriented and controls HW as well as other real-time tasks, for use by others. Despite my best efforts to train the users ("its basically a beta, please let me know if you don't like anything or have problems"), I find out later when confronted by a problem they will often just play with the controls until they get the desired response and never tell me about their workaround! (I usually find out because their attempts at workarounds will crash something and THEN they go looking for me.) They are more interested in moving forward, just like the veterans of TG, and I can understand that. It's very frustrating for me though (at work) and I still don't have a good answer.