While I agree with you mostly, see what you are saying, and think I understand why, I would submit that programming based on even the best funcition analysis remains only hypothesis testing. You look at a behavior or a class of behaviors (a hypothosised class that is). You are thourough looking at everything you can think of. You use a form even from those who have done this before. Then you guess. You should have a very good guess but, you test your theory. And even when it appears that you have been successful, you remain somewhat uncertain. Single case design (Pennypacker and Johnston, Hernsen and Barlow, etc.) can elicidate function more clearly, much more so, but with the vast confluence of variable effecting human behavior, anyone proclaiming 100% certainty about a behavior is a fool. Me, I am 99% sure that you know all of this and kindly are providing information on an area you know. This kind of sharing, ability to reocnsider, to teach and learn both are what we need a lot more of. Dangling participles and all (mine not yours) thanks much!!