My interpretation of mdhughes' point is that it is a failure to say, hey, try my program. And then it turns out you have to define what the program will do yourself in /etc/my.conf, or worse "there's an api! (which you could write a different, useful program with)".
In rms' paper presenting emacs, his emacs had the explicit goal of providing an off-ramp out from secretarial work into authoring original emacs lisp programs. So his intent was to fail secretaries per se.
