At an abtract ("argumentational") level, you are "of course" correct in all that you've been saying. Making students say things in a prescribed way, "ferociously" bringing them back into line or supplying the "obvious" answers, and ignoring or suppressing the effects of the exact lexicogrammar on the "unfolding" discourse
obviously "will" ensure apparent success in "learning".
And, at higher levels, the methods change/converge more somewhat (at least, I seem to recall you saying they did - "shoot fast, from the hip" or something), if they weren't similar in some ways to begin with at the lower levels too.
However, I bellieve that until the teacher contemplates looking at the variety, freedom and ultimately potential "chaos" (that is, potential to always...instead) of real discourse to "inform" the input, and allows
positively "expects" the students to create some themselves, what could be valuable (I won't go so far as to insist it is
real, as opposed to the "learning" above, only that it seems to me to be more real

) learning
experiences (processes) are not taking place.
Students have to accept that they don't know anything until they've
trulylearnt it (struggled for a bit with incomplete, missing or "open" scipts), and being "told" they now know something simply because they've repeated and heard themselves saying it might not always be the obvious shortcut that it seems, especially if the phrases they are being supplied with have been chosen with scant regard for the functional effects the lexicogrammar within them might have (and I have to still wonder if Direct teachers can supply decent models if they also never have to choose/construct them or give them much thought).
I believe it does students a real disservice if it begins and (likely) continues to be pretended if not forcibly insisted otherwise, but then, I doubt if confirmed Direct/Callan/Bertlitz etc teachers would have the skills to do justice to this complexity (not that you are such a teacher despite you protestations to the contrary, woody, or that your insistence that "doing it justice" is beyond "teachers").
Others get around to looking at data and thinking things through in more depth, and somehow manage to come up with a method/process (and obviously have a "product" too, as a result of their research, in mind at least and waiting to input at some stage, perhaps before the practice or task or activity, perhaps after, often at both stages if it seems necessary!).
I suppose you could argue that 'coming up with the method in the process' means the unfortunate and unhelpful end of a "Golden methods era", but as Kev playing Jim Garrison in JFK said, 'Let justice be done to the language, though the heavens fall'.*
The only way to ultimately exercise much control over the long-term fate and well-being of our students as a
result of their linguistic competence** is to attend to meaning in supplying the form. Note that I said the form is supplied along with or for the (
anticipated) meaning, not that people mill around saying very little and somehow "mind-reading" in the absence of saying anything
meaningful (to which we might compare the Direct method of saying "something", not always necessarily "meaningful", just so long as it's what the teacher expects - the hard-pressed teacher without hours rather than an hour of freetime in which they could do a bit of thinking and reasearch

).
This post is way too long and a bit of a mess, but I didn't have time to get it done in one fell swoop and grew tired of reducing it down and rewriting it each time I could get back to it. (I was tempted to delete it all and leave it to another day...but I knew that not finishing what Id started now wouldn't spare me the pain later.

). I think regarding when exactly to input language, I said something on the "Dogme" thread that halfway makes sense and might clarify some of my abbove babbling:
http://www.eslcafe.com/forums/teacher/v ... 1839#11839
*Actually he didn't say 'the language' there, but if he was a linguist talking about investigating the assassination of a John Sinclair by a Chomskyan sympathizer, he might well have).

Bad joke, sorry, but I like the original quote, anyway.
**This competence obviously
will differ depending on the linguistic content of differing syllabuses, and how that content/form relates (is related, passive, "by the writer/developer", which is what a techare should become or certainly develop an eye for) to the situations/contexts and functions that get played out "in the course" of the syllabus, that is, in the exact materials and methods employed.