nj, To my understanding the main difference in how John Grinder's method of modeling works compared to David Gordon's method, is that John strongly suggests taking everything in implicitly by mimicking with micro muscle, at second pos, while in a know-nothing state.  In the end, both produce coded (chunked down) information from which they create patterns to install the behavior, etc. (think about all the games and exercises you played with learning how to match, track patterns, et al, in practitioner training). WITW advocates absorbing the patterns unfiltered, while D Gordon advocates using the filters of NLP to catalog strategies, equivalences, cause-effect beliefs, etc..  I wonder however if John would have absorbed all the patterns he did from M Erickson, if he hadn't had highly refined language skills filtering unconsciously from the outset.  It would be an impossible question to answer, I presume.  One thing David G says that makes sense is that in going deeply into 2nd position you might pick up things you don't want to pick up.  I have had this experience while modeling WITW style -- or real NLP modeling.  And both Richard Bandler and Stephen Gilligan are diabetic, supposedly, as was Milton E, both having gone deeply into 2nd pos with him.  Coincidence? That said, there are probably ways to protect yourself, as John Grinder says, and that must have something to do with setting things up with your unconscious ahead of time. Another great approach to cognitive modeling is taught by Jonathan Altfeld, at www.altfeld.com, called Knowledge Engineering.  It's based on his work in expert systems (a form of artificial intelligence which involves tracking the patterns of experts and coding them into a special AI shell).  It's definitely worth taking a look at, even if to just help with the coding process after the implicit stage of WITW modeling. best, JS