On the basis of this, Shackel builds an analogy to philosophical positions. The 'motte' is a position that is easy to defend, although it's not really the important thing; the 'bailey' is the position you really want, but it may be difficult to defend against close argument. So when close argument comes, you fall back to the obvious things in the 'motte'. For instance, your outer position might be very doubtful and your fallback position might be something almost no one would doubt, so that and when your outer position is attacked, you simply argue for the fallback position. Shackel takes this, plausibly, to happen usually by equivocation.
There are a number of obscurities in Shackel's original discussion. None of them are fatal or problematic for an exploratory discussion as it certainly is, but they do sometimes complicate his stronger claims. For instance, he says:
Diagnosis of a philosophical doctrine as being a Motte and Bailey Doctrine is invariably fatal. Once made it is relatively obvious to those familiar with the doctrine that the doctrine’s survival required a systematic vacillation between exploiting the desired territory and retreating to the Motte when pressed.
But in fact this does not seem to be strictly true, because defensibility is not actually a feature that can read off a position -- it depends on available resources for defense, which can vary from person to person. Thus person A might have an understanding of how to defend the outer perimeter against you while person B does not. Thus the perimeter might be quite defensible with person A and indefensible with person B, with B incapable of doing much more than falling back to obvious basics. This is obscured by the fact that Shackel often talks about the matter in terms of an outer interesting falsehood and an inner trivial truth; but, of course, in real life positions don't come marked with tags saying 'This is false', and whether it is in fact false doesn't really affect much how defensible it is in the immediate context. The vacillation is something that occurs in an argumentative context; but Shackel regularly treats the diagnosis as being of the position as it is believed.
This is a potentially serious ambiguity given that in the abstract a highly defensible position surrounded by a region of things that are just suggested or guesstimated or practically useful is the normal state of philosophical positions as believed. There is nothing wrong with this -- it is how one explores and develops the intellectual territory in the first place, and even if it weren't people can just be honestly not aware of the exact border between the trivial and the nontrivial-but-seemingly-obvious-because-it-looks-like-the-trivial. Likewise, if someone attacks your position and you give it up and fall back to a safer position, where you stay, this is not a motte and bailey situation, because you aren't returning to the original position. Likewise, if you do return to the original position but build up the means to defend it in its own right, this is not a motte and bailey situation, either, because you then stay at the outer position. To be a diagnosis of a flaw based on differential defensibility, it needs the 'systematic vacillation' in actual defense of the position that uses this common structure of positions in a sophistical way. This is not something that happens with positions as such, but only as they are handled in actual interaction with others.