Logic

The "implications" or "discussion" section of your paper should not ad-duce new evidence but merely de-duce the consequences of your research, whether for theory or for practice. In so far as it describes facts, these should be wholly uncontroversial, either because they are familiar to the reader or because you present them only as "plausible". Your argument should not depend on the truth of these claims. You are engaging only with your reader's logical faculty here, your reader's logical sense.

The rest of the paper draws support from outside of itself (even if you insist it can never point "outside the text"). It says something about what people do in the world, or what they say, or what they believe, or the texts they write. But the facts that are stated in the implications or, often, the acts it prescribes, are not grounded outside the paper, they follow logically from the facts already presented in it. That is, to grant the truth of the bulk of the paper we have to believe that you know something about the world beyond its pages. But to grant you your implications we only have to follow your reasoning from what you've already convinced us is true.

There are two major kinds of implications, each of which follow from the disappointment of the results section (the disappointment implicit in the results section, if you will): the reader (and you) may be disappointed in the theory or in the practice. Either the theory will have have to change or the practice will have to change. If we grant you your conclusions, we will have to do something.