Moore's Paradox doesn't support pragmatic encroachment

Why did a certain percentage of people vote they way they did?  I don't know. 

Note that the import of "I don't know" there isn't merely that I lack belief in a proposition having the lofty epistemic status knowledge. 

Rather, "I don't know" generally implicates that one has really no idea. 

This sets up quite the asymmetry with "I know."   I want to bracket one common usage of "I know" which is synonymous with "I am aware that."  Whether this is a loose "sense" of "know" or a loose *use* of "know" is not something I'll take sides about right now. 

In the usage philosophers are usually interested in "I know" implicates more than merely positive epistemic status.  If one is confident enough to assert not just "p" but "I know that p" then they represent themselves as having considerable evidence that p (though not necessarily that they know that they know, nothing forces that on us).  [Side note: It is often taken for granted that the pragmatic implications of assertions are always the same.  However, attention to details goes against this.  How much evidence a speaker represents themselves as possessing depends on *how confidently* the assertion is made.  This fact is conveyed by extra-linguistic features of the conversational context as well as quasi-linguistic features like tone of voice, facial features, and whether the assertion is an answer to a question or volunteered freely or only after some eliciting.]

 

So it seems interesting to me that "I know" and "I don't know" don't just implicate being on one side of a line of epistemic demarcation or the other, but tend to push the implications to the extremes. 

 

But back to Moore's paradox...

 

"p" generally implicates that one has good evidence that p

"I don't know" generally implicates that one has "no idea" whether the target proposition is true.

This is not, of course, a semantic entailment of "I don't know" but merely a conventional implication. 

But there is an entailment from this implicature to the negation of the implicature of the first conjunct.

You can't both have good evidence and no idea. 

This is sufficient to explain the infelicity of "p but I don't know that p."

 

It is also preferable to a standard interpretation which relies on the assumption that an assertion always represents the speaker as knowing what is asserted, for this assumption is at odds with the felicity of the following speech which, though not "utterly common" (a common, though perhaps not utterly so, phrase among those who buy the knowledge account of assertion) is relevantly similar to common speech and clearly appropriate enough.

 

"p"

"Ah, but do you know that p?"

"Perhaps not, no, but I have sufficient evidence to reasonably believe that p."

 

This should *never* be admissible if the knowledge account of assertion is true.  But it is sometimes admissible.  QED.