Bill Poser at Language Log has defended the expression 'more perfect'. His reasoning is that we can speak of things being absolutely perfect, and therefore we already admit of degrees of perfection. So those who say that once something is perfect it can't be more or less so are ignoring the semantics of the word in its regular use. It also impugns the United States Constitution in its use of 'more perfect' to describe our hoped approach toward perfection as a country ("in order to form a more perfect Union"). Christians have a similar notion, expressed in Paul's descriptions of believers growing more and more like Christ, though I don't know if the Greek ever has an expression parallel to this one. The concept is clearly there, though, and that's all you need to show that there's no grammatical insanity or contradiction in such expressions.
This got me thinking about other constructions like this. Grammar police (as distinguished from legitimate grammarians who study grammar as a discipline wihtin linguistics) often fume at 'more pregnant', since one is either pregnant or not pregnant. How can a binary property with only two values admit of degrees? Once you think about it, it shouldn't be hard to consider how almost any supposedly binary term can admit of vagueness. Philosophers like 'flat', since it's got an absolute reading according to which nothing is flat but ideal geometric planes, but all sorts of things are more or less flat without being absolutely flat.
'More pregnant' should be similar. There is the binary use of 'pregnant' without modifiers, which signals just whether someone is or isn't pregnant. Yet there's also the degree use, which signals how far along someone's pregnancy is. There are both degree concepts and binary concepts, and our language reflects that. Grammar police who insist that there's nothing to communicate by such an expression somehow can't see that obvious fact.
The biggest pet peeve of grammar police is 'more unique'. If something is unique, there's only one. How can something be more only one than something else? That's binary if anything is. Well, hold on a second. Uniqueness in an absolute sense is exactly that, but can't we easily think of degrees here just as easily? Everything is unique in one sense, because nothing else is it. If it's something else, it's not it. If something is it, then it's not something else. But is that what we mean when we say something is unique? We wouldn't be saying anything if we meant that. What we really mean when we say something is unique is that it's in a class by itself. There isn't anything else quite like it. It has properties besides its identity that nothing else shares. Well, that can lead to a degree use fairly easily. If something is almost in a class by itself because few things are like it, then it's fairly close to being unique, and by the degree use of seemingly absolute terms we can understand the grammar of 'more unique' to be expressing exactly that. Something is more unique if fewer of its properties are had by other things.
The grammar police complain that expressions like 'more unique', 'more pregnant', and 'more perfect' are meaningless or contradictory, but that involves an insensitivity to the nuances of the semantics of the English language. The philosophical consequences of this are surpisingly important and interesting. There's a whole research field into context-sensitive expressions, which has included such wide-ranging consequences as defending against brain-in-the-vat skepticism by claiming 'know' is like this to my current project of applying something like this to racial terms to show that some of the widely divergent views on what race is may be simultaneously true but only appropriately expressible in different contexts or in a meta-context. I should register my complaint against the work of one philosopher, Peter Unger, who spends lots of time arguing that hardly anything is flat for the sake of showing that we know hardly anything. He basically denies this whole phenomenon of language to make that point. Thus his whole argument for skepticism relies on pretty much everything I've said in this post being false. What would be interesting to see is if you can deny my basic point here and still resist his argument for skepticism.