"The purpose of a poem is to be wonderful."
That's the statement that Andrew Gelman attributes to me (by implication, and as a summary of a post I wrote). Let's first assume that I agree that that sentence represents my views. I think it does, in fact. Let's break it down a little more.
Andrew thinks that this excludes other uses of poetry, so to speak, other purposes. I would think that in order to fulfill these other purposes, it first has to be a poem. In other words, it has to be that before it can do anything else. But does it have to be wonderful?
Obviously not. For example, there are lousy poems that fulfill other functions, and quite well. In fact, a propaganda poem is probably going to fulfill its purpose not so well if it is a wonderful poem.
So this leads us to a strange usage, in which we only want to call poems those that are "wonderful" as poems. Suppose I have a pencil case, but not a good one. You might say, "You call that a pencil case, that's not a pencil case, it's a piece of crap!." You would be wrong, because even bad examples of the category still belong to the category.
With poetry, though, the idea is that not every random utterance will be a poem, but that a poem will be language heightened in some way. Something has to be going on in the language for it to be called a poem. A poet might ironically write a poem that has nothing going on, to make a point. We call this "conceptual poetry," in fact. That is, to make a point about how it is no longer possible to write poems with a lot going on, rhythm, metaphors, etc... This conceptual poetry (or the branch of it that does this or similar things) does not disprove my contention that poetry is language where something interesting is going on. In fact, it proves my point, since it is making a meta-literary point based parasitically on the more traditional definition.
A math professor and I were talking about what he did last night, at a faculty reception. He said he counted stuff. I'm sure it's more technical than that, but basically he told me that his branch of mathematics involved counting things and comparing results, looking for patterns.
So, to be super non-technical here, the study of poetry is the study of "what's going on" in the language of poetry. How it communicates emotion, meaning, how it is structured, etc... The purpose of the poem is to be a poem, wonderful or not, and then also do whatever else any poet or reader wants it to do, like change the world, or win the heart of a lover. Since language is used for many other purposes, then it follows logically that we want poetry to do something extra special, not covered by those other uses.
There's an implicit Poundian ethos in this viewpoint. What I mean is that Pound said that poetry was language charged with meaning. Language + some heightened or improved elements. We are all Poundians, in that everyone more or less agrees that interesting rhythmic structures, or concrete images, etc.. provide the shortest route to making a poem wonderful. It seems like Andrew, whose blog I greatly admire by the way, shares this ethos, because he is able to find these poems mediocre in a similar way as I do, pointing to clichéd language, for example. I don't particularly care if anyone agrees with my particular judgments about poems or poets, since those judgments will always vary anyway.
A poem that's not wonderful is still a poem trying to be more wonderful than it in fact is. When we read a bad or mediocre poem we want it to be better than it is, more fitting with what the purpose of the poem might be. It's no different from what the purpose of an Eric Clapton guitar solo is. It's purpose is to be a great solo, right? Or do we want to say that it's purpose is to fill up 16 measures?