I read Ruskin’s “On Modern Painters”, especially the section “of Truth of water” ( Vol 1, section V) where he castigates landscape painters for throwing a blob of color on a canvass and exploiting a tacit agreement with the viewer that that blob means: ”Water; with or without reflections”…Ruskin said it looked nothing like light on water, and went out with a notebook and pen and carefully described what he saw.  It’s a powerful antidote to modern theorized claims about the Prison House of Language or the slippage of the signifier. Starting with the admission that no one will ever manage to paint the sea, he then praises those who get closest. Nice metaphor there.
As Ruskin says about painting the sea (Vol I part two, section v, ch1 point 5), it’s impossible, it can only be suggested by means of ‘more or less spiritual and intelligent conventionalism’(Vol 1 page 355).
A poem is a conventional literary category, and so there is a general sense of what a poem is and should do, and while that obviously varies across different versions of ‘poetry’, there are dominant assumptions about what a poem should be: in terms of diction, subject matter, and treatment despite the various movements over the past two hundred years. In fact, given the lyricization of poetry over the past two centuries, we have fallen such a long way since Chaucer could write about everything and anything. Put against the works of the great medieval poets, most modern poetry seems flat and anemic.
These often unspoken definitions, internalized over the years, exert a significant pull on the writing of any piece, they are part of the way in which I recognize that I am writing a poem, so the paradox arises that to write a poem one must adhere to the conventions, but I want to avoid the conventions, but if I avoid too many of them is it still a poem.
And that had merged with a distinction Donald Davie makes between two types of poems, one of which he calls “literary” but which I think of as the ‘Poetic Poem’: it’s the one where the poet is being self consciously ‘poetic’ ticking the right boxes, and having poetic thoughts which no one, let alone the poet, has ever had during the experience it purports to describe. Done well it can be interesting. Done badly , which it so often is, the poem is the verbal equivalent of Ruskin’s blob of color…we accept that someone is having a poetic moment and don’t enquire too closely about the truth of it. We respond as well trained seals, especially if the subject matter is about some kind of individual trauma. We parade our sensitivity by responding to poems which invite us to sympathize with the poet and which don’t challenge us ot consider our assumptions or values.
Whatever happened to the "Affective Fallacy"? Or, as Sir Geoffrey Hill put it in a discussion of Owen, "The Poetry had better not be in the Pity'.
That’s another issue. Robert Graves made a simple but useful distinction in poetic Unreason. If we ask if a poem is successful then there are two answers: successful for the poet, and successful for the reader. Far too much poetry may well serve as therapy for the poet, but offers the reader nothing except the opportunity to wear his or her politics or ideology on their sleeve or to parade their own sensitivity. Come to think of it, I.A Richards called these “stock responses” though I don't think he was talking about ideology. There is, somewhere, a fine poem by Thom Gunn about this all.
It may make sense one day.