This standard of writing might be historically contingent (it is) but it is still in force, in that departures from it need to be justified. This is how James Schuyler wrote, or Lorine Niedecker or Denise Levertov at her best.
This is also how we judge poetry of the past, in a sense. Though we might tolerate more rhyme and meter, elevated language and archaism, we still want that concentration of meaning and a strong poetic eye.
II. That is about 90% of it there. Most poems fail because simply because they don't follow those directions and depart from them in quite unintentional ways. Often, a beginning poet will seem to have read no poetry, and has heavy-handed prosaic effects, but with no knowledge at all that he's not supposed to do that.
III. The rest is fine tuning. The main area of fine tuning, once the poem is filled with concentrated visual imagery, is about getting the persona who is speaking the poem exactly right. This means adjustments in register (up or down) and a really fine-tuned hearing of the language.
Everything to do with logopoeia is necessary to make the poem its own unique utterance, not merely a conventionally good poem with lots of sensorial images. You cannot make a poem sound too poetic with words like shimmering. Instead, think of using words that come from a different context: "a sodium pentathol landscape / a bud about to break open" (James Tate; emphasis added).
IV. The poem should seem both inevitable and unpredictable. So if it is predictable, you see what's coming a mile away, or "telegraph" your intentions. Thinking of hitting someone (forgive the violent imagery but no better metaphor comes to mind). A boxer who telegraphs his punches signals in advance what the punch will be, and thus makes the defensive move, and even the other fighter's next offensive move, quite easy. We know the kind of poem that sets up its humorous premise early and then gives the expected answers. Notice too how the use of statistically frequent combinations of words unnecessarily telegraphs your intentions.
On the other hand, the poem should move with some degree of inevitability as well. If it is merely unpredictable it won't make much of an impression. Think of reading a poem line by line and not looking at the next line (keeping it concealed under another piece of paper). Each line surprises, but in a way organic with the rest of the poem. The next line can disappoint by being too predictable or too far afield.
If you know what the poem is going to say beforehand, you will end up being very predictable. You need to discover the meaning of the poem while you are writing it.
V. We want to avoid poetic devices that every other poet uses, like a simile every other line, or a first person speaking in the present tense. There is a contextual sense in which a poet has to be savvy about what the conventions are, and not see them simply as the only possible option. Poetry will seem amateurish if it simply falls easily into certain stale patterns.
VI. If you look at poets like Robert Duncan, you will find he doesn't care about certain things. For example, he will be turgid and abstract, archaic or pretentious in diction, etc... He cares, but he doesn't see anything wrong with that. Or many contemporary poets combine use the "dim lands of peace" construction that Pound condemned. The attempt to write the conventionally good poem, then, could just be a form of timidity. When I depart from these rules, which I do in every poem, I see them as a foray into bad poetry. So any kind of bathos, deliberate use of "dim lands of peace" constructions, overt sentimentality or triviality, is what I tend to favor. That tends to work better for me than earnest attempts to write the good poem. In fact, I modeled myself after Kenneth Koch, who I didn't realize was writing much more in earnest, many times when I thought he was being purely parodic. Or maybe I am wrong about that.