How to Write a Poem That Isn't Boring: A Working Method

Almost every unsuccessful poem I have read fails in one of four identifiable ways. It announces its emotion instead of producing it. It reaches for words nobody uses. It stops at the first true thing it finds. Or it ends on a summary. The good news is that all four are process problems, not talent problems, and a process can be changed.

Step one: start from an object, not a feeling

If you begin with "I want to write about grief", you will write about grief in the vocabulary grief already owns: loss, empty, ache, shadow. Those words are exhausted. Begin instead with a physical thing attached to the feeling — the mug that is still in the cupboard, the wrong shoes by the door, a phone number you have not deleted. The object does the emotional work while your language stays fresh.

Bashō built an entire tradition on this principle. His pond poem has no adjective for feeling in it at all: pond, frog, splash, silence. The reader supplies everything else, and because they supplied it, they believe it.

Step two: write far too much, badly, quickly

Give yourself fifteen minutes and produce forty lines you have no intention of keeping. The purpose is not the lines; it is getting past the first three ideas, which are always the ideas anyone would have had. Interesting material tends to appear around line twenty-five, usually in a phrase you wrote carelessly.

This is also where a generator can genuinely help. Not to write your poem — a machine has no reason to mean anything — but to break a blank page, to throw twenty unexpected pairings at you so that one of them collides with something you actually know. Use it as a flint, not as an author.

Step three: find the turn

Every poem worth re-reading changes its mind somewhere. The sonnet formalised this as the volta, but it is not a formal device; it is a psychological one. Rossetti asks to be remembered and then releases the mourner from remembering. Neruda says she loved him and then reverses who loved whom. Find the point where your poem could become honest at its own expense, and put the hinge there.

If your poem holds the same opinion in the last line as in the first, it is a statement with line breaks, not a poem.

Step four: cut the explanation

Beginner drafts almost always contain one line too many — the line that tells the reader what the previous lines meant. Delete it. Then read the poem again. In my experience nine out of ten poems improve, and the tenth needed a different last line, not the explanation back.

The revision checklist

• Read it aloud. Every place your voice stumbles is a place the line is wrong. This is the single most reliable diagnostic there is.
• Circle every abstract noun — love, time, soul, pain, memory. Keep at most two in a short poem. Replace the rest with things you could photograph.
• Check your line endings. The last word of a line gets extra emphasis whether you intend it or not; make sure it deserves the emphasis. Ending on "the" or "and" should be a decision, not an accident.
• Count syllables in your key lines. Not to enforce a metre, but to notice rhythm you created accidentally. Our Syllable Counter is faster than doing it on your fingers.
• Check the first line and the last line together, with the middle covered. If they make a satisfying two-line poem on their own, your structure is sound.
• Remove the title's job from the body. If the title says "Funeral", the poem does not need to say funeral.
• Leave it for three days. Nothing else reveals a false line so efficiently.

On rhyme, honestly

Rhyme is not old-fashioned; forced rhyme is. The problem with beginner rhyme is that the rhyme word chooses the content — you needed something for "heart", so now the poem is about being apart. Two fixes. First, use slant rhyme, as Dickinson does constantly: soul/all, storm/warm. It gives you a much larger vocabulary and a subtler music. Second, choose the rhyme word first and build the line backwards toward it, so the rhyme never forces a sentence you did not mean.

How long should it be?

Shorter than you think. Bashō needed seventeen syllables. Dickinson needed twelve lines. The instinct to keep going usually means you have not yet found the poem's real ending, which is often four lines above where you stopped. Try deleting your final stanza entirely and see whether the poem survives. Surprisingly often, it improves.

None of this is a formula for a good poem, because there isn't one. It is a way of removing the four failures that keep an otherwise interesting mind from getting onto the page. Do the object first, overwrite, find the turn, cut the explanation. The rest is what you happen to have noticed about being alive, which is the part nobody can teach you and no model can generate.