Poem Type

Poem Type suggests a poetic form by asking what you want from a line, rather than which poets you already read.

I came across the Wikipedia category of poetic forms and wondered which one would actually be my favourite. There are around a hundred and fifty entries, listed alphabetically, with no way in short of reading your way through all of them. The category tells you every form exists and nothing about which one is yours.

This one asks twenty four short statements, answered on a five point agree scale, and builds a profile across ten dimensions. That profile is matched against twenty four forms. You get a form, three poems in it ordered to suit you, the rules in case you ever want to write one, the family it belongs to, and a shelf of four more that came close. About three minutes, no sign up, nothing uploaded.

What it measures

Ten dimensions, and deliberately not personality factors. Reader Type measures openness and empathy because it is guessing what sort of person likes what sort of book. This one measures the features that actually vary between forms, and reads your answers as preferences over those same features:

• Music. How much the sound of a line does work of its own, apart from what it says.
• Rhyme. Whether the chime is a pleasure or a distraction.
• Repetition. Appetite for a line that comes back, changed by what happened in between.
• Brevity. Short and closed, or long enough to spread out in.
• Constraint. Whether a hard rule feels like a game or a cage.
• Narrative. Wanting something to happen to somebody, rather than a held moment.
• Image. Pull toward one exact object over an argument or an idea.
• Interiority. A private voice overheard, against something meant to be spoken aloud to a crowd.
• Wit. How much you want cleverness and comedy in the room.
• Feeling. Open emotional weather, against cool restraint.

Why this model is on firmer ground than Reader Type's

Reader Type is honest that every genre profile in it is a hand-authored guess about what sort of person likes what sort of book. Most of a form profile here is not a guess at all. A villanelle scores top marks on repetition because nineteen lines hold two refrains four times over. A haiku scores top marks on brevity because it is seventeen sound units. A limerick scores top marks on rhyme because it is AABBA.

Those are descriptions, and they are checkable against any prosody handbook. What stays a judgement call is register: how funny a form is, how private it sounds, how much feeling it carries. Those are the softer numbers, and they are the ones the feedback capture exists to correct.

Choosing twenty four forms out of a hundred and fifty

Most of that Wikipedia category is unusable as an answer. A good deal of it is scholarly stubs (Antilabe, Hemistich, Tornada) or one line entries on traditions with almost nothing reachable in English. A quiz that can hand you a form you cannot then go and read is a dead end.

So the filter was two things: a distinguishable character, and enough accessible work that "read these three" is a real instruction. The obscure forms stayed in on purpose, because finding one you had never heard of is most of the point. Alongside the sonnet, the villanelle and the ballad sit the Pashtun landay, the Burmese thanbauk, the Vietnamese luc bat, the Levantine zajal, and Zulu izibongo. Where a tradition is oral or lives mostly in translation, the form says so plainly rather than implying an anthology exists that does not.

One form is a narrow answer

Handing someone a single form out of twenty four is not much use to a person who is trying to work out what they like. So the result opens outward in two different directions, which answer two different questions.

The shelf is the ranking: four more forms below the winner, each opening to its own three poems. The family is structural: the seven groupings the catalogue is built from, so a villanelle names the pantoum and the ghazal as its neighbours because they are all repetition engines. Those are not the same list. Two forms can rank next to each other while having nothing in common, and ranking alone would never tell you why.

It shows you the rules, in case you want to write one

Every result carries how the form is actually built, and a prompt for trying it. The villanelle's is to write one sentence you cannot stop thinking and a second that answers it badly, then build the rest to make them keep arriving. The rules are worth reading even if you never write a line, because knowing that a sestina rotates six end words in a fixed spiral changes how the Bishop poem reads.

The poems are picked for you, not for the bucket

Getting the form right is only half of it. A form is a shape, not a mood: the villanelle holds both Do not go gentle and The Waking, and giving two very different readers the same three poems wastes most of what the quiz just learned.

So every poem carries two tags of its own: whether it consoles, and how much work it asks. The three are ordered against your scores on those, with consolation weighted more heavily, because a poem that lands the wrong way emotionally is a worse miss than one that asks slightly too much.

It tells you when it is unsure

Any quiz can name a winner. Not every set of answers supports one. If the answers sat near the middle, or the top forms are separated by almost nothing, the result page says so and points at the shelf rather than presenting a coin toss as a reading of someone.

That threshold is the one number not carried over from Reader Type. Twenty four candidates crowd the top of a ranking more than sixteen do, so the gap between first and second is naturally smaller, and the older threshold would have fired on nearly a third of results. A hedge that frequent is noise rather than honesty, so it was measured again against this catalogue.

What it does

• Reads the preference, not the reading list. Twenty four statements about sound, repetition, brevity and constraint. None of them asks which poets you like.
• Twenty four forms. From the sonnet and the ode to the landay, the thanbauk and the golden shovel, across seven families.
• Three poems, ordered for you. Chosen from the matched form by how consoling and how demanding each one is, against your own answers.
• The rules and a prompt. How the form is built, and one way to start writing one.
• A shelf of five that opens up. The four runners up each show their own rules and their own three poems.
• Its family, named separately. What sits structurally next to your form, which is a different question from what scored near it.
• Mood asked separately. A few statements ask what you want this month rather than in general, since the two disagree.
• A profile you can read. All ten dimensions are shown with your score, so you can see what drove the result.
• Results are a link. The profile is encoded in the URL, so a result can be kept, bookmarked, or sent to someone without anything being stored or uploaded. It can also be copied or downloaded as text.
• Neutral is a real answer. If a statement does not resonate, say so, and it counts for nothing rather than forcing a guess.
• Light and dark, and the keyboard. Answer with the number keys, go back with backspace.

How it's built

Plain HTML, CSS, and JavaScript. No framework, no build step, no backend, no dependencies. It shares no code with Reader Type, which sits beside it in the same repository: the two are separate apps that happen to agree about how scoring should work. The whole model lives in one data file, so changing what the quiz believes means editing that file and nothing else.

Everything runs in your browser. There is no account and nothing is sent anywhere, so your answers and any feedback you leave never leave your machine.

What was checked, and what was left alone

Matching compares the shape of your profile against each form's rather than the raw numbers, so answering strongly across the board does not match everything. Both sides are standardised first, then each form is calibrated against how it scores for someone answering at random, which stops forms in crowded parts of the space from winning by default. The statements are balanced in both directions, so agreeing with everything produces no result at all.

Run against forty thousand random responders, all twenty four forms are reachable, none dominates, and twenty two of the twenty four rank first for their own ideal profile. The two that do not are near twins of a neighbour by construction: a ballad really is a narrative rhymed sung form and so is the Vietnamese luc bat, and what separates a clerihew from a limerick is that its metre is deliberately incompetent, which is not something the model has an axis for.

Both were left standing rather than corrected. So was free verse, which comes up less often than chance because its profile is low on nearly every axis. It is the most read form in English and the temptation to give it a nudge was real. Nudging one form because it is famous is exactly how an author's taste leaks into other people's results, so where a form comes up too often or too rarely the fix belongs in a rule that treats all twenty four the same, or in real responses, never in a correction to the one someone wanted.

The honest caveat

This is entertainment with a real model underneath, not a validated instrument. The statements borrow the style of personality inventory items but are not validated scales, the register values are judgement calls, and the catalogue is twenty four forms out of hundreds. Treat the result as a good place to start looking, not a verdict.