Scoring a work

This walks through scoring a single work on the Music profile. The same steps apply on any profile — only the axis names change.

Score the current draft

Say you have a song called “Harbour lights” with a lyrics draft as its current version. Open it and go to the Score tab. Each axis is a row: what it is called, what it weighs, the question it asks, and a scale you click.

{ "hook": 6, "lyrics": 7, "emotion": 7 }

Click a mark to set it; click the same mark again to take the axis back to unjudged. The scale is also a keyboard control: Tab moves between axes, the arrow keys move a mark at a time, Home and End jump to either end, and Backspace clears the axis. Judging six axes never needs the mouse.

The tab has the same shape as Versions: the scores given so far on the left, newest first, each with the draft it judged, its date and tier, and what it did to the one before (+5); the open one on the right, its axes scrolling above a footer that holds the total, the tier, the trail of the last few totals (78 → 86 → 91) and Record score. Clicking a score in the list opens its marks; New score at the foot of the list starts from empty scales.

A work that has been scored before opens with the newest recorded marks already on the scales — the verdict that stands, on the scales rather than only in the list. Save stays off until you move a mark, so you cannot record the same verdict twice by accident; move one and the scales are yours again. After a save the scales settle on the score you just gave.

You don’t need every axis filled. Leaving production, originality and visual blank doesn’t count against the work — they’re excluded from the total entirely rather than treated as zero. See Scoring.

The total updates as you go, with the tier it currently reaches. Record it, and kilna stores the total computed from the axes you gave and their weights, and reports which tier that total falls into — hold, audio, picture or clip on the Music profile’s default thresholds.

What the marks mean, and what the next tier costs

If your profile gives an axis a rubric, the sentence for the mark you are on appears under the scale as you move along it — so “is this a seven” is answered by the craft rather than by mood. A mark with no sentence of its own shows the nearest one below it.

Under the total, a ruler lays the tiers out to scale with the card standing somewhere along it, and a line says how far the next tier is and where it is cheapest to get there:

9.3 to Release it · cheapest on Hook (weight 2): 4 more marks

Cheapest means in marks, not in weight. A heavy axis usually moves the total furthest per mark, but not when it is already near its ceiling, and not when the axes run on different scales — so kilna works out what each axis would actually cost and names the one that costs least. If no single axis can close the gap, it says the distance and offers no axis rather than pointing at one that would not deliver.

The same arithmetic marks the scales themselves: on any axis that can carry the card over on its own, the mark where it would cross is ringed. An axis with no ring cannot get there alone.

What each release makes of it

A premiere lives or dies on its dynamics; the same video as an ordinary upload is carried by its fit to the track. If your profile gives a release kind weights of its own, the panel reads one set of answers down every channel you ship to and shows what each makes of it — one score, several honest verdicts. Kinds that weigh the axes the way the profile does are shown greyed rather than left out, so the row means “weighs them your way” and not “does not apply”.

Nothing appears here until some kind actually reweighs something: otherwise every row would repeat the total.

The panel normally shows you where a work stood last time — the history, the trend beside the total, a line per axis. That is useful when you are reading, and a thumb on the scale when you are judging: it is hard to give a six to something you called an eight in July.

Score blind puts the past out of sight. The history, the per-axis trends and the assistant’s opinion are held back until this card has a verdict of its own, and the scales start empty rather than on the recorded marks; recording a verdict brings everything straight back, which is the point — the comparison is the payoff, not the temptation.

What stays visible is your own arithmetic: the total these marks add up to, the tier it reaches, the ruler, the distance to the next tier. Those are a mirror of the verdict you are giving, not a hint about the one you gave before.

Holding a tier by hand

Sometimes the number is right and the verdict is wrong: a song is the only one of its kind, or it repeats a theme you have already published, and the axes cannot see either. Hold at a tier fixes the verdict where you want it and asks why — the reason is required, because a tier nobody can argue with later is a tier with a sentence attached.

A held tier travels: the catalogue shows it instead of the scored one, marks it with a pin, and stops offering a distance to the next tier, which would be advice about a number nobody is chasing. Scores keep being recorded underneath; the tier simply stops following them.

Follow the facts lets go, and the score decides again.

Who judged

A score can carry a name. Leave Who judged empty and the score is yours — that is what the field has meant since the model was laid down. Type a name and the history labels that row, so a second opinion stands next to your own rather than replacing it.

The name stays put after recording, because judging a batch as one person means typing it once rather than once per work.

The score is pinned to this draft

The score you just took is tied to the version of the lyrics that existed at the moment you scored it — not to the work in general. If you look at the work’s score history, this snapshot stays exactly as it was even after you keep editing.

Judging names which version is being scored. Left alone it means the current one, which is what you usually want. Pick another and the score is pinned to that draft instead — for going back and judging a version you had skipped, without disturbing what the work currently points at.

Why this score is a line for yourself. A total tells you what you decided; a note is the only place that keeps why, and three weeks later that is the part you will want. It shows up beside the score in the history.

Revise, then score again

Rewrite the second verse. Save it as a new version. Score the work again with updated values:

{ "hook": 8, "lyrics": 8, "emotion": 8 }

Both scores now exist in the work’s history — the first pinned to the original draft, the second to the revision. The total moving from, say, 62 to 78 is the visible record that the rewrite helped. Neither score is overwritten; a work’s score history only grows.

From the second score onwards a small line appears beside the total, drawn against the full scale rather than against its own range: a work that moved 61 → 63 looks like the small change it was. Green climbs, red falls. The exact numbers, with what each score changed by, are in the list underneath.

Each axis gets its own line too, beside its scale. The line under the total says the card moved; these say which axis moved it — a rewrite that lifted the lyrics and left the hook alone reads that way at a glance. Every score records what all the axes were worth at the time, so these lines come out of history you already have.

Reading staleness

If you edit the work again after scoring — even without adding a formal new version — the catalogue marks the work stale: its most recent score now describes an older draft than what’s currently in front of you. Staleness is a signal to re-score, not an error.

Where scoring feeds into

A work’s most recent score decides two things downstream:

Tier, shown wherever the work appears in a catalogue or list.
Order in the queue, strongest first — which is also the order the auto-layout places them in. See Planning a release.

An unscored work is not treated as bad — it’s treated as unjudged, and sorts last in the catalogue rather than at the bottom.