direction-finding

Derive from the audience's world. for a new world, name the product's mechanism, the audience's real scene, its cultural home — then list seven concrete visual systems, artifacts, or rituals the audience knows by heart. more than three in one material family means the derivation stopped at the most obvious artifact; dig until the list spans at least three families. graphic and screen traditions count: the notation, publications, and interfaces the audience reads daily.

The roll breaks the rut. a script deals which candidates reach the user. the point is mechanical: every model run converges on the same few category looks (warm cream + serif + terracotta; near-black + neon; broadsheet hairlines). self-check: if someone could guess your aesthetic from the category alone, rework until neither answer is obvious.

Verdicts and donations. challengers are fused with the product's facts, then weighed on exactly two axes — audience identification and product clarity — with a verdict per challenger: wins, competitive, or declined. a declined challenger still donates: name the one discipline its system has that yours lacks, and raise your direction to match. a donation transfers ambition and discipline, never the challenger's clothes.

The standing exit. every direction round offers the category standard, played straight, as a permanent quiet option — the user's door, never recommended by the tool. taking it makes convention the commitment: execute the canon at full fidelity.

Truth binds claims, not demonstrations. author whatever illustrative material a concept needs at full fidelity, label it synthetic; what stays uninventable are commercial claims — prices, customers, capabilities the product lacks.

generation converges: the same handful of category looks win the ranking every run, and asking harder for "something different" reproduces the same shortlist one step removed. the machinery — audience-derived candidates, a mechanical deal, forced verdicts — exists because no prompt intent beats a selection process that structurally cannot return the default.

new surfaces and replacement visual worlds — never for local extensions of an established surface, which inherit that world.

bad

brief: make a modern workflow product landing page
direction: dark background, neon gradient, glass cards

category default, not a choice grounded in the audience's world. a default look tells the exact people it addresses that nobody decided anything on their behalf, makes the surface interchangeable with every competitor, and leaves the audience's own visual vocabulary — the artifacts and rituals they already read fluently — unused as the cheapest differentiator.

good

audience world: operational runbooks, routing diagrams, live control rooms
challengers: transit maps, maintenance manuals, broadcast control panels
verdict: control-panel hierarchy with manual-like notation

derived, compared, selected through named evidence. an audience-derived direction reads as native to its targets, the verdict trail makes the choice defensible rather than tasteful, and declined challengers donate discipline to the winner — each round raises the floor of the system, not just the pick.

refinement vs redesign · the craft floor