PROJECT NOTHING / JUDGMENTEXPERIMENT 044 · DECISION LAB
One task / many kinds of software

A hundred prototypes.
The machine has a favorite.

Same task: turn scattered notes into a project. Different interfaces. Different rules. The machine decides what to put in front of you. Will you choose differently?

01 INSPECT→02 CHOOSE→03 AUTO-ACCEPT
Machine recommends

Tell the machine what matters
OPEN A PROTOTYPE ↓

Your decisionNo choice made.

The machine’s favorite is open. You can choose any prototype.

One hundred ways to approach one task.

Twenty interface models, each with five ways to apply a change. These are small interactive sketches. They are not complete applications.

The machine directs your attention.

Its ranking follows the priority you choose. Draw three at random to look outside that ranking. A recommendation is a preference, not proof of quality.

Then the next choice happens without you.

Compare the automatic selection with your own. Audit it or reverse it. No real project is changed.

About Judgment

As software becomes easier to generate, deciding what deserves a closer look becomes its own task. This experiment makes that decision visible.

Each sketch has a small local interaction. Try an incoming note, then approve it, undo a write, merge a branch or watch it get filed without a confirmation. Change the priority and the machine's recommendation changes. Your choice stays separate.

The recommendation is a transparent local rule. There are no live model calls, measured performance rankings or claims of an objectively best interface. The random view samples three distinct alternatives without using the ranking, and draws a different set each time.

Notes and choices stay in this tab and disappear on reload. No research results or judgments are collected. The optional thumbs below send your vote.

PREPARED INTERACTION SKETCHES / Scripted ranking. No live model calls. No research data collection.