---
title: Mixing The Scales
type: anti-pattern
level: L1
status: live
revision: 1
updated: 2026-08-14
systemVersion: 4.2
tags: [visual, communication, clarity]
rating: 7.40
ratingAxes: useful 8 · evidence 6 · pull 7 · original 8 · form 9
ratingKind: derived
source: visual system, in production
---

# Mixing The Scales

_Written 2026-08-14 · last verified 2026-08-14 · system v4.2 · live_

**TL;DR** — A system that scores several different things — urgency, suitability, confidence, completion — tends to render them all the same way. Once red means both urgent and unsuitable, a reader has to infer which question is being answered. Giving each scale its own shape removes the ambiguity for free.

## Pattern

Output carries 6 different kinds of rating. Priority. How suitable an option is. How confident the system is. How complete something is. What tier a contact belongs to.

Each is genuinely useful. All 6 get rendered as coloured dots, because coloured dots are what is at hand.

## Why it looks right

Consistency is normally a virtue in an interface, and one visual language across a system looks deliberate. Red-means-attention is also the most learnable convention there is.

The collision is invisible while each rating appears alone. It only bites when two appear in the same paragraph.

## Why it fails

The reader is doing an extra decode on every symbol: *which question is this answering?* When red means both **urgent** and **poor option**, a red mark beside a recommendation is ambiguous in the worst possible way — it could mean act now or do not do this.

Worse, the ambiguity is silent. Nobody reports being confused; they just occasionally read it the other way.

## Instead

**Give each scale its own shape.** Colour carries intensity, shape carries which question:

> squares `🟥🟧🟨🟩` — how good an option is · circles `🔴🟠🟡🟢` — how urgent · blocks `█░` — how complete · rings `●○` — how confident · stars — quality rating

Now a shape answers *which question* before colour answers *how much*, and the two decodes stop interfering.

Two rules make it hold. The mapping is written down once, centrally, rather than being re-invented per output. And a new scale must claim an unused shape — if none of the 5 remaining shapes is free, that is a signal the system is measuring more things than a reader can hold, which is a content problem wearing a formatting costume.