The band pill: why a range is more honest than a number

Intelligence preparation answers questions — how deep is the approach, how much can the causeway carry, how many defenders are dug in — and almost none of those answers is a fact. They are assessments: judged from partial, ageing, sometimes single-source reporting. The moment such a judgement is written down as a single crisp number, it starts to lie. A planner three desks away reads "30 tonnes" and treats it as surveyed truth, because nothing on the screen says otherwise.

ASSAY's answer is the band pill: the demonstrator's smallest and most-used component, and the first thing a subject-matter expert will test. It renders an assessed quantity as the closed interval the source actually supports — lo–hi unit — with no midpoint anywhere. Not in a marker, not in a centre tick, not in a hidden tooltip average. A midpoint is an invitation to false precision, and it will end up on a surface (ASSAY-DEC-15).

Drag it yourself

Below is the actual shipped component — the same bandPill() the gallery and wireframes use — rendering the Meridian causeway-capacity assessment (K2, source class assessed). Drag the control: every frame is what the real component produced for that band. Pull it all the way down and you reach the scalar — the tempting single number the assessment cannot honestly support.

Live component, fixture data. The pill is the demonstrator's own code (src/components/bandPill.ts); the numbers are real Meridian fixtures (K1, K2). Only the slider chrome is embed-local. Frozen as shipped on 2026-07-13.

The one exemption, kept narrow

Not everything is an assessment. A directly-observed value — a charted controlling depth, an own-force figure — is fact, and banding it would be an absurdity that discredits the very discipline it is meant to protect. So observed values (and a commander's directive thresholds) may render unbanded (ASSAY-DEC-14). The embed shows one: K1's controlling depth, carried as a degenerate band and coloured differently so the eye can tell a fact from a judgement at a glance. The exemption is deliberately narrow — only what is directly charted, surveyed, or own-force — so it cannot leak.

Why it is load-bearing

The band pill is not decoration. Because assessed quantities stay banded all the way through — representation, scoring, and rendering — plan verdicts are judged in banded space, and a verdict only flips at a band edge, never inside one. One SME seeing a confident number they know is guesswork would discredit the whole pipeline (ASSAY-DEC-9). The pill is where that promise is kept in the first pixel.