A per-question cue in the GuideSidebar — a scent of where the meat is — without a column of “6/8” fractions (leaderboard / gap-shaming) and without colliding with the codebook’s state dot.
design-codebook-state-model.md) — a binary state. The first-draft evidence ramp was also a blue dot in a sidebar row, just at varying opacity — so “strong evidence” and “codebook enabled” look identical.
The fix: keep state = blue dot; make magnitude read as a scale — a different shape (count / height / length / size) in a neutral tone. Sentiment hues are spoken-for too, so magnitude stays monochrome.
Six guide questions at descending evidence strength. Column 1 is the codebook state dot we must not look like; column 2 is the current (colliding) ramp; 3–6 are the candidates.
| Guide question | State dot (codebook — avoid) |
V0 · opacity ramp (current — collides) |
V1 · pips | V2 · signal bars | V3 · sparkbar | V4 · dot-size |
|---|---|---|---|---|---|---|
| Last analysis experience | ||||||
| First impressions of export | ||||||
| Biggest frustration | ||||||
| What’s missing | ||||||
| Naming the themes | ||||||
| Anything else |
V0 and the state dot are visually indistinguishable at a glance — that’s the bug. V1–V4 all read as a quantity, in neutral tone, so “on/off” (blue) stays free for the codebook’s meaning.
A dot in a table reads differently from one next to a wrapping label. Left is the codebook sidebar (the reference we must not clash with); the rest are candidates in the GuideSidebar.
Blue = on, grey = off. State, not magnitude. This is the language we must not reuse for evidence.
Ascending height = strength; the cellular-signal metaphor says “level” instantly. Unmistakably a scale, never a switch.
Count = strength. Dead simple, scannable, and monochrome so it can’t be read as the blue state dot. Runner-up.
Keeps the single-dot shape you liked, but magnitude = diameter in a neutral hue — drops both the blue and the opacity trick that caused the clash. Minimal-change option.