A Big Five mapper for family dynamics
Thirteen people, five lenses, one canvas of force-pulled stars. Less a personality quiz, more an instrument for spotting family fault-lines and bridge-builders.
The frame
The constellation. The dossier. The pair. Each surface answers a different question about a person, a pair, or a system.
A PixiJS canvas where every person is a star pulled by d3-force. Family-color halos, weighted edges, click to enter a profile.
One person, five lenses. Big Five aspects, the 16-type map, DISC wheel, tesseract, plus a written dossier. Scroll-spy nav tracks the lens you're reading.
Two people, one diff. Trait-by-trait deltas with a Claude-generated reading on top — where they bridge, where they grind. The same engine extends to groups of 3+.
How it works
Four stages — calibration, drawing, reading, and reconciling. Each one earns the next.
Type in "ENTP-A" or "INFJ" and the system seeds a Big Five baseline. Want sharper? Open the calibrator and answer ten aspect-level questions to convert MBTI into proper Big Five aspects.
Step 1 of 2 · choose a type
Every person becomes a star. PixiJS renders the field; d3-force pulls similar profiles together and pushes opposites apart. Halos carry family color. Edges are weighted by aspect-distance — bridges glow, fault-lines feather thin.
Click a star. The dossier opens with up to five lenses: Big Five aspects, the 16-type map, the DISC wheel, the Tesseract, and a long-form written dossier. Scroll-spy nav tracks where you are; each lens explains the others.
Pick any two stars. A pair page diffs their aspects, then Claude reads the diff and writes one paragraph on where they bridge and where they grind. Same engine extends to groups of three, four, the whole table.
A note on groups
The pair generalizes. Pick a table, a road-trip car, a project team — the system infers a group profile and names the role each person tends to play.
What surrounds it
A small, deliberate set of pieces. Each one earns its slot.
Canvas-rendered stars with pre-baked halo textures. d3-force pulls similar profiles together. Holds 60fps on mid-range mobile.
People, aspects, pair-cache, edges. Single source of truth — the canvas reads, the dossier reads, the pair page reads.
Generates pair-readings and group inferences from the diff. Cached on first compute; regenerated when an aspect moves.
App Router, React 19. The canvas is a client component; everything else is server-rendered for a fast first paint.
Glossary
Plain-language definitions. Skim this once and the rest of the site reads cleanly.
Open the instrument
Thirteen stars. Five lenses each. Pick anyone — start with whoever you understand least.