Qpoint Design

Pages

Teams & Users — Strip Overdrive

Source: qdash /teams — c67 graph-heavy overshoot study, 2026-08-15

The brief: go graph-heavy, draw on the DNA/rich-expression explorations, and deliberately go too far — because the line only becomes visible from the far side. Six treatments of the same 26-team world; each commits named crimes, filed in a report under the graphic. One is designed to survive. Same fixtures as the strip series ($18.4k · 214 people); per-person detail is hash-generated and labeled as such.

Team DNA — the helix wall

amplitude = spend · one rung per person · twist = agents/head

Crime report — metaphor debt: the helix sells the method, not the answer (c62's own verdict on this form). Unreadable channel: nobody can compare sine amplitudes — ML-Infra's wave is 2.4× Support-Ops's and you cannot see it. Rungs vanish at the nodes, so people near crossings disappear. Beautiful, and it answers none of the four jobs.

The colony

membrane area = spend · nuclei = people · cilia = agents

Crime report — misjudged area: blob area is honestly ∝ spend, and it doesn't matter — wobbled organic outlines defeat area judgment (humans are ±40% on circles; worse on amoebas). Aesthetic sorting: positions are hash-scattered for composition, so no order carries meaning. Ink without a question: cilia count agents, which nobody can count. Identity survives only in the labels.

Growth rings

ring thickness = spend · segments = members · red notches = findings

Crime report — fabricated level: rings mean age. There is no per-team history in the data — the form asserts a growth story nothing backs (portal law 5, broken). Radial distortion: inner teams get geometrically shorter arcs for equal shares; ring order is spend-descending, i.e. arbitrary — the tree's "core" is just the biggest spender. Equal-thickness comparison across radii is a known illusion.

The loom

people ⇄ agents ⇄ models · ribbon width = tokens · color = model

Crime report — encoding overload → hairball: 34 ribbons, and the dominant visual event is their crossings, which encode nothing. The beauty is the noise. Ink without a question: "which person feeds which model through which agent" is none of the page's four jobs — it's an inventory query wearing a gala gown. At 214 people this is solid ink.

The barcode — DNA as texture

one bar per person · height = log spend · color = state · all 214 people

Survivor's report — this is the DNA idea read honestly: texture, not metaphor. Nobody reads 214 bars; everybody reads the crowd — Payments-EU's two red spikes, the amber mass of Unassigned, the quiet grey of 21 fine teams. Every bar is a real fixture person with a tooltip and (ship) a click — the portal grammar holds. Length and position do the work; status color means status. Its one crime — sub-pixel bars at fleet scale — is tamed with a 2.4px minimum and per-row width. Graduates: candidate focus-card interior above ~12 people, and a full-width fleet-texture band.

The cockpit — everything, everywhere, pulsing

the anti-page: all six crimes in one frame, on purpose

Crime report — all counts filed: 6 fabricated time series (stamped — the one honest pixel here) · a rainbow scale · a dual-axis panel (the #1 chart crime: two y-scales, one plot, any story you like) · 18 indefinite pulses where the house rule is exactly one "live" mark · red spent as a picture frame · a gauge with no numbers measuring "team health™". Each element is defensible somewhere; together they total to a page that is all signal-shaped and zero signal.

Autopsy

The six ways to err too far

1Metaphor debt

The visual explains its own cleverness instead of the data. You spend the reader's attention teaching the form, then have none left for the answer.

guilty: helix (c62's own verdict: "sells the method, not the week")

2Unreadable channels

Amplitude, organic area, and arc-at-radius are spent where length or position was owed. If a human can't decode the channel within 10%, it's texture pretending to be measurement.

guilty: helix (sine height) · colony (blob area) · rings (radius)

3Fabricated levels

The form demands data that doesn't exist, so the data gets invented — implied age, imagined trends. Portal law 5: the descent stops at real data.

guilty: rings (implied growth) · cockpit (stamped sparklines)

4Aesthetic sorting

Layout chosen for composition, not comparison. The moment position stops meaning anything, the reader's strongest channel is spent on nothing.

guilty: colony (hash-scattered) · rings (order = spend, reads as age)

5Ink without a question

Channels that answer none of the page's ranked jobs. Every mark should trace to a question someone brings to /teams; crossings and cilia trace to nothing.

guilty: loom (the crossings ARE the picture) · colony (cilia)

6Motion & status as decoration

Pulse means "live" — once. Red means "critical" — only. Eighteen pulses and a red picture frame spend the two most expensive signals on ambience.

guilty: cockpit (18 pulses, red frame, rainbow gauge, dual axis)

The line, located

Expressiveness errs the moment a channel stops answering a question and starts decorating one. The barcode survives because every pixel still answers — length is spend, color is state, the crowd is the team, and each bar opens a person. The helix, colony, rings, and loom all spend their best ink on things no one asked (sine height, membrane wobble, ring order, ribbon crossings). And the cockpit shows the sum: maximal signal-shape, zero signal. Ship path for the survivor: the barcode row replaces the dot grid inside the focus card above ~12 people, and a full-width fleet-texture band is the candidate for the roster's header on the 214-person page — both keep the portal grammar (every bar a door).