Week 05 · Foundations

Corner radius6px · 21%

BoxyPill

Nothing Streamlined
Ever Moved

A 1930s refrigerator got the nose of a race car. A pencil sharpener got speed lines. Neither one goes anywhere. Drag the corner below and watch a locomotive and a button round off at the exact same rate — then decide what that roundness is actually telling you.

Predict

Round off every sharp edge on a static object. Does it now go faster?

Part I

The Radius Bench

One card, one input, one button, and two purely decorative add-ons borrowed straight from a 1930s stylist’s toolkit: speed lines, and a teardrop taper on the leading edge.

Drag the radius toward a pill and the card reads calmer. Then start stacking the speed lines and the taper back on top of that same curvature, and watch the two readouts below stop moving together.

The Radius Bench

drag, then stack the speed cues

Friendliness: 37%; modelled — curvature only
Approachability: 37%; modelled — curvature minus speed cues
Radius / height: 0.11; not yet a pill

Corner radius6px

SharpPill

Speed lines0

None5 lines

Friendliness and Approachability are this page’s own invented formulas, not dossier figures — see streamline-math.ts for the exact arithmetic. Friendliness is driven only by curvature, saturating at the true-pill point (dossier 2.3: curved contours read as lower-threat, higher-preference “robust... at object and interface scale”). Approachability starts from the same number and pays it back down for every speed line and the taper — a stand-in for the dossier’s point (2.2) that a symbolic speed cue does associative work of its own, and that work is not always cozy. The radius/height ratio is measured, not modelled: a true pill (a stadium shape) begins exactly at ratio 0.50.

Part II

MAYA, and the Shape That Only Looks Solved

Loewy never wrote a numbered manifesto. The one line of his that survived is an acronym: MAYA — Most Advanced Yet Acceptable.

“The adult public’s taste is not necessarily ready to accept the logical solutions to their requirements if the solution implies too vast a departure from what they have been conditioned into accepting as the norm.” Push a design as far toward novel as it can go, then stop exactly before it crosses the line the audience’s existing mental model can absorb without discomfort. Novelty and comprehension are in direct tension: a completely unfamiliar interface has no learned affordance to draw on, and a completely familiar one has no reason to be preferred over the incumbent it replaces.

Underneath MAYA sits a second, sharper distinction the dossier calls symbolic versus structural function: a form can signal that a problem has been solved without solving it. A rounded corner on a button is almost always symbolic — it changes perceived friendliness, not tap accuracy. A larger tap target is structural — it changes the actual hit rate. Both are legitimate design tools. The mistake is letting a stakeholder believe the symbolic one solved the structural problem.

There is no dedicated Wikipedia article on MAYA under either candidate title (both 404, per the dossier’s own verification pass). The acronym and its definition are confirmed via the Interaction Design Foundation, which also supplies the quote above, word for word — but that quote could not be checked against Loewy’s own 1951 book directly (no digitized primary copy was reachable during verification). Grade the concept CONFIRMED; grade the exact wording of the quote confirmed-via-secondary-source, one remove from the primary text.

Part III

Five Commissions, One Move Each

Loewy’s consultancy worked across nearly every category of American manufacturing. Scrub through five commissions the dossier can actually date and source — each one gets exactly one design move, not a résumé.

Commission

1934 · Sears Coldspot1962 · Studebaker Avanti

1934 · Sears Coldspot — one continuous enameled shell hides the compressor entirely

A boxy, visibly-mechanical refrigerator became one smooth cabinet — Wikipedia credits the redesign with a five-fold sales increase within two years, with zero change to the cooling mechanism itself (Claude.md line 121).verified

Two of Loewy’s most-repeated credits are deliberately not on this scrubber: the Coca-Cola contour bottle (designed by Earl R. Dean at the Root Glass Company in 1915, four years before Loewy reached the US — his only real involvement was a 1955 reformatting) and the PRR GG1 (styled from an existing Donald Dohner design; Loewy refined it, he did not design it). The dossier calls the bottle attribution “the single most important correction in this file” and repeating either here would be exactly the error it corrects.

Part IV

The Bevel Disappears

Slide between a 2010 web button and one from 2023. The bevel, the gloss, the seam it was celebrating — all gone. What is left is one continuous, unbroken shape.

That is the same move as the Coldspot’s enameled shell, aimed at a different material: hide the seams, read as one deliberate object rather than an assembly of parts. Apple’s Human Interface Guidelines and Google’s Material Design both specify continuous, superelliptical — “squircle” — corner curvature as a deliberate choice, and the curvature-preference research behind that choice is real and independently sourced.

What is not documented, and the dossier is explicit about this: no source it fetched — not Wikipedia’s Streamline Moderne article, not Apple’s own design documentation, not Google’s — draws a historical line from 1930s Streamline Moderne to the 2010s–2020s squircle/pill button. The two share a psychological mechanism (curvature reads as friendlier, continuous surfaces read as unified), independently rediscovered in each era. That is a legitimate analogy. “Loewy invented the intellectual ancestor of the squircle” is not — it is this page’s own interpretive bridge, stated as one, not as a citation.

Part V

Spend the Novelty Budget Yourself

MAYA is a dial, not a switch. Push the same interface toward “advanced” and watch a modelled acceptance curve rise, peak, and fall away again.

MAYA dial

Novelty50%

FamiliarAdvanced

The acceptance curve is illustrative, not a dossier figure — no quantified acceptance-versus-novelty function appears in the source material. It stands in for a claim the dossier does make qualitatively: familiarity is (within limits) preferred, courtesy of the mere-exposure effect, and pure novelty imposes a real comprehension cost, so the practical operating point sits somewhere inside the fluency budget the audience can absorb — not at either edge.

Challenge · One radius system

Open

An input (44px), a button (56px) and a card (128px) each have their own corner radius right now, picked with no shared rule. Set all three so that radius divided by height is the same ratio everywhere.

Input radius (height 44px)4px · ratio 0.09

Button radius (height 56px)6px · ratio 0.11

Card radius (height 128px)8px · ratio 0.06

Ratio spread: 0.04 · not yet one system

Watch the ratio printed under each slider — match all three, and keep the curvature real (not zero).

Insight

The curve was never about the wind. It was about the eye.

A refrigerator has no drag to overcome. A pencil sharpener has no airflow to part. Every rounded corner and every speed line on those objects was aimed at a nervous system that reads curvature as safe and a wind-tunnel silhouette as modern — not at a Reynolds number. The same is true of the button under your thumb right now: its pill shape is not solving a structural problem any more than the Coldspot’s shell was. It is speaking directly to the same processing-fluency mechanism Loewy never had a name for but built an entire consultancy around anyway.

Mental model

Whenever you reach for a rounder corner, a smoother transition, or any other polish that changes how a surface feels,

look for whether you are solving a structural problem or a symbolic one, and say which out loud,

because a symbolic fix is legitimate craft only when it sits on top of a system that already earns the promise the form is making — the moment it hides a real failure instead, the gap surfaces later, in a support ticket, not a design review.

Tomorrowpick one rounded element in your product and ask whether its radius comes from a shared token or a one-off guess. If it is a guess, that is the fragmentation risk the dossier's own audit metric was built to catch.

End of experience

← Back to the constellation

Nothing StreamlinedEver Moved

The Radius Bench

MAYA, and the Shape That Only Looks Solved

Five Commissions, One Move Each

The Bevel Disappears

Spend the Novelty Budget Yourself

Where this idea goes next

Nothing StreamlinedEver Moved

The Radius Bench

MAYA, and the Shape That Only Looks Solved

Five Commissions, One Move Each

The Bevel Disappears

Spend the Novelty Budget Yourself

Nothing Streamlined
Ever Moved

Nothing Streamlined
Ever Moved