Specification SYS-02 · Evidence OS · design system · v2

The system, documented as itself.

Evidence OS treats a portfolio as a working record: modular surfaces, sourced claims, logged decisions. v2 reduces the grammar on purpose: three signature devices carry the identity, one red-pen mark per case, and the stationery effects are retired. Structure first, texture second, and the texture stays in the margins.

SYS-02.A · Visual grammar — three devices sign the work

Each signature device owns one job. If a module needs a second device, the module is doing two jobs; split it.

Receipts = outcomes

Every quantitative claim, with source and scope attached.

Decision records = judgment

Alternative · why it won · what it cost. The crown jewel; never abbreviated.

Marginalia = caveats & reflection

Handwritten, red tick, 20px. Nothing essential ever lives in the margin.

Rule · RedPenCircle

At most one per case, always on the outcome panel’s key number, always with a small handwritten caveat. Two circles read as marketing.

Rule · horizontal lines

Rules mark structure, never rhythm: section tops, card frames, and table grids keep their hairlines. List items separate with whitespace alone; dotted leaders and dashed separators are retired (2026-07).

Rule · everything else

Stamps, pulls, route maps, and specimen frames support; they never sign. One per surface, at most.

SYS-02.B · Reading hierarchy

On any surface, content earns attention in this order. When two elements compete, the one higher in the chain wins the size, the position, and the contrast.

Story

prose · headers

Decision

records

Outcome

panels · receipts

Evidence

figs · sources

Caveat

marginalia

Metadata

refs · tags

SYS-02.C · Mono-caps usage limits

Mono caps may appear as

Kickers and section eyebrows.

Figure references and evidence sources.

Status labels: stamps, tags, chips.

Navigation and the theme toggle.

Nowhere else — specifically not

Body copy, deks, or pull quotes.

Headings of any level.

Whole sentences of any kind.

Anything a visitor must read to follow the story.

SYS-02.D · Type ramp

Display · Instrument Sans 700 · 48–58px · −0.025em · 1.1

Simple systems, receipts attached.

Heading · Instrument Sans 600 · 24–26px · −0.015em

Section heading, sentence case

Body · Inter 400 · 16px · 1.6

Body text carries the argument. It stays quiet, reads long, and never competes with the evidence it presents.

Quotes · Newsreader italic · 16–21px · 1.45–1.5

Quoted voices, pulled statements, and sourced corroboration read in Newsreader.

Note · Caveat · 20px · 1.32 · canonical handwritten

Margin notes speak in the author’s own hand, off to the side.

Label · IBM Plex Mono 500 · 10–12px · +0.09em · caps

Evidence receipt · fig. 1.1 · CS-01 · verified

SYS-02.E · Color tokens

paper

#F7F3EA · light bg

card

#FFFDF7 · raised surface

ink

#1C1A16 · primary text

ink-2

#5E574A · secondary text

moss

#33604A · verified metrics

amber

#B06E14 · decor (#8A5510 text)

stamp red

#A83A28 · stamps · fig refs

blueprint

#2C4E8A · routes · focus ring

dark paper

#151310 · dark bg

dark card

#1E1B17 · dark surface

dark ink

#EFE9DC · dark text

dark accents

#8CC7A6 · + E0A050 · E08472 · 93B0E3

Verified AA pairs (WCAG 2.1, normal text 4.5:1)

Pair

Theme

Ratio

Verdict

ink #1C1A16 / paper #F7F3EA

light

15.7:1

AAA

ink-2 #5E574A / paper #F7F3EA

light

6.4:1

AA

moss #33604A / paper #F7F3EA

light

6.5:1

AA

amber-ink #8A5510 / paper #F7F3EA

light

5.6:1

AA

stamp red #A83A28 / paper #F7F3EA

light

5.7:1

AA

blueprint #2C4E8A / paper #F7F3EA

light

7.4:1

AAA

text #EFE9DC / paper #151310

dark

15.4:1

AAA

ink-2 #B5AC9A / paper #151310

dark

8.2:1

AAA

moss #8CC7A6 / paper #151310

dark

9.6:1

AAA

amber #E0A050 / paper #151310

dark

8.2:1

AAA

red #E08472 / paper #151310

dark

6.8:1

AA

blue #93B0E3 / paper #151310

dark

8.5:1

AAA

SYS-02.F · Grid & spacing

Twelve columns, 1200px max width, 24px gutters at desktop. Bento modules sit on the grid with 1px hairlines and 2–4px radii. The reading column caps at 640px; marginalia takes a 220px gutter to its right.

Spacing scale: 8 · 14 · 24 · 36 · 64. Reuse these steps; gutters drop 24→16px on mobile and card padding drops one step, while section rhythm holds at 64px.

SYS-02.G · Component gallery

The full living specimen — every card, stamp, receipt, decision block, note register, work row, methodology drawer, and specimen frame, in every state — is maintained on the component-specimen page. Each component below documents one job:

Primitives

MarkNav

CM

BrandWordmark

Button

View case studies

NavLink

Chip

Evidence-first design

StatusTag

Live

RailTab

01

Context

Stamp

Verified

NoteHand

Note to reader

Margin notes speak in the author's own hand, off to the side.

Rule

CornerTicks

The gates did the work.

RedPenCircle

40%

Content

SectionHead

Impact

The receipts

Impact

The receipts

ScopeFact

My role

Lead product designer · sole designer on the product

My role

Lead product designer · sole designer on the product

CardStat

−20%

Curriculum dev time · beta

−20%

Curriculum dev time · beta

BigNumber

−20%

Curriculum development time

−20%

Curriculum development time

MetricPill

−20%

Curriculum dev time

Beta cohort · 2024

−20%

Curriculum dev time

Beta cohort · 2024

Receipt

Curriculum dev time

−20%

Beta partners completed curriculum-development tasks 20% faster · CourseBook · Minerva · 2024

Curriculum dev time

−20%

Beta partners completed curriculum-development tasks 20% faster · CourseBook · Minerva · 2024

Corroboration

Cody is a machine that converts ambiguity and chaos into clarity and order.

Garret Bassett

Product Manager · Minerva Project

Cody is a machine that converts ambiguity and chaos into clarity and order.

Garret Bassett

Product Manager · Minerva Project

Assemblies

CaseMeta

Org

Minerva Project

Years

2019–24

Type

0→1 platform

Status

In beta with select partners

Org

Minerva Project

Years

2019–24

Type

0→1 platform

Status

In beta with select partners

Org

Minerva Project

Years

2019–24

Type

0→1 platform

Status

In beta with select partners

ScopeBlock

My role

Lead product designer · sole designer on the product

Scope

Product strategy, research, IA, interaction design, testing, design system

Team

1 PM, 5 engineers, academic SMEs

Leadership

Set the product model, drove the greenfield decision, directed the testing program

Shared or outside my scope

Engineering architecture, final commercial rollout

My role

Lead product designer · sole designer on the product

Scope

Product strategy, research, IA, interaction design, testing, design system

Team

1 PM, 5 engineers, academic SMEs

Leadership

Set the product model, drove the greenfield decision, directed the testing program

Shared or outside my scope

Engineering architecture, final commercial rollout

My role

Lead product designer · sole designer on the product

Scope

Product strategy, research, IA, interaction design, testing, design system

Team

1 PM, 5 engineers, academic SMEs

Leadership

Set the product model, drove the greenfield decision, directed the testing program

Shared or outside my scope

Engineering architecture, final commercial rollout

TLDRCard

CS-01

TL;DR — the case in one card

Problem

A platform built for one university was being sold to many: two legacy tools, ten years of debt, no LMS integration, and the costs compounding.

What I did

Pitched a fresh build over merging the codebases, proved it with a working prototype, then designed the whole platform: build, teach, grade, one page model.

−20%

Curriculum dev time

Beta cohort · 2024

+27%

Grading efficiency

Beta cohort · 2024

In beta

Select partners

Timeline

7 months of active design

Status

In beta with select partners

Org

Minerva Project · 2019–24 tenure

CS-01

TL;DR — the case in one card

Problem

A platform built for one university was being sold to many: two legacy tools, ten years of debt, no LMS integration, and the costs compounding.

What I did

Pitched a fresh build over merging the codebases, proved it with a working prototype, then designed the whole platform: build, teach, grade, one page model.

−20%

Curriculum dev time

Beta cohort · 2024

+27%

Grading efficiency

Beta cohort · 2024

In beta

Select partners

Timeline

7 months of active design

Status

In beta with select partners

Org

Minerva Project · 2019–24 tenure

CS-01

TL;DR — the case in one card

Problem

A platform built for one university was being sold to many: two legacy tools, ten years of debt, no LMS integration, and the costs compounding.

What I did

Pitched a fresh build over merging the codebases, proved it with a working prototype, then designed the whole platform: build, teach, grade, one page model.

−20%

Curriculum dev time

Beta cohort · 2024

+27%

Grading efficiency

Beta cohort · 2024

In beta

Select partners

Timeline

7 months of active design

Status

In beta with select partners

Org

Minerva Project · 2019–24 tenure

DecisionBlock

DL-01

Build new instead of merging the codebases

Alternative considered

The team’s plan: integrate the two codebases and account systems, and keep modernizing in place.

Why it won

I partnered with an engineer and prototyped the core mechanism. It proved a fresh build feasible, and less effort than retrofitting.

What it cost

Migration risk. Years of live partner curriculum would have to move, and that risk became mine to own.

DL-01

Build new instead of merging the codebases

Alternative considered

The team’s plan: integrate the two codebases and account systems, and keep modernizing in place.

Why it won

I partnered with an engineer and prototyped the core mechanism. It proved a fresh build feasible, and less effort than retrofitting.

What it cost

Migration risk. Years of live partner curriculum would have to move, and that risk became mine to own.

DecisionLog

Workflow

Before · two tools

After · CourseBook

What it earned

BUILD

Multiple config levels and publish states, in a second tool, with error-prone syncing.

Built directly on the page. The thing you edit is the thing that ships, no publish step.

The 20%: unified authoring replaced two-tool stitching.

TEACH

A stage, a timeline, and controls competing for attention, with no certainty about the student’s view.

One shared page view, clearly marked live. Everyone sees the same thing; nothing private can leak.

Trust. The privacy worry from research, answered by structure, not a setting.

GRADE

A separate, time-consuming interface, away from the lesson it graded.

Inline grading on the page itself, with customizable rubrics and grading schemes.

The 27%: grading moved into the page, so the context switch disappeared.

BUILD

Before · two tools

Multiple config levels and publish states, in a second tool, with error-prone syncing.

After · CourseBook

Built directly on the page. The thing you edit is the thing that ships, no publish step.

What it earned

The 20%: unified authoring replaced two-tool stitching.

TEACH

Before · two tools

A stage, a timeline, and controls competing for attention, with no certainty about the student’s view.

After · CourseBook

One shared page view, clearly marked live. Everyone sees the same thing; nothing private can leak.

What it earned

Trust. The privacy worry from research, answered by structure, not a setting.

GRADE

Before · two tools

A separate, time-consuming interface, away from the lesson it graded.

After · CourseBook

Inline grading on the page itself, with customizable rubrics and grading schemes.

What it earned

The 27%: grading moved into the page, so the context switch disappeared.

OutcomePanel

Measured w/ beta partners

Measured results · beta cohort · 2024

−20%

Curriculum development time

measured, not modeled — method below

+27%

Grading efficiency

Attribution, stated plainly: beta partners completed curriculum-development tasks 20% faster and grading tasks 27% faster than in the legacy workflow. Early results, reported as exactly that.

Measured w/ beta partners

Measured results · beta cohort · 2024

−20%

Curriculum development time

measured, not modeled — method below

+27%

Grading efficiency

Attribution, stated plainly: beta partners completed curriculum-development tasks 20% faster and grading tasks 27% faster than in the legacy workflow. Early results, reported as exactly that.

Measured w/ beta partners

Measured results · beta cohort · 2024

−20%

Curriculum development time

measured, not modeled — method below

+27%

Grading efficiency

Attribution, stated plainly: beta partners completed curriculum-development tasks 20% faster and grading tasks 27% faster than in the legacy workflow. Early results, reported as exactly that.

ProjectCard

CS-01

·

CourseBook

·

Minerva

·

2024

0→1 platform

−20%

Curriculum dev time · beta

CourseBook: one platform, not two

A 0→1 platform to build, teach, and grade in one place, replacing two legacy systems.

Owned end-to-end · discovery through beta

Read the case

CS-01

·

CourseBook

·

Minerva

·

2024

0→1 platform

−20%

Curriculum dev time · beta

CourseBook: one platform, not two

A 0→1 platform to build, teach, and grade in one place, replacing two legacy systems.

Owned end-to-end · discovery through beta

Read the case

WorkRow

CS-01

0→1 platform

·

Minerva

·

2024

CourseBook: one platform, not two

A 0→1 platform to build, teach, and grade in one place, replacing two legacy systems.

−20%

Curriculum dev time · beta

Case study

CS-01

0→1 platform

·

Minerva

·

2024

CourseBook: one platform, not two

A 0→1 platform to build, teach, and grade in one place, replacing two legacy systems.

−20%

Curriculum dev time · beta

Case study

WorkRowCompact

CS-01

Minerva

CourseBook: one platform, not two

A 0→1 platform to build, teach, and grade in one place, replacing two legacy systems.

−20%

Curriculum dev time · beta

Case study

CS-01

Minerva

CourseBook: one platform, not two

A 0→1 platform to build, teach, and grade in one place, replacing two legacy systems.

−20%

Curriculum dev time · beta

Case study

SpecimenFrame

fig. 1

Corner ticks, ruler edge, sourced caption.

Shipped · beta

fig. 1

Corner ticks, ruler edge, sourced caption.

fig. 1

Corner ticks, ruler edge, sourced caption.

Shipped · beta

ContentsRail

Contents

01

Context

02

The reframe

03

Key decisions

04

The solution

05

Outcomes

06

Leadership

07

Reflection

01

Context

02

The reframe

03

Key decisions

04

The solution

05

Outcomes

06

Leadership

07

Reflection

01

Context

02

The reframe

03

Key decisions

04

The solution

05

Outcomes

06

Leadership

07

Reflection

MethodologyDrawer

Footer

SYS-02.H · Identity & icon set — locked 2026-07

Two marks, one tile, one reserved emblem. The primary lockup is the wordmark — “Cody Morrow” with a single red corner bracket (⌐, viewBox 24 · M3.5 3.5 H20.5 V20.5 · stroke 5.5 · rendered 0.68em) at its top-right; it owns nav and headers. P2 “shoulder anchor” — CM with a single red tick growing out of the M’s shoulder — is the icon mark everywhere under 24px.

Rule · which mark where

The wordmark lockup owns nav and headers; CM initials + tick (P2) are retained for small marks — favicon, social tiles, avatars, app icons. Never both marks on one surface.

Rule · the tile

When the context demands an opaque or app-shaped mark, use the tile: P2 reversed (cream + lifted red) on the ink square, rx 10, 5-unit padding. Never light master on ink or dark master on paper.

Geometry · P2, exact

Canvas 48×48 · CM Instrument Sans 700 · size 20.75 · tracking 0.2 · tick path M37.9 9.13 H44.9 V16.13, stroke 3.5 · light #1C1A16 + #A83A28 · dark #EFE9DC + #C96A57.

SYS-02.I · Card system (r3) — schema, variants, hover

One meta schema opens every card and row: REF · PROJECT · ORG · YEARS — TYPE. The ref token is red on every card. Titles are ink, sentence case, no periods.

Rule · V1 / V2 micro-variants

V1 (ref inline in the meta row) is standard for product and design cases. V2 (ref in a dashed left column) is reserved for leadership and practice-building cases — today, Measuring Design Impact only.

Rule · org naming

Org is always the company name alone: Verily, never Verily · Alphabet. A parent company may appear once in body copy, never in a kicker or title.

Rule · hover, all card types

Background tints to the hover fill; the title underlines in red (1.5px, 4px offset) while the CTA arrow nudges 3px right. Freestanding cards add a light shadow lift; rows take the fill only. 180ms, no border shift, no scale. :focus-visible adds the 2px blue outline.

Rule · numbering

One scheme everywhere a project appears: CS-00 through CS-05, identical on home cards, index rows, and case pages.

SYS-02.J · Mobile behavior

Two breakpoints do most of the work: 960px collapses layout, 640px drops density. Mobile is editorial reduction, not just stacking. Per component:

Component

Breakpoint

BP

What changes

Header nav

640px

Long labels swap to short; Home link hides on case views; wordmark subtitle hides. Targets stay 44px.

Contents rail

960px

Desktop rail becomes a sticky single-row tab bar; active tab tracks scroll.

Featured cards

860px

The trio stacks to one column; the CourseBook screenshot keeps its aspect ratio; V2 ref column persists at 56px.

How-I-lead rows

960px

Three columns become two: index stays left, behavior and proof stack right.

Work-index rows

700px

The right column (org · stat · link) drops under the one-liner and runs horizontal; red ref cells stay.

Scope block

960 / 640px

Five cells fold to two columns, then one. Color-coded labels stay.

Decision blocks

960px

Three cells stack; per-cell caps labels carry the structure.

Receipts & Ledger

640px

Card padding 26/28 → 20/16; labels 12 → 11px. Grouping and centered leader unchanged.

Route diagram

960px

The horizontal path hides; waypoints stack vertically joined by a dashed connector.

Margin notes

960px

The margin column folds into the reading column; notes keep their handwritten rotation, indented.

Mobile type ramp (390px)

display · 33px

case title · 32px

section head · 26px

case section · 24px

body · 16px

note · 20px

labels · 10–11.5px

SYS-02.K · Motion

Quiet, brief, always earning its place. Hover states 160ms (the unified card hover runs 180ms); standard transitions 200ms; marginalia fade-ins 280ms with an 8px rise, triggered once when a view opens. Easing is cubic-bezier(0.2, 0.6, 0.2, 1) everywhere. Cases carry a needle-thin 2px progress bar under the header. Under prefers-reduced-motion, everything nonessential is disabled and marginalia renders visible.

hover · 160ms

card hover · 180ms

standard · 200ms

marginalia · 280ms

ease · .2 .6 .2 1

SYS-02.L · Voice & microcopy

First person, calm, precise, annotated. Claims carry sources; caveats sit next to headlines, not in footnotes. Outcome language is attribution-honest: “beta partners completed tasks 20% faster,” never “my design reduced.” No hype, no vague scale words, real numbers only.

Metric caveat

“Measured with beta partners, not extrapolated.”

Missing data

“Not instrumented — ask me about the measurement setup.”

NDA boundary · reserved

“Confidential — no product UI here. I walk through the work live.” (No live use in v2.)

SYS-02.M · Do / don’t

Do

Pin a receipt, with scope, to every quantitative claim.

Ship a methodology drawer with every outcome panel.

Label influence versus execution in every case.

Keep texture in the margins; keep the reading column clean.

Don’t

Decorate with data you can’t source.

Restate a thesis a third time.

Use a second RedPenCircle anywhere in a case.

Fake age: no grunge, no coffee stains, no distressed type.

CMS note: cases, decision records, and metric pills map to Framer collections; the methodology drawer rides the case template as six plain-text fields.