There is a particular satisfaction in building something that nobody sees. A design system is exactly that — a set of decisions so well made that users experience only the product, never the scaffolding beneath it.

I’ve spent the better part of two years obsessing over this idea. Not because I think systems are inherently beautiful (though they can be), but because I’ve watched enough products fall apart to know what absence of structure looks like.

The Problem with “Just Design It”

Most design decisions start with good intentions. A button is blue because blue felt right. A card has 16px padding because the previous card had 16px padding. A font is Helvetica because Helvetica is safe.

This works until the second designer joins. Or the second screen gets designed. Or someone opens the app on a different device and the button is no longer quite the right shade of blue.

Without a system, decisions that felt like decisions were actually just moments. They don’t repeat. They don’t scale. They accumulate technical debt in the form of visual inconsistency.

What a Real System Looks Like

A real design system is not a component library. That is a common mistake. A library is what you end up with. A system is the set of constraints and intentions that produced the library.

Start with tokens. Before any component is designed, you need a vocabulary:

• Colour tokens: not blue-500 but action-primary, surface-elevated, text-muted
• Spacing scale: a mathematical progression (4, 8, 12, 16, 24, 32, 48, 64) that every component uses
• Typography scale: size, weight, line-height — all derived from a single base
• Motion tokens: duration, easing, delay — so animations feel cohesive

With tokens in place, components become nearly automatic. They are just compositions of token values.

The Hard Part Is People

Building the system is the easy part. Getting people to use it is the challenge.

A design system only works if it is easier to use than it is to ignore. That means documentation that is honest about trade-offs. It means components that handle edge cases gracefully. It means a feedback loop where designers and engineers can propose changes without starting a three-month discussion.

The systems that die are the ones that become monuments. They stop evolving because they were declared finished. A living system is maintained like a garden — pruned, seeded, occasionally redesigned from the root.

On Invisibility

When someone uses a product and everything feels right — the spacing is comfortable, the transitions feel natural, the language is consistent — they don’t think about the system. They just feel at ease.

That invisibility is the point. The system succeeds by disappearing. All the decisions, all the arguments about token naming, all the component reviews — they crystallise into a thing that just works.

That is enough. More than enough.

Six months ago I bought a Leuchtturm1917 and a Pilot G2 and made a rule: every idea that mattered had to pass through paper before it went anywhere digital. I didn’t expect it to change how I work. It did.

Why Paper at All?

There is a frictionlessness to digital tools that can be counterproductive. Notion pages can be restructured in seconds. Figma frames can be multiplied infinitely. When everything is editable, nothing feels committed. Ideas live in a state of perpetual draft.

Paper has friction. You can’t undo. You can’t duplicate with Command-D. You can’t search for something you half-remember. These limitations force you to be more deliberate — to actually think before writing rather than thinking by writing and deleting and writing again.

What Changed

The first thing I noticed was the quality of my morning thoughts. Before the notebook, I would open my laptop and immediately start responding to whatever was urgent. The urgent always won.

With the notebook, I started spending fifteen minutes before the screen: reviewing what I was working on, sketching rough ideas, writing observations about things I’d noticed. By the time I opened a design tool, I had something to bring to it.

The second thing I noticed was that ideas surfaced differently. When I’m designing digitally, I tend to iterate in one direction — I start with an idea and refine it. On paper, I’m more likely to take a left turn, cross something out, draw an arrow to a completely different thought. The messiness of the medium seems to unlock a different kind of thinking.

The System I Settled On

After a few chaotic months, I developed a loose structure:

Morning pages (10 min): Freeform. Whatever is on my mind. No editing. Often incoherent. Surprisingly useful.

Project spreads: Two facing pages per project. Left side: constraints and non-negotiables. Right side: ideas, sketches, questions. Updated whenever I sit down to work on that project.

End-of-day review (5 min): One sentence about what I made progress on, one question I’m still sitting with. This single habit improved my memory of my own work dramatically.

Weekly sweep: Everything from the week gets reviewed. Anything worth keeping moves to Notion or a reference system. Everything else is acknowledged and left.

On Losing Ideas

People often ask: what if you lose the notebook? Doesn’t that feel risky?

Yes. And I think that’s fine.

If an idea is truly important, it will survive without the notebook. If it doesn’t survive, maybe it wasn’t as important as I thought. There’s something clarifying about a storage medium you can’t search. It forces you to only keep what is genuinely useful.

A Note on Pens

The G2 is correct. This is not up for debate.

I spent a month last summer doing something completely unrelated to my work: sketching buildings. Not digital renders, not 3D models — pencil on paper, buildings I walked past, façades of train stations, the corner of a parking structure that caught the light interestingly.

I didn’t do it to become a better designer. I did it because I was tired and wanted a practice with no stakes. But it changed how I think about interfaces.

Proportion Is Not Decoration

Architectural proportion is structural. The golden ratio and its variations were not decorative choices — they were discovered to produce buildings that felt balanced at human scale. When the width of a column relates mathematically to its height, and when that height relates to the spacing between columns, the result feels effortless without the viewer knowing why.

This is exactly what good spacing scales do in interface design. When the spacing between elements follows a consistent mathematical relationship — say, a scale of 4, 8, 12, 16, 24, 32 — the interface feels settled. Nothing fights for attention. The eye moves naturally.

When I sketch buildings, I’m constantly measuring. Not in absolute terms, but relationally. How does this window’s width compare to its height? How does the height of the first floor compare to the second? These ratios are the invisible structure of the building’s visual logic.

Negative Space Is Not Nothing

One of the most counterintuitive lessons of architectural sketching: the space between things is as designed as the things themselves.

In a great building, the void between two columns is not leftover space. It is held space. It has weight and intention. The building breathes.

In an interface, whitespace is often treated as what you have when you run out of content. This is backwards. Generous negative space is what makes content legible. It is what creates the sense that a screen has been considered rather than filled.

The buildings I find most calming to sketch are not the most ornate. They are the ones where the emptiness feels deliberate — where the architect knew that leaving space was itself a decision.

Rhythm and Repetition

Façades often have a rhythm: column, void, column, void, column. This repetition creates predictability, and predictability is oddly comforting. You know what to expect. Your eye can rest.

Navigation bars, card grids, list items — these are interfaces using the same principle. Repetition signals system. It tells the user: this is how we do things here. Trust it.

The danger is monotony. Architecture solves this with variation within the rhythm: a slightly wider column here, a recessed element there. Interfaces solve it with visual hierarchy — not every item in a list is the same weight, even if they all follow the same structural pattern.

What I Brought Back

After that month of sketching, I started thinking about interfaces more spatially. Not as screens to fill, but as places to inhabit. Places have proportion. Places have rhythm. Places have held space and considered voids.

The best interfaces I’ve designed since feel less like products and more like rooms. I don’t know if that’s exactly right. But it’s closer to right than what I was doing before.

At some point in the last decade, digital design became allergic to texture. Flat design arrived and everything smoothed out. Gradients disappeared. Shadows became subtle. Noise vanished. The screens got cleaner, more geometric, more confident.

And then, quietly, texture started coming back.

Not as retro pastiche, but as a considered aesthetic choice. Film grain on hero images. Paper textures on backgrounds. Rough brush strokes in illustration. Risograph prints and photocopier aesthetics. Noise layers on gradients. The handmade, infiltrating the machine-made.

I want to think through why this happened, and why I think it’s not going away.

The Problem with Perfect

Flat digital surfaces have no depth cues. No texture means no information about material. No grain means no information about how the image was captured. No shadow means no information about spatial relationship.

This is perceptually strange. Our visual system evolved to read surfaces — to understand from subtle variations in light and texture what something is made of, whether it can be touched, whether it’s far away or close. Flat digital surfaces give our visual system very little to work with.

The result, I think, is a kind of subliminal emptiness. The surface is perfectly legible — you can read every pixel — but it doesn’t feel like anything. It has no material presence. It is pure information without substance.

What Texture Does

When you add grain to an image, you’re introducing variation at the pixel level. This variation mimics the natural variation in analog photography, where film grain appears because light hits individual silver halide crystals. It is noise in the signal.

But our brains don’t read this as noise. We read it as warmth. As presence. As something that was captured rather than generated. The grain carries a kind of indexicality — a trace of a physical process even when the image is entirely digital.

Paper textures do something similar. A slight cream warmth and surface roughness suggests physical substrate. It activates associations with books, letters, documents — things we trust because we’ve held them.

This is not nostalgia. It is a sophisticated response to a real perceptual need.

Using Texture Well

The designers who use texture badly use too much of it, or use it arbitrarily. A rough paper texture behind a sans-serif in a tech startup product makes no sense — the texture and the content are not in dialogue.

The designers who use texture well treat it as material: they ask what the texture says, and whether that meaning aligns with the content it sits beneath.

Grain on photography says: this was a moment, not a render. Paper texture says: this is considered, slow, worth reading. Risograph texture says: this is handmade, limited, a little imperfect on purpose.

None of these are neutral. They are all claims about the work’s relationship to time, craft, and making.

A Personal Note

I’ve been adding a very subtle noise layer to almost every gradient I design for the past year. Not because it’s fashionable — because gradients without noise look slightly wrong to me now. Banded, synthetic. Like they haven’t committed to being a surface.

The grain makes them real enough. And real enough is all I’m after.

My copy of The Death and Life of Great American Cities is barely readable. Not because the text is worn — the text is fine — but because the margins are full. Pencil marks, asterisks, questions, occasional objections, one very long note that continues onto a Post-it and then onto a separate sheet tucked between pages 214 and 215.

This is marginalia. The notes we leave in the edges of things.

I have been thinking about marginalia a lot lately, because I think it describes something important about how we actually process ideas — and because I think most digital reading tools handle it badly.

What Marginalia Does

The note in the margin is not summarising. It is thinking. The reader is having a conversation with the text — not receiving it passively, but pushing back, making connections, flagging uncertainty, celebrating a particularly well-made sentence.

Highlighting is passive. Underlining is passive. A marginal note is active. It requires the reader to formulate a response, however brief.

Cf. p.47 means: I noticed something here that connects to something earlier. ? means: I don’t believe this. Yes. means: exactly this. How? means: the claim isn’t justified.

These are annotations with semantic content. They record not just what the text said but what the reader thought.

The Digital Marginal Problem

Kindle highlights are stored in a separate system. Notion comments float separately from the text. Hypothesis annotations are overlaid on web pages in a parallel layer. In every case, the annotation is spatially separated from the text — it exists about the text rather than with it.

Paper marginalia exists with the text. The handwriting is cramped into available space. You can see at a glance how much was noted on this page versus the previous one. The density of annotation is itself informative.

When annotation moves into a separate layer — even a linked one — something is lost. The conversation between reader and text becomes a file attached to a document rather than a lived relationship with a physical object.

I don’t know exactly how to solve this in digital reading interfaces. But I notice the problem.

Incidental Space in Design

The concept transfers to interface design. The margins, the padding, the whitespace — these are typically understood as structural (they create breathing room, separate elements). But they could also be understood as the space where the user’s attention rests between active engagements with content.

What happens in that space? Usually nothing. The user moves on. But the incidental space of a well-designed interface is where rhythm is established, where the eye pauses and resets, where the pace of reading is controlled.

The margins, in other words, do work. Not the work of content, but the work of pacing.

My Marginalia Practice

I have a rule: if a book doesn’t deserve pencil marks, it doesn’t deserve my time. This sounds harsh. What I mean is: passive reading is not really reading for me. If I finish a book without a single marginal note, I’ve probably consumed it rather than engaged with it.

The notes don’t have to be good. They can be one word. They just have to be there — evidence that I was present for the text, not just in the room when it happened.

There is a specific quality of attention that appears after midnight. I don’t know if it’s neurological, environmental, or some combination of both. But I know it’s real, because I’ve been chasing it for ten years.

Between midnight and 2am, the world quiets. Not fully — the city doesn’t fully quiet, and neither does the mind. But the surface noise drops. The emails stop arriving. The notifications thin out. The day’s decisions, made and unmade, recede. What’s left is a kind of stripped attention: fewer things competing for it, more available to direct.

For me, this is when certain types of creative work become possible.

What Changes at Night

The quality that I notice most is a reduced self-censorship. During the day, I’m aware of the work in context — aware of how it might be received, what the deadline is, whether this is the right direction. This awareness is useful. It keeps me practical and calibrated.

But it also slows the generative phase. Ideas get evaluated before they fully form. The inner editor is too quick. The practical mind interrupts the imaginative one.

At night, the practical mind is tired. It stops making objections. Ideas surface that wouldn’t have survived the daytime internal critique long enough to be useful. Some of those ideas are genuinely bad — the night removes one filter. But some are things I wouldn’t have found otherwise.

The Particular Texture of Night Work

There is also something about the physical experience of working at night that I find irreplaceable. The desk lamp in the dark room. The city visible through the window as a pattern of lights rather than a surface of detail. The particular silence that is not quite silence.

I have done a lot of design work looking out at a city at night. The bokeh of the streetlights. The movement of cars that are far enough away to be abstract. There is a way in which this visual input is perfectly calibrated for creative work: present enough to ground you, abstracted enough not to compete.

I think this is related to why so many writers have written in cafés, or to music — some ambient noise and visual motion that activates the senses without directing the attention. The city at night is the best version of this I’ve found.

A Note on Sustainability

I’m aware that none of this scales. Midnight sessions as a creative practice work when you’re single, childless, freelance. They work less well with a job that starts at nine. They stop working at all if they become a compensation for insufficient daytime focus rather than a supplement to it.

I’ve had periods where I worked exclusively late because I couldn’t concentrate during the day. That’s not what I’m describing here — that’s avoidance, and it’s exhausting.

What I’m describing is selective. A few nights a month when the work calls for a different kind of attention. When something needs to be made that won’t survive the day. When the stripped attention of 1am is the right tool for the problem.

The Work That Emerges

The things I’m most satisfied with — certain pieces of writing, a few design directions that changed the whole project — have almost all happened at night. Not all of them. But enough to make me take the association seriously.

I’m not sure what the mechanism is. I may never know. What I know is that I’ve stopped feeling guilty about it, and started treating it as a feature rather than a bug.

The city lights don’t explain it. But they’re always there when it happens.