# Not precious, not in lanes

Before anyone reads the title wrong: deep craft still matters, and real specialists matter more than ever.

I came up through design, starting out as a mobile UX/UI designer and growing into leading design across forty-three iOS and Android apps at Zapporoo here in Miami between 2014 and 2016. That wiring never really leaves you. I still notice immediately when a primary button sits four pixels outside comfortable one-thumb reach, or when white type blooms muddy over a dark photo.

What *has* changed in 2026 is everything happening at the seams between our job titles. When trying out an idea takes twenty minutes in a cloud container instead of three days in an editor, two old habits stop paying rent: getting precious about your first draft, and staying parked inside your lane.

## Precious protects a pixel; caring protects the person using it

When a draft takes three days of manual labor to mock up, throwing it in the trash hurts mostly because of the sunk three days. Drop the cost of a realistic try down to twenty minutes, and killing your own lukewarm ideas becomes weirdly fun.

While building [Home Port](https://andrewdurden.com/), I threw my own suggestions overboard constantly:

- Early on I wondered if a Miami Vice neon palette might pop (`"Sure let's try Neon for now?"`). We rendered it over real room geometry, realized it looked cold and gimmicky, and swapped to warm twilight walnut and plaster before lunch.
- An early glass-and-brass card strip looked fussy (`"I don't like that brass plate aesthetic for cards"`), and my first pass at intricate spot illustrations felt overcrowded (`"it's a bit busy with these"`), which pushed us straight into the tactile cloth-bound book covers that anchor the room now.
- When a job was about to try cooler lamps while our living-room rug still looked washed out, I killed the side quest in one all-caps text: `"DONT WASTE YOUR TIME ON COOLER LAMPS, CANCEL THAT JOB - MAKE THE RUG THE DARKER BLUE YOU JUST SHOWED AND MOVE ON"`.
- Twice during the build, instead of trusting our own echo chamber, I exported the build to a second AI assistant to rip apart our UX assumptions and blind spots, kept the sharpest critiques, and threw out the rest.

Being precious and caring deeply look similar from across a conference table, and they're opposites in practice. Being precious protects a pixel or a framework you got attached to. Caring protects the stranger opening your link on a shaky train connection.

## Engineering shifts left, product shifts right

Watch great design engineers work on the web right now, people like [Emil Kowalski](https://emilkowal.ski/), [Rauno Freiberg](https://rauno.me/), or [Jakub Krehel](https://jakub.kr/writing/details-that-make-interfaces-feel-better), and you see immediately that engineering constraints don't show up at the end of a handoff ticket anymore. They sit right inside sketch one.

On Home Port, byte ceilings (`450 KB` max mobile first view; ours lands around `218 KB`), `3:1` contrast checks across all four lighting hours, and `44 px` tap minimums were locked before the room even had furniture. On `CrashCalc`, a plain early question (`"ensure we aren't loading a 1gb app on peoples devices"`) became an automated `30 MB` install ceiling. On `Messina OS`, `tv-guard` fails compilation the second modern CSS tries to sneak onto a frozen 2020 `Chromium 85` TV browser.

At the exact same time, product management is shifting right toward production. Instead of tossing static PRDs over a wall and hoping for the best, PMs who like getting their hands dirty can spin up live TypeScript spikes to test assumptions in days, write the `CLAUDE.md` and `SKILL.md` harnesses that encode product taste, and curate the trace-grounded evaluation suites that gate releases.

## Going one layer deeper where your knowledge is shallow

On Home Port, Blender Python geometry, Cycles bounce budgets, `OKLab` color math, and `AVIF` compression internals were all **shallow knowledge** for me. Coding agents carried the heavy implementation mechanics.

Shallow can't mean incurious, though. As [Felix Rieseberg wrote](https://felixrieseberg.com/how-to-be-more-ambitious-a-tutorial-for-humans-with-models/) about pairing depth with range, models give you just enough vocabulary in adjacent fields to step one layer deeper and ask the right physical question. I'd never configured a ray-traced glass shader in Python, yet squinting at a flat daytime window on my phone was enough to ask `"shouldn't the glass actually reflect the environment and not you just coloring the glass? making sure"` and catch a painted shortcut. On our wedding camera, my personal depth lived in table hospitality rather than graphics programming; working with Claude Code meant never having written a `512 x 512` HALD LUT shader in WebGL2 stopped being a wall.

## Knowing when a real domain expert rules the room

Crossing lanes is the exact opposite of pretending you know more than someone who has spent decades in the field.

In `CrashCalc`, my dad's decades of forensic crash reconstruction rule every equation. When an investigator gets cross-examined in court, nobody cares how fast my multi-agent Git worktrees ran. Frozen reference answers (`golden vectors are law`) govern every calculator, and any discrepancy goes straight to my dad's ruling, never a casual edit by an agent or by me. Knowing which lane *isn't* yours to improvise in is half of adult product judgment.

## Still the kid in the blue polo

The twelfth book on the [Home Port](https://andrewdurden.com/) shelf, *Blue Polo*, goes all the way back to taking desktop PCs apart on my parents' carpet as a kid and working shifts on the Best Buy floor during undergrad at FIU here in Miami.

Standing in a retail aisle on a Saturday afternoon cures you of jargon fast. Nobody walks in excited about HDMI bandwidth specs or acronyms; they just want movie night with their kids to work on the first try. Years later, working alongside engineers, applied scientists, and designers far more talented than me, the job underneath the title hasn't really changed: stay endlessly curious about how the machinery works under the hood, drop your ego when a draft misses, and turn all that complexity into something warm and obvious for the person standing in front of you.
