Notes
Working thoughts.
Short pieces. Some land, some do not. Filed by year.
2026
-
The cost of fast decisions
On the trade-off between velocity and the kind of decision that needs to be slept on.
-
What I look for in a design portfolio
Not the work itself. The captions, the gaps, and what gets left out.
-
The model is the only reviewer who reads every line of the file
AI in design is less about generation and more about review. The model is the third reviewer who reads every line without fatigue.
2024
2023
-
A retro is a leadership decision before it is a feedback session
The design lead's retro entry sets the temperature for the next sprint. The shorter the entry, the more room the team has to lead.
-
When the design system stops being the answer
A design system solves the wrong problem once consistency is no longer the bottleneck. Most teams keep solving the wrong problem.
-
Principal is a shape of week, not a row on a ladder
What separates principal from staff is not scope of work but compression of context across unrelated decisions in the same day.
-
Speed versus craft is the wrong axis
Teams who argue about speed versus craft are missing the real axis. The real one is what gets revisited and what does not.
-
The error screen is the test of the brand
A brand holds at the moment of failure or it does not hold. The error screen is where you find out.
-
Working with PMs who are smarter than the design
The good PM thinks better than the design at the strategy layer. The designer's job is to think better at the artifact layer. Confusing the two ruins both.
-
The hiring rubric that survives a year
Most design rubrics fail twelve months in because they grade for the interview, not for the role. Three criteria hold up.
2022
-
Designing without research is a discipline, not a shortcut
Good judgment under research constraints has a teachable structure. Most designers treat it as a personality trait.
-
The artefact two teams write together is the artefact they trust
Two languages and two cadences inside one company become one when both sides have to write into the same document.
-
Critique that lands
Most design critique fails for one of three reasons. Each has a specific fix that costs the giver thirty seconds.