Architecture begins where reversal becomes expensive
Treat architectural significance as a gradient: spend design effort where reversal, evidence, and accumulated coupling make change costly.
Field notes
Concise observations on architecture, reliability, delivery, and the organisations around the software.
Treat architectural significance as a gradient: spend design effort where reversal, evidence, and accumulated coupling make change costly.
The useful output of architecture work is not a diagram. It is a set of decisions a team can act on, challenge, test, and revisit.
A struggling programme usually has capable people trapped inside a broken decision and delivery system. Start there.
Reliable systems come from architecture, operations, incentives, product decisions, and incident learning working as one system.