Complex systems are mostly decisions. Which data is authoritative. Where a boundary sits. What happens when a dependency fails. What the system must never do. The code follows from those decisions, and its quality is capped by theirs.
This is why we keep teams small and senior. The people who make architectural decisions also write the code that tests them, and they stay on the project long enough to live with the consequences.
What that looks like in practice
- Engineers talk directly to the people who own the problem, not through layers of account management.
- Architecture is written down early and revised openly, as short decision records rather than a large upfront document.
- Design and engineering work in the same review, because an interface often exposes an architectural flaw first.
What it is not
It is not a promise of speed through heroics. Small senior teams are faster mostly because they avoid work: features nobody needs, abstractions nobody uses, migrations that did not have to happen.
It is also not a rejection of scale. When a system needs a larger team to operate it, part of our job is to make it legible enough for that team to own it confidently.