1 min read
1 min01 · 1 MIN READ
The cost nobody estimates
Every technology you add has a running cost that never appears in the estimate: someone has to learn it, operate it, patch it and be on call for it at two in the morning.
That budget is finite. Spend it where the problem is genuinely novel, and take the dull, well-understood option everywhere else.
“ Every company gets about three innovation tokens. You can spend them however you want, but the supply is fixed. ”
02 · 1 MIN READ
What we default to
- PostgreSQL until the data genuinely is not relational
- One deployable service until a team boundary demands two
- The framework's own patterns before a bespoke abstraction
- Boring, observable infrastructure over anything self-healing
None of this is a rule. It is a default, and a default only has to be right more often than it is wrong.
03 · 1 MIN READ
What it bought us
- median time from idea to production
- 3 days
- databases to back up
- 1
- bespoke queue implementations
- 0
