🗓️ 06062026 2100

STRATEGIC VS TACTICAL PROGRAMMING

Ousterhout frames two programming mindsets. Tactical: get it working now. Strategic: produce a design that works now and supports future development. The choice compounds — tactical shortcuts accumulate into systems nobody wants to touch.

Tactical programming​

  • Goal: make the feature work, ship it
  • Each shortcut is small and seemingly harmless
  • Complexity grows incrementally — no single commit ruins the codebase, but hundreds of small compromises do
  • Tactical programmers are often rewarded short-term (fast delivery) while the debt stays invisible until it isn't

The "tactical tornado"​

Ousterhout's term for the prolific developer who ships fast but leaves a trail of fragile, tangled code. Often celebrated by management, dreaded by the team that maintains the result.

Strategic programming​

  • Goal: a great design that also works
  • Treats every change as an opportunity to improve structure
  • Sweats the small stuff — zero tolerance for accumulating complexity
  • Slower initially, but faster over time as the codebase stays workable

The investment math​

  • Ousterhout estimates ~10% extra effort upfront for strategic design
  • Payoff window: 6–12 months before the investment yields net gains
  • After that, strategic codebases accelerate while tactical ones slow down
WARNING

Startups and tight deadlines push toward tactical. Facebook's early "move fast and break things" culture shipped fast but required years of cleanup. The cleanup rarely fully happens.

How to apply​

  • Not binary — you don't redesign everything, and you don't skip all design
  • When touching code: leave it slightly better than you found it
  • When building new: invest in the interface and module boundaries first
  • When under pressure: know you're taking a loan, not getting something free

References​