engineering management

Engineering management after the cost of code collapsed

3 months ago

The cost of producing plausible code has collapsed, but broad claims about dramatic speedups, obsolete practices, or halving headcount are unproven.
Management practices should be audited based on their underlying assumptions: those resting on code cost need review; those on human coordination, trust, or verification remain valid.
Evidence of speedups is smaller than noise; honest instrumentation is needed to measure actual acceleration, not felt speed.
Old metrics like velocity and PR counts become misleading because volume is cheap; focus on business outcomes and treat code volume as a cost.
Plumbing and mechanical verification (machine-checkable correctness) are fast, but semantic verification requiring human judgment and ownership remains slow and does not compress.
The junior engineer pipeline is broken because AI absorbs tasks that built judgment; no proven solution exists yet.
Old rules like 'director should not code' need recalibration; 'shield the team' still valid but requires deliberate context selection; 'consensus before commit' applies only to irreversible decisions.
Headcount requests should be scrutinized because output per person changed, but coordination costs remain.
The job is sorting which practices rest on typing costs vs. human nature, not blindly keeping or discarding the old playbook.
Management's information-routing function is at risk from AI, but judgment, ownership, and signing off on decisions remain human; orgs will shrink to those who own accountability.

Hasty Briefsbeta

engineering management

Why I Still Write Code as an Engineering Manager

Engineering management after the cost of code collapsed