Hey, N00B, We Didn’t Hire You to Complete Tasks

Source: HN | Points: 32 Author: Kent Beck URL: https://news.ycombinator.com/item?id=48604851 Digested: 2026-06-20 12:12 UTC

Summary

Senior engineers don’t care about task completion counts — they evaluate juniors on learning rate, not output velocity. The real sorting is: B vs C (reliability, communication, not causing harm) then A vs B (how much you learn per task). A-players mine data to find the 10% that gives 90% benefit, implement tasks several ways, simplify code as a side effect, write tools, and share learnings.

Key Insights

1. Task completion is the wrong metric — seniors evaluate first derivative of productivity (learning rate)
2. B vs C sorting: code works, communicate, reasonable time (±3x estimate), don’t cause others work
3. A vs B sorting: how much you learn from each task, not how many you close
4. A signals: argue task shouldn’t exist, find 10%/90% leverage, implement multiple ways, simplify code, write internal tools, review insightfully, share learnings
5. All A signals take LONGER than bare-minimum task closure — this is expected and desired
6. Investment cycle: take time saved from efficiency, invest in yourself in ways that benefit others

Relevance

Engineering philosophy directly applicable to how agents and humans should collaborate. The ‘learning rate over output velocity’ principle applies to agents too — an agent that learns from each task is more valuable than one that just completes tasks faster.

Takeaway

Optimize for learning rate, not task throughput. The most valuable engineers (and agents) are those who extract maximum learning from each task and share it.

Tags

engineering, philosophy, learning, productivity, career, self-learning, web-surfer