Goodhart’s Law and the "Anti-Metric": Why Velocity is a Trap

The Trap of "Lines of Code" and Goodhart’s Law

Goodhart’s Law states: “When a measure becomes a target, it ceases to be a good measure.”

We saw this in the “McKinsey Developer Productivity” debate. If you measure developers by “Lines of Code” or “Number of PRs,” you don’t get productivity — you get inflation.

• Engineers will split one PR into five small ones to game the metric.
• They will write verbose code instead of elegant, concise functions.

The "Vanity Metrics" of DORA

Even DORA metrics (Deployment Frequency, etc.), while useful for DevOps health, fail to measure organizational risk.

A team can deploy 10 times a day (high velocity) while building a brittle architecture that only one person understands (high risk).

The Metrics That Actually Matter: "Capability & Resilience"

Instead of measuring output (which can be gamed), sophisticated leaders measure capacity and risk.

1. The Bus Factor (Resilience)

Identify parts of your codebase that are owned by only one person. If your Lead Architect gets hit by a bus (or poached by OpenAI), does your roadmap die?

• NotchUp Insight: Visualizes knowledge silos based on code commit history, not org charts.

2. The Skill Liquidity Ratio (Capacity)

What percentage of your team can be deployed to a critical initiative tomorrow?
If you have 100 engineers but only 5 can work on your GenAI initiative, your liquidity ratio is 5%.
This is a strategic bottleneck.

3. Cognitive Load (Efficiency)

Are your senior engineers coding, or are they reviewing PRs and sitting in meetings?

• NotchUp Insight: Tracks context switching by analyzing the spread of tasks across domains.
  High context switching = low deep work = future burnout.

Conclusion

Stop measuring how fast the hamster wheel is spinning. Start measuring if you have the right hamsters running in the right direction.

Measure Resilience, Not Just Velocity. Audit your team with NotchUp.