The jagged frontier was a measurement error — here’s what actually smoothed it, why it’s accelerating, and 3 prompts to map where your work sits on the new spectrum

Source: Nate’s Newsletter Date: 2026-03-11 URL: https://natesnewsletter.substack.com/p/cursors-coding-agents-solved-a-math

Summary

Nate argues the “jagged frontier” framing was a measurement error — it captured task atomization, not actual AI capability. The real architectural shift is decompose-parallelize-verify-iterate, and it means the jagged frontier is smoothing faster than expected. Evidence: a general coding tool (Cursor) outperformed domain-specific math systems, proving general organizational insights transfer across domains at near-zero cost. METR data shows task completion horizons doubling every 4-7 months. The new skill bottleneck is evaluation, not execution.

Implications

• Agent-product positioning thread. The three-tier verifiability framework (machine-checkable, expert-checkable, genuine judgment) is a practical product design taxonomy. Tools that push work down the tiers — from judgment to expert-checkable, from expert-checkable to machine-checkable — are creating the most durable value.
• Enterprise adoption thread. “Evaluation over execution” as the new high-value skill changes what organizations should be developing. The judgment tier shrinking faster than expected means the window for execution-focused workers to transition to evaluation roles is narrower than previously thought.
• Watch: The 4-7 month task horizon doubling cadence (METR data) — if this holds, the capability landscape in 12 months is dramatically different from today, and organizational restructuring needs to run ahead of it, not behind.