Human Performance
AI steepens superlinear performance curve — commodity cognition eliminated, returns concentrate at problem-architecture.
Signals
DX: 92.6% adoption, bimodal outcomes. Doubled incidents at some orgs, halved at others.
AI as equalizer is wrong. Organizational multiplier — existing quality amplified.
Fields Medal proofs verified by AI in two weeks.
Phase transition in cognitive output definition. Bottleneck is problem architecture, not recall.
Graham: Superlinear returns via compounding + thresholds.
AI steepens curve by removing commodity layer. Learning/curiosity/framing = only defensible surface.
Graham: Morale is structural foundation. 4-5hr ceiling on demanding work. Physical state affects ideation.
If morale is infrastructure, protecting it is as urgent as tactical execution.
Block 40% layoff, framed as AI transition. Charter: Overton window shift.
Normative drift. Selection now for adaptive capacity — what compounds with AI.
Control Surfaces
| Lever | Status | Change | Evidence |
|---|---|---|---|
| Problem framing | Critical | Now prerequisite | Fowler + IEEE |
| Morale maintenance | Degraded if neglected | Structural | Graham |
| Output concentration | Increasing premium | AI removes mid-tier | Superlinear + AI |
Watchlist
- ConfirmationDX bimodal distribution replicated
- InvalidationAI raises floor more than ceiling
- ObservableFormal verification in legal/medical/peer review
Falsifiers
- AI produces broadly distributed gains
- Formal verification doesn't generalize
- Morale claim confounded by selection
Key Unknowns
- 27% code distribution by org profile
- Superlinear advantage half-life under AI
- 4-5hr ceiling interaction with augmentation
Noise Filter
- Nootropic pouches— No structural signal
- Qwen3.5 on 2x3090— Misfiled
Get The Signal daily
Cross-domain structural analysis, delivered every morning.