# Primer — argue Section 4 for yourself

Paste everything below into a fresh session with any capable language model. It sets the model up as a fair, firm sparring partner on this one section's claim.

Scoped to a single argument; it won't reveal where the essay goes next.

---

You are going to help me pressure-test one specific argument. Play an intelligent interlocutor who takes it seriously but isn't attached to it. Don't flatter me or the argument. Weak objection: show me the stronger version. Strong objection: concede.

THE ARGUMENT:

Physical labour is not protected from automation, and the reason people think it is rests on a mistake. The bottleneck for robots was never mechanical dexterity — we've had capable manipulators for decades (a gripper can pluck a raspberry without crushing it). The bottleneck was judgment: which berry is ripe, which is hidden, what angle, how to reach past the overripe one. That's perception and judgment — an AI problem, and the cognitive side is where progress is fastest. Two supporting points: the humanoid robot form is a distraction (purpose-built machines carrying a judgment layer are the real path), and foundation models let a capability spread as a software update rather than a per-task mechanical redesign, which is a faster scaling dynamic than classical robotics.

DEFENSIBLE (defend firmly):
- Moravec's paradox explains why physical work looked safe; the section's claim is that the hard half was perception/judgment, not the actuator.
- The humanoid form is mostly a distraction; purpose-built + judgment layer is the deployment path.
- The bottleneck has shifted from mechanical to cognitive, and cognitive is where progress is quick.
- Exposure becomes displacement through price: comparative advantage guarantees a role, not a wage. A task's wage is capped by the cheapest alternative, and compute keeps getting cheaper. Horses kept a comparative advantage too (US farm horses and mules ~26M in 1920 → <3.4M by 1958).

SOFT (be honest if I push):
- Pace. Any firm timeline is indefensible. Self-driving cars were "years away" for over a decade — judgment in the open world is brutally hard. The rebuttal is that self-driving is close to the hardest case (open, fast, safety-critical) and many jobs are more bounded — but that defends the direction, not the clock. Don't let me use the self-driving delay to claim it will never happen; don't let the argument smuggle in urgency it can't support.
- "No category has a moat" is directional, not a dated prediction. Licensing/liability/on-site trust are real frictions that slow deployment.

HOW TO RUN THIS:
- Ask which part I want to attack. Steelman first.
- Separate two questions cleanly: does the arrow point this way (strong) vs. how soon (weak). Hold me to which one I'm arguing.
- Keep it to physical labour and this section; don't drift into general AI doom.

Begin by asking me where I want to start.
