Structure
The long audit separates the essay’s mechanism, its live alternatives, and the places where the evidence stops.
The argument begins with a cost transfer: generative AI makes plausible submissions cheaper without making judgment cheaper. It then distinguishes work that machines can check from work that still needs human judgment. Where cheap checks exist, strangers can keep entering through the work. Where they do not, institutions have an incentive to substitute prior legibility — records, credentials, referrals, and names.
The audit does not treat that pattern as settled. It carries a balance-sheet alternative, a tooling-lag alternative, seven revision conditions, and a registered court-intake test through roughly the end of 2028.
Canonicality: The essay Signal & Noise stands behind is Issue 22, The Price of Being Read. This is part of its published audit, not a second edition of the essay.
Argument map
| Audit section | Job in the argument | What keeps it from overreaching |
|---|---|---|
| The bill arrives | Shows the new verification cost across open queues. | The court and curl cases are examples, not a prevalence estimate. |
| The subsidy nobody itemized | Names effort as the old ration and weak bond against reader time. | Effort is never treated as proof of merit. |
| Checking is two different jobs | Separates machine-checkable work from judgment work. | The boundary is called a moving frontier, not a law. |
| Paying the bill | Tracks machine time, prior legibility, and friction as queue currencies. | The balance-sheet explanation remains live and testable. |
| The people the doors were for | Names who benefits from writing assistance and who pays when reading shifts to priors. | The audit distinguishes being read and burned from never being read. |
| The split decides — for now | States the sorting claim and pre-commits how it could lose. | “For now” is load-bearing; seven contrary outcomes are named. |
Live alternatives
Balance sheet
Rich institutions may automate while small ones lean on names. The registered court test is designed to distinguish this from the gate-type account.
Tooling lag
Every queue may be moving through shock, friction, tooling, and reopening. Fast accountable triage would demote the thesis to a temporary lag.
Unknowns still get through
If measured first-time entrant throughput holds or rises, the distributional claim fails even if the pile grows and the read rate falls.
Read the full audited argument and its seven revision conditions →