7.1Three stations, not two
A warning must be received, believed and acted on by someone entitled to act. Wohlstetter's note places the decisive failure at the first station; Betts's at the second and third. The play shows each failing on its own terms: the beacon is received at once and doubted because of its reporter; Cassandra's warning is understood in part and disbelieved; the death cry is believed by most of the elders, though one still asks for proof, and no one has the standing to act.
What the scheme adds is the separation of the last two stations. Betts's note joins belief and action; the elders' debate shows them coming apart. That matters for remedies. A remedy for disbelief (a critical evaluator, priority for the bad prognosis, a second channel) does little for a group that largely believes and cannot act; a remedy for standing (named roles with authority to stop, assigned in advance) does nothing for a group that will not believe.
7.2Wohlstetter and Betts after the play
The play is a poor test of the view that warning fails for want of information, since it concerns a house whose crimes are known. What it can show is the limit of noise as an explanation: where the warning fails most sharply, the receivers say the signal is clear. Wohlstetter's note does include acting on the signal, and the play is consistent with that part; it is the placement of the main obstacle in discrimination that the play does not bear out.
Its support for Betts's placement is also limited. The obstacles to belief in the play are a curse and a prejudice about women. Betts's note names political and organisational obstacles; the prejudice is social and in that sense political, but no institution could remove a curse. Read strictly, the play shows that credence can fail for reasons independent of the message, which is consistent with Betts's note and says nothing about which institutional causes operate in a real case.
7.3What AI-written judgments do with a warning
The records add something the play cannot: a plan, made before the event, for how a warning will be handled. Across both records the AI-written judgments do three things consistently. They make receipt traceable: original text, timestamps, receipt confirmation, each correction linked to the message it corrects and checked for arrival. They assign standing in advance: named roles, on fixed cycles, with authority to stop or reopen, and in the second record with standing placed at the receiving end. And they decline to rule on credence: receipt confirmation is distinguished from consent to content, and acknowledgment, truthfulness and sovereignty are kept apart.
The judgments treat credence as the business of the parties, not of the office that carries the message. That is a principled division: an office that decided which warnings to believe could suppress true ones, which is exactly the worry of the minority judgments in the first record.
The division also has a cost. Betts's note places the main obstacle to warning at credence and action. The records build strongly for action, in the form of standing, and leave credence untouched. If the parties at the receiving end disbelieve a true warning, the procedures will show afterwards that the warning arrived, in its original words, with its corrections and its dissent attached. They will not have made the warning more likely to be believed. In the terms of the play, the records guarantee that Cassandra's words reach the elders and are kept; they do nothing about the curse.
The Cassandra frame asks how unheeded warnings and false-alarm corrections can fall under the same accountability. The judgments answer by making both traceable through the same record: they answer the question of accountability (who can be shown, afterwards, to have received what), not the question of heed. That is narrower than the frame invites, and it may be the only answer a relay can honestly give.
7.4What the evidence can and cannot carry
The play is a single work of fiction, read in a single translation. Its warnings are true by the author's design, which removes the hardest part of the real problem: in practice, a receiver does not know whether a warning will be vindicated. A reading of the play can show what the stations of a warning are and how each can fail; it cannot show how often each fails in practice, or which matters most.
The records are synthetic, written by AI, and nothing in them happened. They show how a warning problem is posed and how thirteen AI-generated judgments plan for it, not how any council has acted. The two records were chosen because their author attached the Cassandra frame to them, which makes them apt and also unrepresentative: other records in the collection may handle warnings differently. The judgments' consistency may reflect a common origin rather than independent agreement.
The ledger can say that both records name a decider, a deadline, unknowns and conditions for reopening, as all fifty do, and that the first is among the thirteen cases where the judgments chose more than one option. Its program sees numbers, thresholds and times, not roles or events, so it cannot say whether the named roles would observe what they are assigned to observe.
7.5Alternative readings
Three other readings deserve mention. One holds that Cassandra's scene is about fate, not warning: the curse makes the outcome certain, so nothing the elders do could matter, and reading the scene for institutional lessons misreads it. Cassandra's own line, that it is all one whether she is believed, supports this. The elders' debate after the cry, which treats the outcome as still open until the door swings wide, cuts against it.
Another holds that the beacon episode shows a healthy demand for confirmation, not a failure of credence, since the elders end by believing the Herald. The elders' own words, which place the doubt in the reporter's sex rather than in the signal, tell against it.
A third concerns the records: the refusal to judge credence may be less a design choice than a habit of judgments written under a brief that offers no conclusion and asks them to separate setting, inference, hypothesis and value. Two records cannot rule that out. It would not change what the procedures do, only how much credit the judgments deserve for it.