Abstract

When a warning fails, is the failure one of information or of belief? Two classic works on surprise attack place the weight differently. Roberta Wohlstetter (1962) placed the failure before Pearl Harbor in telling signals from the noise around them; Richard Betts (1982) found that surprise attacks have usually succeeded because warnings were not believed, or not acted on in time. This article reads Aeschylus' Agamemnon, in Gilbert Murray's verse translation, as a text that stages both failures and a third that the two positions tend to fold into one another. The play opens with a true signal, the beacon relay from Troy, that the elders of Argos doubt as a woman's credulity; it turns on a true warning, Cassandra's, that the elders partly understand and do not act on; and it closes the warning scene with the elders debating, too late, who should do what. Read this way, a warning passes through three stations: receipt, credence and standing to act. Two decision records published by Meridian, a platform of the RISCON family, serve as a modern counterpart. Both are modern summaries written by AI about fictional polities, not archival documents, and both name Cassandra as their own interpretive frame: a truce liaison council choosing how to relay notices while its crisis line is broken, and an interstellar communications council deciding what to send across an eighteen-year delay to a community whose government may have changed. The thirteen AI-generated judgments on each write careful procedures for receipt and for who may act, and deliberately decline to judge whether a warning is true. The article asks what that division of labour does to the Wohlstetter and Betts debate. It draws no conclusion about which option either council should choose.