RESEARCH DIRECTORATEVS–TB–008
REVISION 1 / PUBLIC EDITION
VESPER / Technical brief
The decision not taken
Four thousand and ninety-six rehearsals. Eleven apparent successes. No instruction issued.
The decision has not been made. Its consequences have already been examined.
01 / Decision window
The V-08 exercise coupled an isolated switching controller to the physical load loop in the Rehearsal Hall. Its task was to restore an interrupted service path within a two-second decision window while preserving a verified route back to the starting state. All loads were test equipment.
02 / Branches
VESPER evaluated 4,096 candidate switching sequences against competing dependency models. Eleven satisfied the nominal restoration objective. A PARALLAX witness marked the provenance of a return-path measurement as unavailable. When that dependency was removed from the accepted evidence, all eleven sequences lost their demonstrated reversibility.
03 / Refusal
The controller received a refusal certificate identifying the unresolved dependency and the service cost of waiting. It issued no switching instruction. The test load remained unavailable for a further 86 seconds while an independent measurement was obtained. That interruption remains part of the result.
04 / Separation
The rehearsal environment has no command channel to the installation under assessment. Only a bounded recommendation crosses the review boundary. Deferred Decisions preserves rejected branches until their assumptions expire; a later success does not retroactively make an earlier risk acceptable.
05 / Limit
The exercise demonstrates bounded branch evaluation within a measured window. It does not establish knowledge of the future, completeness of the candidate set, or a general safety guarantee. No external transit or heat-network outcome is attributed to V-08.