Once, Adrian sent an analysis at two in the morning.
Isaac called him immediately.
"Why are you awake?"
"You're calling me at two in the morning."
"I'm a mechanic."
"That's not an explanation."
"Go to bed."
"Did you look at the plot?"
"Yes."
"And?"
"Tomorrow."
Adrian later told Eleanor that Isaac's greatest contribution to his career was not the resonance model.
It was teaching him that urgency did not justify thoughtlessness.
"He works fast," Adrian said, "but he never behaves hurried."
Eleanor understood exactly what he meant.
The follow-up technical review of OBSIDIAN-7 occurred six weeks after the remediation.
The coupled harmonic remained suppressed. Pressure regulation had stabilized. Propellant consumption returned close to the revised mission baseline.
The hardware damage already accumulated could not be reversed, so the mission team adopted additional monitoring and a more conservative maneuver profile.
The satellite would not achieve every objective originally planned.
It would remain controlled.
That was success enough.
At the debrief, one executive used the phrase heroic save.
Isaac corrected him.
"Nothing heroic happened."
The executive looked surprised.
"Mr. Merritt, the vehicle was weeks from loss."
"And thirty-seven people spent three days checking my work before anybody touched it. That's not heroics. That's engineering."
"The original insight was yours."
"The original warning was mine. The fix belonged to everybody who tried to break it and couldn't."
Eleanor said nothing during the exchange.
Later, in the hallway, she told him, "You know they'll still turn you into a story."
"Then make it an annoying story."
"How?"
"Make them include the part where I was wrong twice during validation and other people caught it."
"You really do hate mythology."
"Mythology is how organizations stop learning."
That sentence reached the oversight board.
The board's reform package was less dramatic than the anomaly itself and probably more important.
Future critical programs had to preserve unresolved technical dissent in the primary risk record rather than relegating it to appendices without explicit closure.
Engineers who refused safety-related certification would trigger independent technical review before any adverse personnel action related to the same program could proceed.
Cross-discipline environmental testing had to include coupled-condition assessment when subsystem interactions could plausibly produce emergent behavior.
Contractor performance metrics were modified so schedule success could not offset unresolved safety exceptions in executive scoring.
The rules did not guarantee courage.
No rule could.
They made cowardice harder to disguise as procedure.
Isaac read the policy draft sitting at the same workbench where Eleanor had first opened the telemetry folder.
He circled three paragraphs in red.
"Too vague," he told her over the phone.
"Which?"
"Independent review. Who chooses the reviewer?"
"Program leadership with oversight concurrence."
"Then program leadership can pick someone friendly."
"What do you want?"
"Random selection from a qualified pool outside the reporting chain for certain classes of dissent."