"Then tell it accurately."
Isaac looked through the glass wall at employees crossing the corridor.
"Accurate is uglier."
"Usually."
He agreed to testify.
The first reconstruction session focused on the design review.
Investigators obtained the forty-seven-slide presentation, the final risk register, the qualification package, and three versions of Isaac's report. They also obtained email traffic among Cameron, Giselle, program management, legal, and engineering leadership.
The early emails were ordinary.
Cameron asked for clarification of assumptions. Program managers questioned schedule impact. Finance noted potential contract penalties. Engineers debated whether the coupled condition justified new testing.
Nothing in those messages proved corruption.
That mattered.
Isaac read them and remembered how easily memory, sharpened by injury, could turn every past disagreement into evidence of conspiracy.
Some people had genuinely doubted his model.
They had been wrong.
Being wrong was not misconduct.
The line crossed later.
One message from Cameron to Giselle, sent after Isaac refused to sign the certification, described the risk in different terms.
Merritt is unlikely to support the current approval basis. If his position becomes part of the customer record, we will lose the milestone and potentially the launch window. Legal should advise on managing the personnel and documentation exposure.
The phrase managing the personnel exposure appeared twice more in subsequent messages.
A second message from Giselle asked whether the analysis could be categorized as exploratory rather than design-limiting because it had not completed peer validation.
Cameron responded that doing so would keep it outside the primary risk register.
Then came the security review.
The same file-transfer practice used against Isaac had been recorded dozens of times across the engineering group with no disciplinary action. Investigators found no evidence that Isaac sent controlled data outside approved systems or to unauthorized recipients.
What they found instead was an internal security inquiry opened four days after his refusal to sign.
The triggering complaint had come from Cameron's program office.
That timing did not automatically make the complaint false.
The selective enforcement did.
"Did you know other engineers used the same transfer method?" the review chair asked Cameron during a recorded interview.
"I was not responsible for monitoring every engineer's file behavior."
"But you reported Mr. Merritt's."
"It came to my attention."
"How?"
Cameron paused.
"I don't recall."
The investigators showed him his own email requesting a security review.
"Does this refresh your memory?"
Another pause.
"I was concerned about compliance."
"Why only his compliance?"
Cameron asked to consult counsel.
The interview paused.
Giselle's role proved more complicated.
Her executive account had opened Isaac's unredacted report, but she said a staff member routinely downloaded documents for her review queue and that access metadata did not prove she personally read every file. That statement was plausible.
Other evidence narrowed the question.
A handwritten note from the original board packet contained three words in Giselle's handwriting beside the schedule impact estimate: FOUR MONTHS IMPOSSIBLE.
Another note beside the technical recommendation read: NEED SECOND OPINION.
Those notes did not prove she understood the full failure mechanism.
They did prove she knew the recommendation existed.
During her interview, she resisted the suggestion that she had ordered Isaac removed.
"I did not instruct anyone to terminate Mr. Merritt."
The review chair asked, "Did you know his refusal to certify created a program obstacle?"
"Yes."
"Did you know legal and Cameron were discussing personnel options?"
Giselle looked toward counsel.