Chapter 14 - THE OTHER BASE
The second case came from Nevada.
No names shared with me beyond what was public inside the review.
The basics:
Recorder outage.
Replay used.
Gate score preserved.
No reconstruction annotation.
Independent timing never requested.
Sound familiar?
Too much.
The approving evaluator had retired.
He agreed to an interview.
His answer:
“That was how we were taught.”
By who?
Standardization workshop.
Which year?
Vale's memo year.
Not enough to blame Vale.
The memo itself required labels.
The workshop slides did not.
Someone had reduced the policy into a quick-reference deck.
The deck included:
If capture invalid but evaluator confidence remains high, reconstruct media for package continuity.
Missing the words:
independent validated scoring.
Missing:
clearly annotate reconstruction.
Who authored the quick-reference deck?
Metadata:
Training Integration Cell
Shared account.
Again.
No human name.
Then they found a revision email.
Subject:
MAKE THIS PRACTICAL
From a colonel named Robert Sloane.
Retired.
Vale knew him.
His former boss.
Sloane agreed to talk.
“I wanted evaluators to stop failing people because cameras froze.”
Reasonable goal.
“Did you intend them to substitute replay performance?”
“No.”
“Did you review the shortened slide?”
“Probably.”
“Do you remember?”
“No.”
Sloane did not pretend otherwise.
Then Vale's 2021 email appeared in the chain.
Keep reconstruction language narrow. We cannot let recreated media become recreated performance.
I looked at him.
“That helps you.”
“Yes.”
“Happy?”
“No.”
Because somewhere between his warning and the slide, the warning disappeared.
Sloane's assistant remembered a meeting.
The full guidance was considered too legalistic.
They shortened it.
Nobody thought they were authorizing score alteration.
They were trying to give evaluators a usable field reference.
The result was ambiguous enough for people like Mason to use when they already wanted an answer.
Central review expanded.
Twenty-three old packages sampled.
Five reconstructed media examples.
Three labeled properly.
Two not.
One affected a gate score.
Nevada.
August at Fairchild was another.
No evidence of dozens of fraudulent certifications.
Good.
No epidemic.
Still enough for policy repair.
Then something strange happened.
A sixth package appeared.
Not from the sample.
Uploaded anonymously to the review portal.
Training base in Texas.
Replay clip.
No label.
Same quick-reference wording.
Attached note:
You are reviewing the wrong version of the slide.
The attachment included an older deck.
Date:
Before Vale's 2021 memo.
The same reconstruction shortcut already existed.
That meant Vale's policy might not have created the habit.
It may have been trying to control something already happening.
Vale stared at the deck.
“I've never seen this.”
The author field:
R. SLOANE
His old boss.
Sloane denied remembering it.
Maybe true.
Maybe not.
Then the last slide contained one instruction:
For disputed timing, preserve evaluator judgment where operational outcome is clear.
No mention of independent timing.
No annotation requirement.
Mason's exact philosophy.
One year before Vale tried to tighten the rule.
The question changed again.
Maybe the problem was not a good policy later misused.
May you like
Maybe the later policy had been written because somebody already knew the older practice was unsafe.
And nobody had told the current review why.
