Several autonomous tractors intermittently stopped during planting.
No error persisted.
Logs suggested sensor disagreement.
Software teams chased timing bugs.
The engineer asked:
“What do operators say?”
An engineer shrugged.
“There are no operators.”
She stared.
“There is always someone standing near machine eventually.”
They interviewed farm technicians.
One said failures happened most often after washing.
That clue mattered.
Water entered a connector behind rear fender.
Not enough to fail immediately.
Enough to create resistance drift hours later.
Software responded correctly to bad input.
The problem was not software.
The engineer stood in meeting.
“Stop debugging code until connector redesign.”
A younger developer objected.
“Logs show state-machine transition.”
“Yes.”
“Because machine believes two sensors disagree.”
“Why?”
“We think race condition.”
“You think.”
“I know connector is wet.”
The room quieted.
A field test confirmed.
Afterward, one young engineer approached.
“I thought you were old-school.”
She smiled.
“I am.”
“Then how—”
“Electricity has been annoying farmers longer than you’ve been alive.”
He laughed.
F7’s lesson had survived into machines with satellite guidance.
The exact technology changed.
The logic did not.
Identify what you actually have.
Observe before replacing.
Question shared assumptions.
Respect field evidence.
At sixty-three, she stepped down from corporate director role.
She did not call it retirement.
“I am reducing meetings.”
That was accurate.
She kept consulting.
Taught two evenings a month.
Worked farm.
Served on heritage board.
The first meeting of the board nearly made her quit.
They spent forty minutes debating the shade of cream paint for a restored 1948 combine.
She raised hand.
“Does it run?”
The curator looked offended.
“Yes.”
“Then why are we spending forty minutes?”
“Accuracy.”
She nodded.
“Fair.”
Then added:
“But if roof leaks on storage building while we argue paint, we are failing.”
The roof leaked.
She got funding moved.
She was not opposed to preservation.
She opposed confusing appearance with purpose.
That became recurring argument.
One board member wanted F7 fully restored cosmetically before permanent display.
The proposal included stripping red paint and revealing larger areas of gold undercoat.
She read it.
“No.”
The member said:
“Visitors would better understand.”
“No.”
“We can document layers.”
“You would destroy original evidence to make evidence easier to see.”
“We would professionally conserve.”
“You would create version of tractor that never existed in its working history.”
The member looked frustrated.
“It would be more visually compelling.”
“That is not same as more true.”
Proposal failed.
Years later, conservators thanked her.
Original layered paint became important to study factory process.
At sixty-seven, she discovered something that changed one small piece of F7 history again.
A box arrived from the family of a deceased Prairie King engineer.
They had found old photographs and correspondence.
One envelope was addressed to her grandfather.
Never mailed.
Inside, a handwritten note:
Field Unit 7 ignition fault confirmed at Davenport demonstration. Your reroute held through final test. If dealers sell unit, keep revised route. Factory drawings will not carry test harness.
The engineer sat in her kitchen reading.
Her grandfather had known.
The warning existed.
Why did wire return to wrong path decades later?
She and a historian reconstructed.