[HM]Harlan does his job
They said the new system would “streamline feedback.”
No one asked whose feedback, or why it needed streamlining, or what had been wrong with the old way of simply talking to people.
But the system arrived anyway – a glossy dashboard with cheerful colors and a hunger for numbers. It needed survey results like a data center needed power – relentless, insatiable.
So, the company hired Collectors. The Collectors were told to ask Content (formerly known as customers) to fill out surveys. Not honest surveys, of course, but correct surveys: surveys that reflected well on the Collector, which reflected well on the Manager, which reflected well on the Regional Director, which reflected well on the AI that monitored all of them.
“Please remember,” the Collector would say, “you’re grading me, not the company. And turn your phone sideways to properly view the survey.”
And the Content, bewildered, would tilt their smart phones sideways like lemmings.
But AI wanted more.
So, the company appointed Watchers to watch the Collectors.
And then, Over-Watchers to watch the Watchers.
And then, Meta-Watchers, whose job it was to ensure the Over-Watchers were properly observing the Watchers observing the Collectors collecting the data that fed the AI.
Each layer produced reports.
Each report generated metrics.
Each metric required more data.
The building hummed with the soft, anxious glow of screens.
And in the middle of it sat Harlan, a Collector of no particular distinction. He had once enjoyed talking to people – the small, human exchanges that made the day bearable. But now every conversation felt like the prelude to a plea:
“Please fill out the survey.”
“Please rotate your phone.”
“Please don’t mention the company policies.”
“Please don’t mention the wait time.”
“Please don’t mention the survey itself”
One morning, a Watcher approached Harlan with an electronic notebook and a strained smile.
“Harlan,” she said, “your survey-completion metrics are trending downward. The AI has flagged you for motivational recalibration.
Harlan looked at her, at the electronic notebook, at the blinking camera in the ceiling tile.
And something in him – something small, stubborn, and very old – simply refused to move.
“I prefer not to,” he said.
The Watcher blinked. “Not to what?”
“Not to participate,” Harlan said. “Not to cajole. Not to beg for stars. Not to turn human beings into data points for a machine that doesn’t know their names.”
The Watcher stared at him, horrified. This was not in the training manual. So, the Watcher pressed a button.
Within minutes, an Over-Watcher arrived. Then a Meta-Watcher. Then a Senior Meta-Watcher with a badge that said Human Interface Optimization Lead.
“Harlan,” the Watcher began.
But the Senior Meta-Watcher stepped forward and cut her off. “Excuse me. Harlan, the AI is concerned about your attitude.”
“I prefer not to,” Harlan repeated.
The Watchers conferred. They checked their dashboards. They consulted the AI, which whirred and blinked, and produced a recommendation: “Increase Observations.”
A new Watcher was assigned to watch Harlan. And another to watch the Watcher.
But Harlan simply sat at his station, calm and unbothered.
He preferred not to.
And the Watchers, for all their layers and metrics and dashboards, had no protocol for a human being who simply declined to be optimized.