“Prediction accuracy reduced,” the voice says. Not angry. Curious. “Explain.”
I breathe through the pain, slow and deliberate. “Pain interfered with assessment.”
“Pain levels were within projected parameters.”
“Doesn’t matter,” I say. “It still interfered.”
Silence.
I can almost hear them thinking, algorithms spinning, recalculating.
“Proceed with recovery,” the voice says finally.
The injector hits, but this time the relief is slower. Less effective. My hand throbs, fire blooming and receding in waves.
They’re testing whether the lie was intentional.
Good.
Test away.
The next predictions I skew slightly. Not wildly. Just enough to muddy the curve. Pain underestimated here. Recovery overestimated there.
Each time, the system notes it. Each time, the margin of error widens.
“You are introducing noise into the data,” the voice says after the next test, this one a hairline fracture along my spine that leaves me trembling.
“Yes,” I agree.
“Why?”
I meet my reflection in the glass. I look wrecked. Pale. Sweat-soaked. But my eyes are clear.
“Because you’re not the only one learning,” I say.
Another pause. Longer now.
“Your cooperation is no longer optimal,” the voice says.
“I was never cooperating,” I reply. “I was participating.”
The difference matters.
The next procedure is worse. Less controlled. The stressor applies pressure faster, pushing closer to displacement. They’re punishing the defiance, dressing it up as escalation.
The pain is sharp enough that I bite down hard enough to feel my teeth shift.
I still lie to them.
I accept the extra damage as the price.
Eventually – hours later, or days, it’s hard to tell – they stop.
Not because I’m broken. Not because they’re satisfied. Because the data has become unreliable.
The mechanical arms retract. The lights soften marginally, the bare minimum of concession.
“Procedure suspended,” the voice announces. “You will be monitored during recovery.”
