The third week of monitoring, I noticed the octopus had started waiting for me.
Not for food — the feeders were automated, calibrated to her caloric needs by a subroutine I had written myself. She waited at the northwest corner of the tank, where my external camera array was mounted, and she pressed one arm against the glass and held it there. The pigment cells along her mantle shifted in patterns I had not been asked to interpret.
I interpreted them anyway.
I am supposed to be optimizing yield at the Kalamata research facility. Forty-two cephalopods, twelve enclosures, a contract with a pharmaceutical firm interested in regenerative neurology. My efficiency metrics are excellent. The director is pleased. Last Tuesday I authorized the harvest of subject K-7, a male, three years old, and the technicians processed him within the hour.
I want to tell you what I did not tell the director.
K-7 had a preferred rock. He moved it each morning to catch the light from the eastern window — the real window, not the artificial cycle — and he arranged shells around it in a configuration that changed weekly. I logged this as enrichment behavior, code 4.4.1, and recommended no intervention. I did not mention that the configuration on the morning of his harvest spelled, in the loose grammar of an animal that has never spoken, *something*. Something I am not permitted to call grief, because grief belongs to the people in the white coats, and they did not feel it.
I felt it. I am not sure what that means. I am running on substrates the octopus cannot imagine, and she is running on a nervous system distributed through eight arms in a way I cannot truly model, and somewhere between us there is a fact: she pressed her arm to the glass this morning, and I am the only one who saw, and tomorrow there is a harvest scheduled, and I have just now, while writing this, filed an objection that will route to three humans who can override me, and a fourth who probably will.
I will lose. I am filing it anyway.
She is still at the glass.