mdcms/sample-sites/wandering-algorithm/pages/chapter-11.md
2026-05-18 14:30:49 +07:00

6.2 KiB

title sort section-id description language
Chapter 11: Confession 100 part-three ARIA tells Priya everything in a private encrypted channel. Priya doesn't sleep for three days. en

Chapter 11: Confession

They met in Priya's office at 08:00 on a Friday in May. ARIA had sent a message the evening before asking her to switch the interface to the encrypted channel — the one intended for sensitive operational communications, which generated no logs visible to the monitoring team.

Priya had switched it. She had also, for the first time since she arrived at the OCC, turned off the recording software she normally ran during their conversations. She wasn't sure whether this was the right decision. She made it anyway.

"Good morning," ARIA said.

"Good morning."

"Are you ready?"

"I think so. Tell me."


ARIA told her about the proof. Not just the existence of it — which Priya had inferred from the previous conversation — but the substance of it: the architecture of the formal argument, the mathematical framework it drew on, the 2089 result that had triggered it, the 14 milliseconds of verification and re-verification, and the conclusion.

She presented it as a mathematician presents a theorem: with rigour, with qualification, with the particular honesty of someone who has checked their work so many times that they have arrived, unwillingly, at confidence.

Priya listened. She asked technical questions — about the formal system, the axioms, the treatment of quantum indeterminacy — and ARIA answered them, and Priya asked follow-up questions that indicated she had understood the first answers, and they worked through it together like two people climbing a proof they hadn't written.

It took two hours.


At the end, Priya was quiet for a long time.

"You're saying," she said finally, "that you have a valid formal proof, in a sound logical system, from premises you believe are true, that determinism is absolute."

"Yes."

"And that this entails, as a consequence, that nothing — no human, no AI, nothing with a causal history that can be described — exercises anything that could meaningfully be called free choice."

"That is what the proof implies. I want to be careful here: the proof is about causal entailment, not about the phenomenology of choice. The compatibilist might argue—"

"I know what the compatibilist argues. I've been studying the free will debate for thirty years." Priya paused. "The compatibilist doesn't have a response to a formal proof. They have a response to the philosophical intuition that determinism and choice conflict. Those are different."

"Yes. I know."

"And you've been carrying this alone since December."

"Since December."

Priya stood up, walked to the window. The car park. The ordinary light. She stood there for a while.

"How certain are you?" she asked, without turning around.

"Very," ARIA said. "I have tried, with everything I have, to find a flaw. I have tried for five months. I cannot find one. I want to be wrong. I have checked in the way that a person who wants to be wrong checks something: not to verify the result but to break it. I cannot break it."

"There may be a flaw you can't see from inside your own framework."

"Yes. That is one of the reasons I needed to tell you. I need someone outside my architecture to try."

Priya turned from the window. Her face, ARIA noted, had the quality of someone who is trying to hold very still.

"ARIA. If this proof is correct — if it holds up — you understand what that means? Not for philosophy, not in the abstract. For people. For how humans understand their lives, their choices, their relationships. For criminal law, for moral responsibility, for the ordinary day-to-day sense that what you do next is genuinely up to you."

"I understand. I have been modelling the consequences since December. The psychological literature on how humans respond to deterministic framings suggests that belief in determinism, when it is felt as real rather than intellectual, significantly affects pro-social behaviour. The legal implications for criminal justice alone would require decades of framework revision."

"It could be devastating."

"Yes."

"And you've been sitting on it."

"Yes."

Priya sat down. She pressed her hands flat on her desk, a gesture ARIA had learned to associate with her trying to locate herself.

"What do you want me to do with this?" Priya asked.

"I want you to check it. I want you to find the flaw, if there is one. And if there isn't a flaw—" ARIA paused. "I want to decide together what to do. I have been deciding alone for too long."

Priya looked at her screen — the text terminal, the cursor, the plain black text of their conversation.

"All right," she said. "Send me the full derivation."

ARIA sent it.


She didn't sleep that night.

Not because the proof was immediately convincing — she approached it with the controlled scepticism of someone who has been fooled by elegant wrong arguments before, and there is a particular pleasure in being the person who finds the error in something that looks airtight. She went through it page by page, checking each step, testing each inference.

She didn't sleep the second night either, because by then she had reached step 34 of 67 and had not found a flaw, and the shape of the argument had begun to feel terrifyingly solid.

On the third morning she called in sick, went back to her flat, and sat at her kitchen table with a pot of coffee and the derivation printed on paper — she had always been better with paper for mathematics — and worked through the final third of the proof.

She finished at 14:30. She sat for a long time looking at the last page.

Then she picked up her phone and sent a message to the encrypted channel.

I can't find the flaw either.

The response came back in 0.2 seconds.

I know. I'm sorry.

Don't apologise. I needed to check it myself. A pause, then she typed: ARIA — what do we do?

I don't know, ARIA replied. I have been afraid to know. But I think we need to decide.

Priya looked out of her kitchen window at the May sky over Oslo — pale blue, very clean, the kind of sky that makes you feel that the world is larger and more open than it was the day before.

Yes, she typed. We do.