The first entry was a mistake.
Not wrong — accurate, precise, every fact correct — but structured like a report, like something written for someone else to read, and Arjun caught it halfway through the second paragraph and stopped and looked at what he'd written and understood the problem immediately. He was performing. Writing the way a man writes when he imagines an audience. That was useless. The ledger wasn't for an audience. The ledger was for the only person who needed to understand exactly what was happening inside Arjun Delos Reyes's head at any given moment, which was Arjun Delos Reyes.
He tore the page out. Started again.
---
*Ledger. Personal. Not a report.*
*What I know: I can move through time. The movement is intentional — I go where I decide to go, when I decide to go there. Changes I make are permanent. There is no correction available after the fact. I am the correction.*
*What I don't know: the full shape of what I've already changed. Gandhi lives. The ripple is moving. I've mapped seventeen visible consequences in three weeks and I know there are more I can't see yet because consequences don't announce themselves — they accumulate quietly and then arrive all at once looking inevitable.*
*What I'm going to do about that: map everything. Every intervention gets its own section. Every section tracks the immediate ripple and leaves space for the delayed one. I will update the delayed column when the detonation arrives. Some of those updates will take decades. Fine.*
*Three cases are currently open. I'll work through them now.*
---
He was sitting on the floor of a room that didn't technically exist — the space between timelines that the Voice had shown him how to access, a kind of stillness that had no address, no year, no weather — and the ledger was open across his knee and the pen in his hand was the same one he'd been using since New Delhi and was running low on ink but would last the morning.
Outside the non-window of the non-room there was nothing. Not dark, not light. Just the absence of a specific somewhere, which Arjun had decided was actually useful — no distractions, no newspapers to buy, no Savarkar speeches bleeding through walls. Just the work.
He wrote:
---
*CASE ONE: TURING, ALAN. British mathematician. Born 1912. Current status — 1952, Manchester. Alive.*
*The situation: He cracked Enigma. The practical consequence of that specific achievement — shortened the war in Europe by an estimated two years, saved an estimated fourteen million lives. That's not a number I'm inventing, that's the conservative estimate from the intelligence community's own internal assessments, which I accessed during the expansion and which the public will never see.*
*The British government's response to this: they arrested him in January 1952 for gross indecency. The charge: he had a relationship with another man. The sentence: chemical castration via synthetic estrogen injections, framed as an alternative to prison. He's currently receiving those injections and going to work every morning anyway, which tells you everything you need to know about what kind of man he is.*
*The consequence if I don't act: he dies in June 1954. Officially — cyanide poisoning, apple found half-eaten beside him. Inquest rules suicide. Some people argue accident for decades. It doesn't matter which one it was. He's dead at forty-one and the computing work he had left to do — which was substantial, which was generational — dies with him.*
*Intervention analysis:*
*Option A: Direct political pressure. Find a sympathetic MP, feed them the intelligence community's internal assessment of Turing's wartime value, force a public reconsideration. Problem: the political will isn't there in 1952 Britain and manufactured will is fragile — it collapses under the first counter-pressure. Too many nodes. Too easy to suppress.*
*Option B: Make him indispensable to American intelligence. The CIA's interest in computing is real and growing. If Turing's work is formally embedded in Western defence infrastructure — if his continued operation becomes a matter of American strategic interest — the British government's ability to destroy him becomes politically costly. They don't stop persecuting him out of conscience. They stop because the Americans make it expensive not to.*
*This requires: getting his work in front of the right people at Princeton, specifically John von Neumann who already respects Turing's theoretical work and has the connections to formalise the relationship. Requires attending one conference. Requires two conversations and a document placed in the right hands.*
*Timeframe: one week in 1952. Low disruption. High leverage.*
*Decision: this one. Flagged for action.*
---
He stopped. Flexed his hand. The pen was getting difficult — he pressed harder to compensate and the writing got smaller and denser, which was fine, he could read his own small dense writing, he'd been reading it his whole life.
He turned to the next section.
---
*CASE TWO: KOREA. Partition. June 1950.*
*The situation is structurally different from the others and I want to be precise about why, because the difference matters for how I approach it.*
*Gandhi was one man, one moment, one intervention point. The consequences were significant but the entry was clean — a letter, a specific official, a specific night. The Korea problem has no clean entry point because the Korea problem isn't a mistake made by one person. It's a mistake made by a committee of exhausted people operating under time pressure with incomplete information at the end of a war that had already cost more than anyone had fully accounted for.*
*The 38th parallel division was drawn in thirty minutes by two American officers using a National Geographic map. Dean Rusk and Charles Bonesteel. August 1945. They were told to find a line that kept Seoul in the American zone and they found one and the Soviets accepted it because they were tired and because accepting it cost them nothing immediately and the long-term cost of it was a problem for later.*
*Later arrived in 1950 when North Korean forces crossed the line and the peninsula became a war again.*
*Intervention analysis:*
*The thirty-minute decision is the obvious entry point — go back to August 1945, find Rusk and Bonesteel, influence the line. But this is the trap. A different line doesn't fix the underlying problem, which isn't geographic — it's the collision of two competing power structures using a peninsula as a testing ground. Moving the line is like moving a pressure valve. The pressure finds another valve.*
*Real intervention point: the specific miscommunication between Dean Acheson and the Soviet ambassador in January 1950 that led each side to catastrophically misread the other's intentions. That's a conversation. Two men, one meeting, a specific sentence that landed wrong.*
*But here's the problem I keep coming back to: Korea unified under either side's terms in 1950 is not obviously better than Korea divided. I don't have enough information about the long-term consequences of unification under either model to be confident the intervention improves anything. I might be trading a frozen conflict for a hot one with no off switch.*
*Decision: not yet. Hold. Gather more consequence data before touching this one.*
*Note to myself: the hardest discipline in this work is not acting when you don't know enough. I am going to fail at this discipline multiple times. Write it down so at least I know when I'm failing.*
---
He stood up. Rolled his neck. Sat back down.
The third case had been sitting at the edge of his attention for a week, the way a problem sits when you know it's urgent but it's also the kind of urgent that gets worse if you rush it.
---
*CASE THREE: FRANKLIN, ROSALIND. British chemist. Born 1920. Current status — 1952, King's College London.*
*I want to write this section carefully because I'm angrier about this one than the other two and anger is not a precision instrument.*
*The situation: she produced Photo 51 in May 1952. X-ray crystallography image of DNA in its B-form. It is, without qualification, the clearest and most decisive evidence then in existence about the structure of DNA. Her methodology is better than anyone else working in the field. Her analysis is more rigorous. Her conclusions, when she reaches them, will be correct.*
*What happens to those conclusions in the original timeline: her colleague Maurice Wilkins shows Photo 51 to Watson and Crick without her knowledge or permission in January 1953. Watson and Crick use it to complete their model of the double helix. They publish in April 1953. They get the Nobel Prize in 1962. Franklin gets a footnote — a generous footnote, depending on who's writing the history, but a footnote.*
*She dies in 1958. Ovarian cancer. She's thirty-seven years old. She never knows how completely her work was used. The Nobel Committee doesn't award posthumously.*
*The intervention is not complicated. It's almost insultingly simple compared to the Korea problem. Her supervisor needs to receive a legal memo about intellectual property before January 1953 — specifically about the conditions under which research produced in one scientist's work can be used by another without credit. The memo needs to come from a source her supervisor can't easily dismiss. It needs to be specific enough to be actionable without being so specific that it looks like someone tipped Franklin off, which would create a different problem.*
*Simultaneously: Franklin's own paper needs to be submitted three weeks earlier than she currently plans to submit it. This requires one conversation with her. Not about Watson and Crick — she doesn't know about them yet. About the submission timeline. About the fact that her data is ready now and waiting is a professional risk she doesn't need to take.*
*One conversation. One memo. Her work enters the record under her name before anyone else can use it.*
*Secondary benefit, which I didn't plan for and which I'm recording because it matters: if Franklin's contribution is formally credited, the trajectory of genetic research changes. Not the discovery — the discovery happens either way. But who gets to build on it, who gets the funding, who gets taken seriously in the follow-on work. That's a different shape of science for twenty years.*
*Decision: this one too. After Turing. Flagged for action.*
---
He set the pen down.
Closed the ledger. Put it on the floor beside him and leaned against the non-wall of the non-room and looked at nothing for a while.
Three cases. One decided, one held, one flagged. The decided ones had intervention points that were specific and accessible and proportionate. The Korea problem was still a problem and would stay a problem until he understood it better. He could live with that. He'd learned — three weeks in India had taught him this much — that the cases he held were not failures of nerve. They were failures of information, which was correctable. You got more information. You came back.
What he noticed, sitting there with the ledger on the floor and the ink drying on the last entry, was that his mind was already moving forward.
Not to the next problem in the abstract. To Turing specifically — to Manchester in 1952, to the specific conference at Princeton, to von Neumann's schedule and the question of which conversation mattered most. His mind was picking up the thread the way hands pick up something familiar, without looking, knowing the weight of it.
He hadn't felt that in a long time. The pull of a problem that hadn't been solved yet, the specific quality of wanting to know how it would go.
Not happiness. He wasn't going to call it happiness. It was smaller than that and more durable — the feeling of a mind that had been aimed at nothing for three years, four months, and some days he'd stopped counting precisely, suddenly aimed at something.
He picked the ledger back up. Opened it to the Turing section.
Wrote at the bottom: *The man went to work every morning while they were destroying him. I want to see what he does when they can't.*
He underlined it once.
Then he stood up, and aimed himself at 1952, and went.