SATURDAY, MAR 21, 2026
He stayed in the warehouse until three in the morning, after Sasha had dropped Felicia at a location she didn't specify and Marco had gone home and it was just Dan and the warehouse and the second Roxxon crate sitting on the workbench.
He split the cargo into two lots the way he had the first time — different buyers, different timing, different channels to reduce the probability of any single transaction's buyer connecting to the other — and set the first transaction in motion and left the second for Monday. When the first transaction cleared at two forty-seven AM he transferred Felicia's cut from the airport job — $190,000 in untraceable digital currency — and noted it in the operational log. The Byzantine reliquary payment had gone out four days earlier, the same system, the same forty-eight hour window he'd promised. Both cleared without issue. The accounting was clean.
Then he sat in the mezzanine office with the camera feeds on the wall screen and thought about Felicia.
She knew too much about how he operated. He'd known this was building for weeks and had continued the partnership anyway because the operational calculus was sound — her capabilities justified the information exposure, and he had assessed, each time he ran the calculation, that the value of her as a partner exceeded the risk of what she knew.
He ran the calculation again now with the full accounting of what tonight had added: she'd moved cleanly through a JFK airside staging area under time pressure with Port Authority in range and had done it faster than his plan had allowed for, which meant the plan's buffer time had been conservative, which was good. She had improvised nothing and executed everything. Sasha had said it was the smoothest job she'd driven, which meant Sasha was calibrating her against a standard that Dan also applied, and the result was the same.
The information exposure. He catalogued it honestly. She only knew his first name. She didn't know his address. She didn't know about Columbia or Castillo or Gwen or Yara or any of the people who occupied the other half of his life. She knew the Red Hook warehouse's location, which he'd accepted when he'd first brought her to plan OP-009.
She knew Sasha and Marco by first name and could describe them. She knew his planning methodology at a level of specificity that went beyond most people he'd worked alongside — she'd been inside enough of his plans now to understand the underlying logic, the distribution framing, the preference for information over improvisation, the four-minute buffers.
She understood how he thought, which was the most intimate kind of knowledge someone could have about an operational partner and the most difficult to protect against once it had accumulated.
He ran the threat model. If she wanted to leverage what she knew against him — take it to someone who would pay for a profile on a rising independent operator, or use it to preempt a target he'd already researched — the damage was real but bounded.
The identifying information gap was significant. If she wanted to use her operational knowledge to take a job from him, the math didn't support it: one job's value against the ongoing partnership that had cleared, in five months, more than three times what either of them would have earned independently. She was not calculating incorrectly. She had demonstrated this at every step.
Threat model outcome: low probability, medium consequence. Acceptable, as it had been acceptable last time he'd run it, and the time before that.
He opened the operational notebook and wrote: "F. — information exposure reviewed. Risk: acceptable. Trust level: 72%. Trend: upward. Note: trust level has moved 22 points in five weeks. Pace is unusual. Examine whether pace reflects genuine reliability or optimism bias."
He looked at the note. He had written optimism bias with the intention of taking it seriously, which meant examining it rather than dismissing it. Optimism bias would look like: weighting evidence of reliability heavily because he wanted it to be reliable; discounting risk because the alternative was that a person he was enjoying working with was a risk he should be managing more carefully.
He went through the evidence. The Roxxon envelope. The split on OP-009. The airport execution. Every meeting on time. Every agreed protocol observed. No use of her information advantages at any point that he could identify.
That was not optimism bias. That was evidence. His trust level was at seventy-two because seventy-two was what the evidence supported. He scratched out the note about optimism bias and replaced it with: Rate of trust development reflects rate of evidence accumulation. Partnership operates at high tempo. Evidence accumulates accordingly.
He closed the notebook. He went home through the April dark, driving the armored sedan because it was late and because the threat level was elevated and because comfort was not a reason to lower your operational standards. He parked and walked the last block and let himself into the 113th Street building and went up the stairs and into his room and stood at the window for a moment looking at the garden strip below.
The March light had finally done something to the trees, a faint suggestion of color at the tips of the branches that was not yet leaves but was the intention of leaves, the first visible step in the direction of a different season. He had watched these trees for five months. They had been bare when he moved into this room and they were going to be different soon and that fact felt, standing at the window at three in the morning, like something more than a botanical observation.
He did not send a text, because it was three in the morning and because whatever he would have said wasn't a three-in-the-morning kind of thing to say. He knew what it would have said and he held it and didn't send it, and that was fine.
Some things were better for the waiting. He had not always believed this — he had been, for most of his time in this world, someone who moved quickly on complete information and waited only when the information was incomplete. This was different. The information was not incomplete. He was simply choosing the right moment, which was a different kind of waiting and one he was just beginning to understand the value of.
He slept five hours and woke at eight and made coffee and stood at his window with the cup in both hands watching the spring light come in over the garden strip and thought: this city is doing something to me and I don't entirely object to any of it.
Then he opened his academic notebook, because Castillo's seminar was at ten and he had not yet finished the reading, and the reading deserved his actual attention, and giving things your actual attention was not diminished by the fact that other things were also happening to you at the same time. That was the lesson the last five months kept teaching. He kept learning it.