[Midtown hotel ballroom — May 2, 2013, 6:52 PM]
Maren arrived twenty-two minutes late.
This was not an accident. The boutique-bar cocktail reception was the kind of event that rewarded late arrivals: the early arrivals had already assembled their networking clusters, and walking in at the twenty-two-minute mark meant the room was warmed but not yet committed, and the clusters were still permeable. She moved through the space with the ease of someone who had calculated this before walking in.
Don had been there since 6:30, which was when the reception formally started and when he preferred to arrive at events — early enough to choose his own geometry rather than have it assigned. He was at the edge of the room near the drink service, talking to a partner from a mid-size litigation firm he had known since the first bar association CLE, about a property dispute that had nothing to do with Don's practice and that he was engaged with at twenty percent of his attention while the other eighty percent tracked the room.
He registered Maren entering at exactly the moment she registered him.
The specific quality of her registering him was: she was pleased. Not in a performed way — in the way that a person is pleased when they have expected a thing and the thing has appeared.
She moved toward a cluster of private equity people at the far end of the room and began talking. Don watched his immediate conversation without appearing to watch the far end of the room, which was the skill that the last two years of Detection had made automatic.
She was making predictions.
This was what she had been doing since the New Year. He had heard three versions of it in different contexts — a conference, a casual industry breakfast, one casual corridor encounter — and the pattern had been consistent enough to constitute a behavior rather than isolated events. She made small market calls and pop-culture calls, not her serious investment-thesis-level work but smaller-scale predictions, the kind you make in social settings where you want to be known as sharp. Some of them were specific in a way that exceeded what good analysis alone could produce. Some of them were off on timing in a way that was consistent across the sample.
Degraded foreknowledge. Not information access. Not sources. The specific signature of someone working from a model of how the future was going to go that had been accurate once and was becoming less accurate over time.
It was his signature. He had been living inside it for two years, had been tracking the decline through the Error-Bar Log and the Divergence Diagnostic. He knew what it looked like from the inside.
Now he was watching it from the outside.
She had drifted closer over the next forty minutes. Not toward Don specifically — she moved through the room the way she always moved through rooms, with purpose and without appearing purposeful. But she was at the drinks table at the same time he was refilling, and she was in the adjacent cluster when he moved toward it, and eventually she was simply next to him with a glass of something white that she had been carrying since she arrived.
"I see you watching me," she said.
"I'm standing in the same room."
"You're paying specific attention." She said it without any edge. The observation was accurate and she had decided to name it because naming it was more interesting than not naming it. "I've been predicting things. Were they accurate?"
"Most of them."
"Most of them." She drank. "You noticed the ones that were off."
Detection ran the sentence: she was telling him she knew about the off-ones. Not asking. Telling.
"The tech acquisition timeline you mentioned to Reyes," Don said. "You said it was closing in three weeks. It closed in six."
"Yes." She said it with the specific acknowledgment of someone who has made an error and has catalogued it carefully. "I had the right shape and the wrong timing." She looked at her glass. "It happens."
The Library, running on 2.5 LP of the 3 LP I had authorized for this conversation: #maren-vance — prediction pattern: degraded foreknowledge signature confirmed. Three-data-point pattern: Redstone Holdings (Ch.86 clue #1), Darby dissolution timeline (Ch.96/122), tech acquisition timing. Pattern: correct outcome, wrong by 40-60% on duration. Consistent with foreknowledge degradation rate comparable to Don's own: ~3% decline per month from initial accuracy.
The pattern was not lucky speculation. Lucky speculation had random timing errors. This had systematic timing errors — consistently running shorter than reality, the specific signature of someone whose model was calibrated against a future that was shifting.
Maren's version of the Divergence Diagnostic was running at her own expense, and she didn't call it that.
"It happens with good sources too," Don said.
She looked at him.
The look had the quality it always had: direct, interested, deciding how much to show. "Sometimes," she said. "Sources have their own timing issues." A pause. "Some information is structural rather than source-dependent. Structural information has a different kind of timing problem."
Structural information. Not: I have good sources and they sometimes get the date wrong. Structural: I have a model of how things will go, and the model is correct about what will happen, and incorrect about when.
He was looking at himself.
He picked up a canapé from the passing tray and ate it. It was good — the same reception caterer who had done the antitrust CLE in 2011, the first time he had been in a room with Victor Reyes. Two years ago, approximately. The canapés had been good then too.
"Victor Reyes is looking for you," she said.
"I know."
"You're avoiding him."
"I'm managing the timing."
She laughed — genuinely, the brief laugh of someone who has heard a sentence they found accurately observed. "I like that. Managing the timing." She turned the glass. "He wants your name on the antitrust coalition letter."
"He does."
"Are you going to give it to him."
"Conditionally. If the Columbia competition analysis confirms the theory, yes."
She looked at Don for a long moment. "You know," she said, "you are one of the only people in this room who is consistently right about things without being consistently confident about being right."
"Being confident about being right is how you end up wrong."
"Yes." She turned the glass again. "I know something about that."
The sentence had a weight to it that the other sentences had not had. Detection caught it: not the self-deception micro-fracture, which had been present and widening since November. Something else. A moment of genuine recognition — of something she was not hiding from Don but was carrying with the specific quality of a fact she was still working out what to do with.
Then Reyes appeared at Don's left shoulder.
"Klein. I need five minutes."
"Give me two." Don turned to Maren. "Excuse me."
She made a small gesture — not dismissal, the gesture of someone releasing you to an obligation while reserving the right to continue the conversation later. It was, in its way, the most Maren-like gesture she had made all evening.
Reyes pulled Don to the room's edge and outlined the coalition status: four boutique names, the Columbia analysis in progress, a projected filing date of summer 2013. He needed Don's conditional signature on the draft letter by Friday.
"If Elena Fuentes's analysis confirms the theory of harm holds under current doctrine," Don said.
"She's been in contact with my office. She thinks it holds."
"Then you'll have my signature."
Reyes shook his hand with the grip of a man who had been building toward a specific thing for eighteen months and had just received the piece he needed.
Don went back to where Maren had been standing. She was not there. She was across the room, making another small prediction to a pair of analysts, about a film release, about when it would open — and the date she said was six weeks off from the actual release date in Don's memory.
Six weeks. The same systematic offset as the tech acquisition. The same systematic offset as the Redstone call, adjusted for duration.
The Library returned: #maren-vance — Janus hypothesis probability: 82%. Based on three confirmed degradation-signature events. Alternative hypothesis (superior sources + systematic source-timing error): 8%. Insufficient data to confirm Janus hypothesis fully. Arc 4 confirmation pending.
Eighty-two percent.
He watched her work the room with the ease of someone who had been doing this for a long time — making small calls, letting some of them land and some of them drift, reading the room's reaction. She was, in her way, running a version of Detection at the social level. Not supernaturally. But very well.
He put down his drink.
He did not confront the hypothesis. He carried it, the way he had been carrying it since January, the weight of a thing that was almost certainly true and was not yet confirmable.
She looked across the room at him.
She saw that he had set down his drink.
She knew exactly what had happened — he could read the knowing in the quality of her look, the specific micro-expression of someone who has made a move and seen the response.
She smiled — briefly, not performing it, the genuine smile of someone who is enjoying a game played at the right level.
She went back to her conversation.
Don picked up his coat from the coat check, said the necessary goodbyes, and walked out into the May evening.
The hypothesis was at eighty-two percent. Arc 4 would begin inside thirty days.
The Janus question had a clock.
✦ ━━━━━━━━━━━━━━━━━━━━━━━━ ✦
Read A LOT more chapters for free at unwrittenrealm.com
✦ ━━━━━━━━━━━━━━━━━━━━━━━━ ✦
TL;DR — Patreon has the chapters ahead.
Silver / $6 / 15 or more chap ahead
Gold / $9 / 20 or more chap ahead
Platinum / $15 / 25 or more chap ahead, no weekly wait
That's it. patreon.com/fanficwriter1
If reviews aren't your thing, no pressure — they do help though.