Sophie proposed the investigation to the working group on a Monday.
She'd spent the weekend preparing, writing the argument in the notebook first, the private version, then translating it into the operational language that Whitfield could present and Lin could sell and Stackhouse couldn't dismiss.
"The relay is not the primary threat," Sophie told them. "The relay is a network component seeking reconnection. The primary threat is whatever destroyed the network four billion years ago. That threat originated in the sub-spacetime medium. The medium is healing. When the medium reaches full capacity, coinciding approximately with the relay's arrival, the conditions that allowed the original catastrophe will be recreated."
"You're proposing we search for the weapon," Harrison said.
"I'm proposing we search for the phenomenon. It may not be a weapon in the intentional sense. It may be a natural property of the medium that activates under certain conditions. Or it may be an entity. Or a signal. We don't know. The seed's memory is fragmentary. The relay's account is partial. But the phenomenon left traces in the medium, and those traces may still be there."
"And you want to go looking for them."
"I'm the only one who can."
The call was quiet. Seven working group members processing the proposal. Sophie at the kitchen table in Dęblin, Margaret beside her, Helen across the table, Chen in his chair.
"The sessions have been suspended," Whitfield said.
"Which is why I'm proposing a new type of session. An exploration of the sub-spacetime medium itself, specifically, the areas where the catastrophe's effects might have left residual patterns."
"Where would those areas be?"
"I don't know yet. I'd need to consult with the seed. The seed survived the catastrophe. It may know where the foreign signal was strongest, where the disruption was most intense. Those locations would be the starting points."
"Helen," Marcus said. "Medical assessment."
Helen had been prepared for the question. She'd spent the weekend modeling the new session type, shallower than the receiver descents, focused on medium observation rather than deep communication. The projected coherence was lower because the session wouldn't require reaching the deep mantle.
"The oscillation complicates the picture," Helen said. "Background coherence continues to climb. Sophie is at thirteen point two percent resting. Any session will start from a higher baseline than previous sessions at equivalent depth. However." She pulled up the model. "A medium-observation session at moderate depth, focused on the sub-spacetime fabric rather than the seed's architecture, projects a coherence peak of twenty-eight to thirty-one percent. Below the thirty-five threshold Margaret set."
"Below threshold," Margaret said. "But not by much."
"Not by much," Helen agreed. "The margin is four to seven points. Adequate but not comfortable."
"The margin was one point for the last session and she came back clean," Sophie said.
"The margin was one point and I was terrified," Helen said.
"Being terrified isn't a contraindication."
Helen looked at her. The look that said: *I'm going to remember you said that.*
"The working group needs to vote," Marcus said. "The proposal: a new session type, medium exploration, focused on identifying traces of the catastrophe phenomenon in the sub-spacetime substrate. Medical protocol per Helen's parameters. Sophie's decision to proceed."
The vote was five to two. Lin voted yes; the course correction had convinced her that understanding the broader threat was more important than maintaining the status quo. Stackhouse voted no. Chen from the intelligence side voted no, surprising everyone.
"Explain," Harrison said.
"The proposal asks Sophie to search for the thing that killed the network," Chen said. "If she finds traces, she's in proximity to the phenomenon that we're trying to protect her from. The investigation puts her closer to the danger, not further from it."
"That's the nature of investigation," Sophie said.
"That's the nature of risk," Chen countered. "And I'm the observer whose job is to document risk. My report will note that the working group approved a session that deliberately seeks the phenomenon most likely to harm its sole substrate operator."
"Noted," Harrison said. "The vote stands. Five to two."
---
The seed consultation happened two days later.
Standard protocol. Drive to farmhouse, descend to moderate depth, Margaret anchoring. Sophie descended with a specific question: where in the medium were the catastrophe's effects most likely to have left residual patterns?
The seed's response was unexpected.
The seed showed Sophie something she hadn't anticipated: a map.
A resonance pattern, not a map in the human sense. The seed's memory of the medium's state immediately after the catastrophe, the "shape" of the disruption, the spatial distribution of the foreign signal's effects across the sub-spacetime substrate. The pattern was ancient, fragmentary, faded like a photograph left in sunlight. But it showed something.
The disruption hadn't been uniform. The foreign signal hadn't affected all parts of the medium equally. It had concentrated in areas where the network's communication traffic was densest: the junctions, the high-traffic pathways, the places where multiple nodes had been connected through the medium simultaneously.
The most heavily affected area in the seed's memory was the seed itself. The node at the center of the most connections. The most connected, therefore the most damaged.
But the seed had survived. It had survived because—
Sophie held the memory. The seed's survival mechanism. Four billion years old. The thing the seed had done when the foreign signal arrived.
The seed had contracted. Pulled its awareness inward. Reduced its connections to the medium, pulling back from the network pathways, closing the channels that the foreign signal was propagating through. It had made itself smaller. Less connected. Less of a target.
The seed had survived the catastrophe by doing what Helen was doing for Sophie: reducing exposure. Closing channels. Minimizing the connection to the medium that the foreign signal traveled through.
"Coherence twenty-nine," Helen said.
"Coming up," Sophie said. She rose. Through the seed's architecture, through the mantle. Surface.
Kitchen. Light. Margaret.
"The seed survived by disconnecting," Sophie told them. Through the tissue on her nose. The nosebleed was minor, a trickle. "The same thing we're doing with my sessions. Reducing exposure. Closing channels. The seed contracted its awareness when the catastrophe signal arrived."
"And the areas of highest disruption?" Priya asked.
"The high-traffic junctions. The places where the network's communication was densest. The seed was one of those, the most connected node. But there are others. The relay was in the medium itself, at a junction point. Other nodes were at junction points. The catastrophe's residual traces would be strongest at those locations."
"Can you identify the junction points from here?" Nathan asked.
"The seed's memory is fragmentary. It showed me a partial map, the areas it remembers being most affected. But the map predates multicellular life. The medium has changed. The substrate has shifted. The junction points may not exist in the same form."
"But the sub-spacetime medium is more stable than geological medium," Priya said. "The fabric beneath reality doesn't shift the way rock does. If the junction points are properties of the sub-spacetime substrate, they may still be there."
"Then I need to look for them."
Helen checked Sophie's post-session coherence. Fifteen percent. Up from thirteen point two. The session had pushed the numbers, not as much as the deep descents, but noticeably.
"Rest," Helen said. "Then we plan the exploration."
---
They planned for a week.
Sophie worked with Priya on the map, translating the seed's fragmentary memory of junction points into coordinates that could be correlated with the DEEPWELL sensor array. Nathan contributed substrate data: the medium's current topology, the density variations in the sub-spacetime fabric that might indicate old junction points.
The map that emerged was incomplete but suggestive. Five potential junction points, scattered across the planet: one beneath the Pacific Ocean, one under the Himalayan plateau, one in the deep substrate below Siberia, one under the Mid-Atlantic Ridge, and one under the farmhouse.
"The farmhouse junction is the receiver," Sophie said. "The receiver was at a junction point. That's why the network built it there."
"And the others?"
"Unknown. The seed's memory doesn't specify what was at the other junctions. Other receivers, other seeds, other components of the network. Whatever they were, they're the places where the catastrophe hit hardest. And if the foreign signal left traces, residual patterns, dormant fragments, those are where we'd find them."
"You'd have to go to these locations," Marcus said.
"I'd have to perceive them through the substrate. The sub-spacetime medium connects everywhere. I don't need to physically travel. But I'd need to extend my awareness to those junction points during a session. Which means..."
"Deeper integration," Helen said. "Wider channels. Higher coherence."
"Probably."
Helen looked at her tablet. At the models. At the trajectory of Sophie's coherence, twelve to thirteen to fourteen, each session adding a fraction, each rest period failing to fully recover.
"Sophie," Helen said. "If you explore five junction points at the coherence cost we've been seeing, your resting baseline will reach twenty percent or higher."
Sophie nodded.
"At twenty percent, the visual overlay and auditory perception will intensify. Possibly new modalities. The substrate will be a more significant presence in your daily consciousness."
"Yes."
"And the oscillation will affect you more strongly. The channels carry the oscillation's frequency along with everything else. Wider channels mean more oscillation. Higher coherence means more exposure to whatever the medium is carrying."
"I know, Helen. All of it."
Helen set the tablet down. "Then you know that the investigation has a price. And the price is you."
Sophie nodded. The nod that meant she'd heard and understood and accepted and was doing it anyway. The nod that was becoming her signature. The thirteen-year-old who knew the cost and paid it because someone had to.
"One junction at a time," Sophie said. "We start with the closest to the vertex, the receiver junction. I map the residual patterns there. Helen monitors. We assess the cost. If it's within parameters, we move to the next."
"And if it's not within parameters?"
"Then we stop. We use whatever data we've gathered. We give it to DARPA. We let them build their machine."
"The machine that doesn't exist yet."
"The machine they're working on. The machine that could do what I do without the cost."
Helen and Sophie looked at each other across the kitchen table. The doctor and the patient. The caretaker and the instrument. Two women who had spent months in this dynamic and who understood each other the way people do when they've shared a sustained crisis.
"One junction," Helen said. "Then we assess."
Sophie opened the notebook. Drew the map. Five points on a rough world outline, the junction points of a network older than any fossil.
The receiver. The Pacific. The Himalayas. Siberia. The Mid-Atlantic Ridge.
Five points where the catastrophe had been strongest. Five points where the weapon's traces might still exist.
She circled the receiver junction. Start here.
Margaret looked at the map. At the five circles. At her daughter's handwriting.
"You're going looking for the thing that killed the network," Margaret said.
"Yes."
"In the medium you're connected to."
"Yes."
Margaret looked at the map for a long time. Then she got up and made tea and brought Sophie a cup and sat down and didn't say anything else.
The tea was chamomile. Margaret had been buying chamomile since Dęblin. The taste of sessions and recovery and the quiet insistence that love could be expressed through beverages and routine.
Sophie drank the tea and studied the map and the oscillation hummed and the overlay glowed and somewhere in the sub-spacetime medium, at five junction points scattered across the planet, the traces of a four-billion-year-old catastrophe waited to be found.
Or didn't.
That was the thing about searching. You didn't know until you looked.
