Chapter 15
Chapter Fifteen
MAREN
AIR AND WATER BIOLOGICAL SCREENING: ELEVATED SPORE MARKERS
STATUS: WITHIN SAFE PARAMETERS
I read it once. Then again.
Holden looks up from his tablet. Dutch’s attention shifts from the door to me. Reyes steps closer to the display.
“Spore markers?” Holden asks.
“Coral-derived compounds,” I say. “Filtered through guest systems during tours at controlled exposure levels. The reef produces them naturally.”
“Elevated where?” Reyes asks.
I open the report.
Air circulation: atrium low-level increase. Water processing: maintenance lateral east elevated trace. Guest suites: within normal. Research wing: normal. Petting tank corridor: elevated but within safe parameters.
Everything within safe parameters. The facility’s new favorite lullaby.
I pull the last month.
The markers have been rising for nineteen days. Politely. Inside thresholds.
“Why didn’t this flag?” Dutch asks.
“Because it’s within safe parameters.”
He gives me a look.
Holden’s voice is quieter. “Could elevated exposure affect staff?”
The old data sits in my head. Controlled exposure: perceptual softening, vivid colors, emotional openness. Unfiltered exposure: hallucination, dysregulation, radical vulnerability. Human protocols assume controlled exposure. Creature data is newer, uglier.
The coral spore neural acceleration data is three weeks old. Three weeks in a reef that no longer respects my calendar.
I open the compound breakdown.
“Current levels are below human-effect thresholds in all occupied areas,” I say. “But we increase filtration anyway. Atrium, petting tank corridor, east maintenance, sub bay. Full sweep. Air and water. No guest messaging.”
“How sure are you?” Dutch asks the question without changing tone.
I look at the spore chart.
Nineteen days rising. Still safe. Still below threshold. Still based on thresholds written before the reef began making new arguments.
“Very,” I say.
Dutch nods.
Reyes sends the filtration orders to engineering. Holden’s notes grow longer.
By 0630, the facility has entered a state of gorgeous fraud.
Hospitality messages update in gentle language.
Staff move with calm faces and altered routes.
The atrium spore filtration increases by fifteen percent with no change to scent balance because someone in branding once had an opinion and apparently even crisis has to smell correct.
The petting tank dome remains opaque. The submersible bay is quiet, all vessels nested and ready, the emergency pod still tucked behind its partial concealment like a secret I no longer enjoy having.
I spend the next hour in motion. Operations, containment, research wing, sub bay feed, back to containment. Every step has a reason. Every order has a sequence. Fear is easiest when it has to keep up with me.
At 0740, Holden follows me into the documentation station off operations. He waits until I stop moving long enough to become reachable. Crafty bastard.
“I need the official account for the board,” he says.
“It’s in the shared file.”
“The shared file says localized boundary failure in east infrastructure with no pressure compromise.”
“That’s accurate.”
“It doesn’t mention organic residue.”
“It’s in the attachment.”
“Attachment three.”
“Are we ranking attachments now?”
He looks tired. Emotionally tired in a way that makes him look less polished and more like the man who once fell asleep on my floor with a journal article open on his chest.
“Attachment three is where people put things they don’t want read first,” he says.
“Attachment three is where supporting material belongs.”
“Does unknown chitinous residue supporting a boundary failure feel like supporting material to you?”
I close the file I’m editing. “Would you like to draft it?”
“No.”
“Then stop criticizing the architecture of my incident report.”
“The architecture is load-bearing.”
I hate him.
For one second, the briefing room, the breach, the wet little horror in the walls, all of it thins enough for old shorthand to show through.
“Don’t romance the attachment hierarchy,” I say.
There we are. A lab at two in the morning. Him telling me I’d buried the most interesting result in figure seven because I was afraid the first six figures would hate me. Me throwing a pen at him. Him catching it. Both of us laughing too hard because exhaustion had made us stupid and close.
Want moves through me, messy and rude.
Not only want for his body, though my body isn’t innocent of that want. Want for a version of us that didn’t end with me standing alone in the rain holding my mother’s voice like a knife I had swallowed.
It lasts less than a second. I kill it with professionalism.
“Attachment one can include residue summary,” I say.
“Thank you,” he says.
I move the residue summary into attachment one. “Anything else, Dr. Armitage?”
“No, Dr. Vale.”
I leave before the air starts making decisions.
By 0900, the enhanced protocols have become routine enough to scare me.
The danger of competence is anything can become a process if the right people are tired enough and good enough at their jobs.
Staff stop flinching at access changes. Guests stop asking why tours remain delayed because Lina has them convinced the reef is being optimized for a once-in-a-lifetime viewing window.
Tom serves coffee, spritzes, and no crimes against nomenclature.
Dutch escorts Dane Whitcomb away from the petting tank corridor twice without visibly moving him.
Reyes sends node temperature projections every twenty minutes, each one slightly less polite than the last.
I trust all of them.
At 1017, the filtration upgrade completes.
Spores remain elevated but trending downward in occupied zones.
I open the behavioral modeling suite in the research wing.
The lab is quiet now. The whiteboard waits with yesterday’s equations under new emergency notes. My coffee has gone from bad to historical. DEVIATION remains open on the second monitor, the word sitting there like a small, honest animal.
I load the updated data and start with the smaller fauna.
Hallucigenia variant H-3 has been classified as low containment risk since introduction. Primarily benthic. Slow in open water. Substrate fidelity high. Environmental hazard low. Behavioral adaptability limited.
All based on models from introduction data, fossil inference, and the first two years of observation.
I add the new data.
The model rejects it. A poor fit score and an advisory to review inputs. The system assumes the data is wrong before it assumes the model is insufficient, which is rude of it and also extremely human.
I review the inputs. They’re correct. I run the model again.
Poor fit. Again.
This is the moment good science is supposed to feel clean. Hypothesis meets data. Data doesn’t fit. Adjust hypothesis. Don’t protect the old model out of vanity. Don’t make reality flatter your prediction. Follow the evidence.
I know this. I believe this. I’ve built my entire life on this.
So I adjust the parameters.
Substrate fidelity: reduced. Compression behavior: expanded. Exploratory range: increased. Response to EM irregularity: added. Potential spore-linked neural complexity variable: provisional.
The fit improves. Not enough.
I adjust again.
Locomotion flexibility. Environmental learning. Sensory response to vibration. Attraction or avoidance around sonic deterrent pulse irregularity.
The model accepts the new data.
I stare at the new confidence interval.
Wider now. So much wider.
The creature can do more because the model now allows it to have done more. That’s valid. That’s how revision works. That’s how science survives contact with reality.
It’s also how a person teaches herself to stop being surprised.
I save the updated model.
The system asks if I want to apply revised parameters to current containment risk assessment.
My cursor hovers.
If I apply them, H-3 moves from negligible to moderate. The system will recommend more restrictions, possibly guest containment, maybe preliminary evacuation staging.
The facility will narrow. People will know.
I think of three men refusing to leave the room after I showed them teeth. Then I think of the guests asleep or drinking coffee or complaining softly about delayed experiences in a place I told them was safe.
I click APPLY.
The system processes. New alerts bloom across the screen.
PETTING TANK SUPPORT INFRASTRUCTURE: RISK ELEVATED
EAST SERVICE GAP NETWORK: RISK ELEVATED
GUEST-ADJACENT SERVICE PANELS: REVIEW REQUIRED
SUBMERSIBLE BAY SECONDARY ACCESS: REVIEW RECOMMENDED
For one moment, I feel relief.
Then the final advisory appears.
RECOMMENDED ACTION: CONTINUE ENHANCED MONITORING. STATUS CONTROLLED.
The model has widened enough to fit the threat and still call it controlled.
I followed every number. I rejected no data.
I revised the model against my own vanity and I did it well.
Doing it well is the problem, because a worse scientist would have produced a worse lie and I produced a valid one.
CONTROLLED, says the system I just taught to say it.
I didn’t romance the model. I let the model romance me back, in good faith, with a clean confidence interval, and called the surrender rigor.
On the second monitor, DEVIATION watches me. On the first, the revised model sits clean and valid and more dangerous because it now sounds like the truth.
I save the model.