The different plan came from the Collective.
Not in the morning. Not after deliberation. At 2:33 AM, in the space between wakefulness and sleep, in the hollow hours when the fluorescent lights hummed the same frequency they always hummed and the lab was dark except for the scanner's standby glow and Tanaka's workstation display showing the screensaver she'd programmed — a rotating model of the consumption architecture that she'd built from scan data, the three-dimensional map of Zeke's channels spinning slowly in the dark like a planet made of pathways.
*We have been thinking,* the Collective said. The voice quiet. The nighttime register — lower, less chorus, more singular, the ten thousand voices compressing into something that approached a single speaker the way a crowd's murmur approached a hum. *We have been thinking about containment.*
"Go away."
*We understand the instruction. We do not comply. The thinking is relevant. The thinking has produced a — conclusion. A possibility. A thing we did not know we knew until we thought about it long enough that the knowing surfaced from the history we carry.*
Zeke opened his eyes. The ceiling. The fluorescent tubes, off now — Tanaka had installed a dimmer three days ago, the researcher's accommodation for a patient whose sleep was already insufficient without the constant optical assault of institutional lighting. The tubes dormant. The room in the grey-blue glow of the screensaver's architectural model, the channels rotating, the pathways turning.
"Talk."
*The previous government — China, twenty-three years ago — built a containment facility for the previous eater. The facility was designed for two purposes: consumption optimization and transformation containment. The consumption optimization failed — the pipeline accelerated the timeline. But the transformation containment — the design of the facility's containment architecture — was based on research that the Chinese government conducted independently. Research into the nature of curse-eater transformations. Research that produced data.*
"Data that you have."
*Data that the previous eater's architecture absorbed during his time in the facility. The data entered his channels through the facility's monitoring equipment — the sensors, the scanners, the instruments that the Chinese researchers used to study his consumption architecture. The instruments transmitted electromagnetic information into the architecture and the architecture absorbed the information the way it absorbed everything — passively, continuously, without distinguishing between curse energy and non-curse data. We carry the data. We have carried it for twenty-three years. We did not know it was relevant until the containment question was asked by a colonel on a phone call that was not happening.*
"What does the data say?"
*The data says that curse-eater transformation is not a binary event. The previous understanding — the institutional framework, the HA's projections, the US government's casualty models — treats transformation as a switch. At 100 percent saturation, the switch flips. The eater becomes a Curse Entity. The entity is uncontrolled, destructive, apocalyptic. Binary. On or off. Human or monster.*
"And?"
*And the Chinese research found that the binary model is — incomplete. The transformation has a gradient. A spectrum. Between 95 and 100 percent, the eater exists in a transitional state — not fully human, not fully entity. The transitional state has characteristics of both. The consciousness persists. The power increases. The body changes but the change is progressive, not instantaneous. The Chinese researchers documented this gradient in the previous eater's final months. They watched the transformation happen in stages, not in a single event.*
Zeke sat up. The cot protesting — the springs complaining, the mattress redistributing his weight with the reluctant flexibility of furniture that had been designed for temporary use and that was being used permanently. The sitting changing his perspective — the ceiling giving way to the room, the horizontal glow of the screensaver replaced by the three-dimensional space of the lab, the world expanding from a tube's-eye-view to the actual geography of the place where he lived and might die.
"A gradient. Not a switch."
*Not a switch. The binary model was adopted by the HA because the HA's data comes from the aftermath of transformations — the destruction, the casualties, the entity-state that the transformed eater enters after the process completes. The aftermath is binary. The entity is an entity. But the process of becoming the entity has stages, and the stages are — controllable. Potentially. The Chinese research suggested that a curse eater in the transitional state — between 95 and 100 percent — could theoretically maintain consciousness and direct the transformation process rather than being consumed by it.*
"Theoretically."
*The previous eater did not maintain consciousness. The pipeline had eroded his capacity for resistance. Four years of accelerated consumption had degraded the neurological substrate that consciousness required to persist through the transformation. He entered the gradient without the cognitive resources to navigate it. The transformation consumed his consciousness in the first stage and the remaining stages produced an entity without a pilot. The entity — without consciousness, without direction — defaulted to its base function: the expression of ten thousand curses simultaneously. The casualty event.*
"So the transformation killed people because the eater couldn't think through it."
*Because the eater's mind was — spent. Depleted by the pipeline. The consumption volume had exceeded the neurological recovery capacity and the deficit was permanent and the permanence meant that when the transformation began, there was not enough of the eater left to steer. The entity emerged mindless. The mindlessness produced destruction. Not because transformation inherently produces destruction. Because uncontrolled transformation produces destruction. The distinction is — crucial, crucial, crucial.*
The distinction. The word that changed the model. Not a switch but a gradient. Not an uncontrollable event but a potentially controllable process. Not an automatic apocalypse but a transition that could theoretically be navigated if the person navigating it still had enough mind to navigate.
"You're saying I could transform without becoming a monster."
*We are saying the data suggests that a curse eater who reaches the transitional state with sufficient cognitive integrity could potentially direct the transformation's progression. Could choose — at each stage — what the transformation produces. Could shape the entity that emerges rather than being consumed by it. The word 'potentially' is doing — significant work. The data is from one case study. The case study was a failure. The failure does not disprove the theory — the failure demonstrates what happens when the preconditions for success are not met.*
"And the preconditions are?"
*Cognitive integrity through the transitional state. A mind that persists. A consciousness that doesn't dissolve when the saturation reaches the gradient. The Chinese research identified the cognitive threshold — the minimum neurological function required to maintain directed transformation — and the previous eater was below it. The pipeline had taken too much. The question is whether you — at a slower consumption rate, with a body that is already rejecting to limit further damage, with an architecture that has been modified for efficiency rather than volume — whether you would reach the gradient with sufficient cognitive resources to navigate it.*
"The body's rejection."
*Yes. The rejection. The body's autonomous refusal to consume is — we have reconsidered the rejection. We told you we were terrified. We were. But the reconsideration has produced a different analysis. The body is not refusing consumption because the body has reached its limit. The body is refusing consumption because the body is protecting the neurological substrate. The brain. The consciousness. The thing that would need to be intact for directed transformation. The body is — preserving the pilot. The body knows something that we did not know until the Chinese data surfaced from our memory: that the mind matters more than the saturation, and that the consumption was destroying the mind faster than it was filling the channels, and that the only way to preserve the mind was to stop the consumption before the consumption stopped the mind.*
"The body is smarter than both of us."
*The body has access to data that neither you nor we can consciously read. Biological intelligence. The accumulated wisdom of a biological system that has been monitoring its own neurological function and that has determined — through mechanisms we cannot access — that the current consumption rate is incompatible with cognitive survival through the transitional state. The rejection is not a failure. The rejection is a strategy.*
Zeke stood. The standing carrying him to the center of the lab, to the chalk circle on the floor — the boundary that Tanaka had drawn for consumption procedures, the circumference that defined the space where the eating happened and that now felt less like a medical boundary and more like a starting line.
"If the US government's concern is containment — preventing the transformation from producing mass casualties — then the answer isn't a facility in a low-population area. The answer is directed transformation."
*The answer is the possibility of directed transformation. The possibility requires research. Understanding. Development of techniques to maintain cognitive integrity through the gradient. The Chinese research provides the theoretical framework. The practical application requires — work. Study. Time. The time that the hazard exemption might provide. The resources that the counter-proposal's intelligence sharing might fund.*
"The counter-proposal isn't just a distraction. It's the mechanism."
*The counter-proposal gives the US government what it needs — intelligence, data, access — while the hazard exemption provides the time for the research that might make containment unnecessary. If you can demonstrate — if Tanaka can demonstrate, scientifically, with data — that directed transformation is possible, the containment model changes. The US government's models project an uncontrolled transformation producing mass casualties. A model that includes directed transformation projects — something different. Something that isn't necessarily safe but that is — manageable. Reducible. The kind of risk that institutions can accept because the risk has been quantified and the quantification is below the threshold that triggers the institutional reflex to contain.*
"We need Tanaka."
*We need Tanaka. We need her awake and we need her listening and we need her scientific mind processing what we have told you because the processing is the thing that turns historical data into actionable research and actionable research is the thing that turns a death sentence into a — *
"A maybe."
*A maybe. Which is better than the current nothing.*
---
Tanaka woke to the sound of her name.
"Tanaka."
The waking faster than usual. The days of insufficient sleep had produced a paradoxical adaptation — the body's alarm system tuning to higher sensitivity, the startle response shortening its latency, the wake-up time decreasing because the body couldn't afford the luxury of slow arousal in an environment where the next crisis might arrive before the arousal completed.
"What?"
"The Collective has data. Chinese research from twenty-three years ago. Absorbed through the previous eater's monitoring equipment. The research says transformation isn't binary — it's a gradient. A transitional state between 95 and 100 percent where consciousness can theoretically persist and direct the transformation process."
Tanaka blinked. Once. Twice. The blinks serving as processing cycles — each one a clock tick in the computation that was running behind her eyes, the information entering her analytical framework and being evaluated against everything she knew about curse consumption architecture and saturation thresholds and the biological mechanics of what happened when a body's curse load exceeded its design capacity.
"Say that again."
He said it again. The Collective supplemented — providing the technical details, the Chinese research parameters, the gradient model, the cognitive integrity threshold, the data points from the previous eater's final months. The information flowing from the internal voices through Zeke's mouth to Tanaka's ears to Tanaka's brain, the chain of transmission converting twenty-three-year-old classified Chinese research into a 3 AM briefing in a hospital basement.
Tanaka was at her workstation before the briefing finished. Her hands on the keyboard. The screensaver dismissed. The displays lighting up — scanner data, architectural models, saturation tracking, the full suite of analytical tools that she'd assembled over weeks of monitoring and that she was now reconfiguring for a different question.
"A gradient," she said. The run-on building. "A transitional state with preserved consciousness. The Chinese research documented stages — stages that the HA's binary model doesn't account for because the HA's data is all post-transformation, aftermath data, the destruction that the entity produced after the transition completed. If the transition has stages and the stages can be navigated and the navigation requires cognitive integrity then — then the body's rejection makes sense. The body isn't refusing consumption because the body is full. The body is refusing consumption because the body is protecting the brain. The neurological substrate. The consciousness that needs to survive the gradient."
"That's what the Collective said."
"The Collective is describing a theory. I need to test the theory." She pulled up the scanner's neurological monitoring function — a secondary capability that she'd installed three days ago for the tissue dispersion analysis and that she now repurposed with three keystrokes. "I need to measure your cognitive function relative to your saturation level. If the Chinese research is correct, there should be a correlation — a measurable relationship between the saturation level and the neurological health of the brain tissue. If the consumption is degrading cognitive function, I should be able to detect it. If the body's rejection is protecting cognitive function, I should be able to see the protection in the data."
"May I?" she asked. Already standing. The scanner in her hands. The ritual request. The permission that she always sought.
"Go ahead."
The scanner on his head. Not his chest — his head. The neurological readings populating the display with a different kind of data than the saturation measurements — brain activity, neural pathway integrity, cognitive processing metrics, the quantified health of the organ that the body was apparently protecting at the expense of everything else.
"I'm running a baseline cognitive assessment. Standard neurological metrics — processing speed, pattern recognition, working memory, executive function. These measurements will establish your current cognitive profile. If — when — the saturation increases, I can remeasure and track the degradation curve. Or the non-degradation curve, if the body's rejection is successfully protecting brain function."
The scan took twelve minutes. Tanaka silent throughout — the focused silence of a researcher conducting an assessment whose results would determine whether the theory was theory or data and whether the data supported a plan or demolished one.
"Results." She looked at the display. Then at Zeke. Then back at the display. The double-look — the verification pattern of a scientist who had seen numbers that didn't match her expectations and who was checking the source before trusting the numbers.
"Your cognitive metrics are normal."
"Normal?"
"Normal. Within standard deviation for your age and baseline profile. Processing speed, pattern recognition, working memory, executive function — all within normal parameters. There is no measurable cognitive degradation." She paused. The pause of a scientist encountering data that surprised her and whose surprise was being processed alongside the data. "At 86.91 percent saturation, with the consumption history you have, the standard model predicts — Zeke, the standard model predicts approximately 12 to 15 percent cognitive degradation. You should be measurably slower, less sharp, less capable of complex reasoning. You're not. Your cognitive function is — intact."
"The body's protecting it."
"Something is protecting it. The rejection. The modification. The seed. Some combination of factors is insulating your neurological function from the saturation's effects. I can't identify the specific mechanism without more data, but the result is — your brain is fine. Your brain is undamaged despite carrying a saturation level that should be producing measurable cognitive impairment."
"Which means—"
"Which means if the gradient theory is correct, you would enter the transitional state with cognitive resources sufficient for directed transformation. You would have — you would have a chance. Not a certainty. A chance. The previous eater didn't have this chance because the pipeline had already destroyed his cognitive capacity before the gradient began. You — your cognitive capacity is intact. The body protected it. The protection worked."
The room. The 3 AM basement lab with its chalk circle and its scanner and its researcher who hadn't slept enough and its patient who carried ten thousand voices and its data that said, for the first time in weeks, something that sounded like the beginning of hope.
Not hope. Not that word. The word was too large for the evidence. The evidence was one cognitive assessment, one theoretical framework from twenty-three-year-old Chinese research, one interpretation of the body's rejection behavior. The evidence was preliminary. Incomplete. The kind of data that a responsible scientist would label "suggestive" rather than "conclusive" and that a responsible scientist would not build a plan on without further investigation.
Tanaka was a responsible scientist.
"I need to write this up," she said. The keyboard under her fingers. The words assembling on the screen with the speed of a mind that had found a problem worth solving and that was solving it in real time, the sentences building the way her sentences always built — fast, connected, the run-on threatening to escape the structure but held in place by the discipline that science required and that Tanaka provided. "A research proposal. Directed transformation theory. The Chinese gradient data, the cognitive preservation evidence, the body's rejection-as-protection hypothesis. If we present this to the DG — through Soo-Yeon, through the unofficial channel — as a research agenda that addresses the containment concern, the DG has a third option. Not the US facility. Not continued HA suspension with an inevitable transformation. A research program aimed at making the transformation survivable."
"The third option."
"The option that serves everyone's interests. The HA retains jurisdiction — they're not giving up their authority over the world's only curse eater. The US gets intelligence sharing and medical data — the counter-proposal stays in effect. And the containment concern is addressed not through a facility but through research that might make containment unnecessary. The DG can present this to the delegation as the superior solution. Better than custody because custody doesn't actually solve the containment problem — the previous eater transformed in a government facility and the facility didn't prevent the casualties. Directed transformation does."
"If it works."
"If it works. Which requires the research. Which requires the time. Which requires the hazard exemption and the counter-proposal and the DG's cooperation and all of the pieces that are already in play, arranged not randomly but deliberately, each piece serving the plan that the plan didn't know it was serving until the Collective provided the data that made the plan visible."
She was typing fast now. The proposal taking shape on the screen — sections, subsections, the organizational structure of a scientific document that was also a diplomatic document that was also a survival document. Research objectives. Methodology. Timeline. The sections that every proposal required and that this proposal would present to a Director-General who needed a reason to refuse the United States and who would find in the proposal not just a reason but a better future than the one the delegation had brought in their briefcases.
Zeke watched her work. The watching — passive, the body in the chair, the mind in the space between exhaustion and alertness that the 3 AM hours produced. The Hemingway in his pocket. The old man. The fish.
The alternatives were a pipeline or a kill order. The third option was a chance.
Tanaka typed until dawn.
