The cafe is nearly empty when I arrive. A Tuesday afternoon in July, the lunch crowd gone, the dinner rush hours away. Just the hiss of the espresso machine and the low hum of a refrigerator.
Gwen is already there, tucked into a corner booth with a laptop and a notebook covered in equations. She looks up when the door chimes, and something in her expression shifts when she sees my face.
I slide into the seat across from her.
"You look terrible," she says.
"Thanks. That's exactly what a guy wants to hear."
"I mean it. You look like you haven't slept."
I haven't. Not really. I spent most of last night staring at the EDEN blueprint on my Digivice, trying to find something I missed. Some shortcut. Some angle that would make the numbers work.
I didn't find it.
"The spatial encoding problem," I say. "It's worse than I thought."
Gwen closes her laptop. "Tell me."
So I do. I tell her about the flickering, about the way the gymnasium floor dissolved into void when we pushed past four users. About the geometry warping in Jessica's tower, the shadows disappearing, the physics engine stuttering under the load.
"The system can't maintain environmental integrity under simultaneous build conditions," I say. "Four users is the threshold. Past that, everything degrades. And Gwen, the fix isn't a patch. You said it yourself. The encoding architecture needs to be rebuilt."
"How long?"
"You said six months. Maybe more."
She nods slowly. "That's accurate. The spatial encoding is the foundation. If the foundation is flawed, everything built on top of it fails."
I lean back in the booth. The leather creaks. "So we're stuck. Four users in controlled conditions. That's not a network. That's a toy."
"What's the specific failure mode?" Gwen asks. "When the encoding degrades, what breaks first?"
"Environmental textures. Then geometry. Then physics. The system prioritizes user position and collision detection, so those hold the longest. But everything else falls apart."
"Interesting." She pulls her notebook toward her and starts writing. "So the encoding is trying to maintain too many simultaneous environments without a hierarchical priority system."
"Exactly."
"What if we didn't try to maintain everything at once?"
I frown. "What do you mean?"
Gwen taps her pen against the notebook. "In traditional data structures, you have linear hierarchies. Parent nodes, child nodes, everything connected in a predictable chain. But what if the spatial encoding used a non-linear approach? Something more like a neural network, where the system dynamically allocates resources based on user attention and interaction."
"That's not how the blueprint is designed."
"No. But the blueprint is failing." She looks up at me. "The Digivice architecture you showed me. The way it handles Digimon data storage. It's not linear either, is it?"
I think about the way the Digivice stores multiple Digimon, the way it accesses their data instantly regardless of how many are stored. The way it compresses and decompresses information on the fly.
"No," I say. "It's not."
"How does it work?"
"I don't know the specifics. But it's efficient. More efficient than anything I've seen in this world."
Gwen's eyes light up. "That's what we need to study. If we can understand how the Digivice handles data compression and dynamic allocation, we might be able to apply those principles to the spatial encoding."
I pull out my Digivice and set it on the table between us. The screen glows softly in the dim cafe.
"The System designed this," I say. "I don't know if we can reverse-engineer it."
"We don't need to reverse-engineer it. We just need to understand the principles." Gwen reaches out and touches the edge of the Digivice, careful not to activate it. "This device stores multiple Digimon, each with their own data sets, evolution trees, and abilities. And it accesses them instantly. No lag. No degradation. That's the kind of efficiency we need."
I look at the Digivice. I've been carrying it for months, using it every day, and I never thought about the underlying architecture. It just works. Like breathing.
"What kind of research are we talking about?" I ask.
"Unconventional data structures. Non-linear algorithms. Maybe some quantum computing theory, though that's still mostly theoretical." She flips to a fresh page in her notebook and starts listing topics. "I'll need to dig through academic papers. Talk to some people at YGD. Maybe reach out to Dr. Connors, he's done work on neural networks and pattern recognition."
"That's a lot of work."
"Yes."
"And no guarantee it'll lead anywhere."
Gwen looks at me. Her expression is steady, focused. "Ethan, there's no guarantee with any of this. We're building something that's never been built. The spatial encoding problem is just the first wall we've hit. There will be others."
She's right. I know she's right. But hearing it doesn't make the weight on my shoulders any lighter.
Except.
Except she's here. She's not walking away. She's not telling me the problem is too big or the timeline is too short. She's making a list.
"Thank you," I say.
Gwen blinks. "For what?"
"For not telling me this is impossible."
She smiles. It's small, but it's real. "Nothing's impossible. Just really, really difficult."
I laugh. It comes out rough, but it's genuine. "That's one way to put it."
We spend the next hour brainstorming. Gwen asks questions about the Digivice's functionality. I describe what I know, which isn't much, but she pulls threads from it like she's untangling a knot. Every answer leads to another question. Every question leads to a new possibility.
The cafe fills up around us. The dinner crowd arrives. The espresso machine hisses and steams. We don't notice.
At some point, a waitress brings us coffee. I don't remember ordering it. Gwen adds sugar to hers and keeps writing.
"The key," she says finally, "is dynamic resource allocation. The system needs to know what to prioritize and when. Right now, it's trying to maintain everything equally. That's why it fails under load."
"So we teach it to be selective."
"Exactly. We create a hierarchy of attention. What the user is looking at, interacting with, moving toward. That gets full encoding priority. Everything else gets compressed or temporarily suspended."
I think about that. "Like how the Digivice only fully materializes one Digimon at a time. The others are stored as data until they're needed."
Gwen points her pen at me. "Yes. That's exactly what I'm thinking."
We look at each other across the table. The Digivice sits between us, glowing softly.
"I'll start compiling research tonight," Gwen says. "There are papers on neural networks, on dynamic data structures, on attention-based systems. It'll take time to sort through everything, but I'll find something useful."
"How much time?"
"A few weeks. Maybe a month."
A month. That's better than six months. That's better than never.
"Gwen," I say. "I mean it. Thank you."
She closes her notebook. "You don't have to keep thanking me. This is my project too."
"I know. But still."
She reaches across the table and squeezes my hand. Her fingers are warm from holding the coffee cup.
"We'll figure this out," she says. "Together."
I nod. The weight on my shoulders is still there. But it feels different now. Less like a burden and more like a challenge.
We gather our things and step out into the evening air. The city is loud and bright, cars honking, people shouting, the subway rumbling underground.
"Same time tomorrow?" Gwen asks.
"Same time tomorrow."
She walks toward the subway station. I watch her go, then turn toward home.
My Digivice buzzes in my pocket. A message from Gatomon.
How did it go?
I type back with my thumb. Good. We have a plan.
That's what you always say.
This time I mean it.
I pocket the Digivice and start walking. The night is warm. The streets are crowded. And for the first time since the gymnasium floor flickered and dissolved into void, I feel like we might actually pull this off.
***
Donate Power Stones to support this novel
Advance chapters in patreon.com/Najicablitz