Team Apex had publicly declared the Riemann Hypothesis as their next challenge.
But the team was not in a situation relaxed enough to devote themselves solely to this.
“Seoha! The approval to use ‘The Core’ has come through!”
Suri came rushing frantically into the lab.
On the tablet in his hand, an email from the operations committee was displayed.
“Did it really happen this easily?”
Sujeong’s eyes went wide.
Teo shook his head.
“As if. It’s because of what we’ve accomplished that we’re getting special treatment.”
The Core was not a facility that just anyone could use.
Just obtaining a single account required dozens of documents.
Research objectives, plans, data import and export routes, legal review, security training—and before final approval, one had to go through ethics review and a vote by the operations committee.
“Then we can finally build the Oracle we’ve only ever imagined.”
Seoha rose from his seat and fetched a box he had stored in the storage room. It contained documents organizing the fundamental concepts of the Oracle algorithm.
A design whose calculations were so complex that ordinary workstations couldn’t even begin to attempt them.
“I can’t believe the day has come to use it… I’m nervous already.”
Suri rubbed his palms together as if looking forward to it.
“Suri, you especially need to be careful. Press the wrong key once, and you might get hauled off somewhere.”
A supercomputer is something fundamentally different from an ordinary workstation.
Stopping a job once it’s in the queue isn’t easy, and it wasn’t rare for a single poorly written line of code to waste thousands of cores for hours on end.
So for researchers using The Core, it was standard practice to do not just double, but triple checks before hitting Enter.
“Shall we take a look at this together today?”
Thud—
Seoha set the box down on the desk.
Inside were formulas and diagrams, and memos densely written by hand. It was a blueprint closer to a thought experiment than code.
“Is this all of it?”
Sujeong, whose eyes had briefly clouded over, quickly came to her senses and approached.
Seoha looked at her with warm eyes.
The team members exchanged opinions late into the night, adding flesh to the skeleton of the algorithm Seoha had created.
It was an endlessly complex and abstruse system, but for some reason, Sujeong didn’t find it difficult. The reason was none other than this:
‘At least this one has a definite answer.’
For several days, she couldn’t get food down properly.
She tied her hair up tightly and immersed herself in designing the algorithm.
***
“Welcome.”
Mayor John had come to the MIT campus.
Seoha, wearing a white lab coat, came out to the door to greet him. His face was noticeably gaunt, as if to show that the process of getting Oracle passed through Congress had not been smooth.
“Good to see you again, Seoha.”
He extended his hand with a bright expression.
“Things have been hard lately, haven’t they?”
At Seoha’s question, John gave a bitter smile.
Now that the bill had passed Congress, citizens were expecting traffic to improve any day. But as they say, after one mountain comes another; the negotiations were still going through painful labor.
“The legal side keeps putting on the brakes.
How to use the data, who takes responsibility, who can be sued if an accident happens… They’re demanding answers to all of it right now.”
Seoha nodded. They were all expected reactions.
Beep—
When the door opened, the full view of the lab was revealed.
The research team dispatched from New York had all returned here a few days ago.
Dozens of personnel were busily coming and going, doing their assigned tasks.
A researcher running with a laptop tucked under his arm, mathematicians writing something in front of a whiteboard, engineers gathered in threes and fives debating heatedly over data processing methods.
John looked around the lab with interested eyes.
He had immediately prioritized injecting manpower and funds here as soon as the budget passed.
“People are probably misunderstanding. They must be imagining a giant AI machine secretly running deep underground beneath New York City Hall.
In reality, Oracle is being built in a place like this.”
“Like Minority Report? If I get the chance, I’d love to experience it.”
Seoha laughed in a cheerful voice, as though the very thought was amusing.
In reality, of the 10-billion-dollar budget, only about 20 percent went to equipment and infrastructure. The largest portion of the budget would be spent on refining and integrating data.
“But everyone looks very busy.”
“Right now, this place is data hell.”
Seoha approached a nearby desktop and showed him the screen.
A massive city like New York is already split into thousands of systems.
Traffic signal systems, subway operations, police and fire dispatch systems, emergency medical call centers, road maintenance, energy management, public housing administration…
Countless streams of data were being transmitted here. The problem was that all of these were systems built at different times, by different entities, each with their own unique language and standards.
“Wait! This data hasn’t been adjusted for daylight saving time. Everything’s off by an hour!”
“The criteria for ‘delayed’ emergency response are different? For hospitals it’s one minute, for the NYPD it’s three minutes.”
“The legal team said never to use this data? They said there’s a risk of personal identification.”
Shouts rang out here and there.
John momentarily wore a blank expression at the reality so different from what he had seen on paper.
“I didn’t know it would be this chaotic.”
Refining data for a megacity like New York requires enormous budgets and time, and crucially, the efforts of countless highly skilled personnel.
They check data consistency and remove errors and duplicates. They also communicate with responsible officials to fill in missing information and unify spatial and temporal standards. And throughout this entire process, they must de-identify personal information that could become problematic.
“We’re still at the beginning stage.”
Seoha guided him to the deepest part of the lab.
Security was getting progressively tighter.
The access card levels went up one by one, and finally one had to pass fingerprint and iris scans.
Beep—
“Welcome. This is ‘The Core,’ where the brain of Oracle is being made.”
When the door opened, the air changed.
Unlike the noisy commotion outside, an indescribable tension could be felt here.
“Oh!”
John looked around with an amazed expression.
Towering servers far exceeding human height were lined up along one wall.
Cold air blasting from the cooling units, status indicators blinking regularly—it was the high-performance computing cluster that MIT boasted of.
Equipped with tens of thousands of CPU cores, large-scale GPU nodes, and even an ultra-high-speed interconnect system rarely seen in university research facilities.
While not on the level of national supercomputers (ORNL, LLNL), it was in the top tier in the world for research purposes.
“Dean Whitman went through quite a lot so we could use this. Though of course, the budget came from the City of New York.”
John couldn’t contain his admiration as he walked along the server racks.
“To think it’s on this scale… It would have been impossible to design this in New York anyway.
In fact, petitions saying Oracle is realistically impossible are pouring into City Hall every day. But seeing this, I feel confident that we can do it.”
“Shall I put your mind at ease a little more?”
Seoha guided him in front of a large monitor in the center.
Logs were streaming across the screen in real time.
“In fact, we’ve already started testing.”
“Is that true?”
John’s eyes widened.
Suri was sitting in front of the keyboard. Beside him, Sujeong and Teo each had their laptops open, watching the monitor.
“What we’re running now is the first module of the Oracle algorithm. It’s the pattern recognition engine.”
Seoha pointed at the screen with his hand.
“When we built the satellite AI before, we had pre-refined the traffic data. We trained it on ten years’ worth.”
“Ten years? How long did that take?”
“It would have taken over a year on a regular workstation. The Core did it in forty-eight hours.”
“Remarkable.”
John suddenly looked around as if realizing something.
“But is this everyone? There are only four of you.”
It was far too few people to design the core brain of a 10-billion-dollar project.
Seoha made a playful expression as he looked at John.
“We’re an elite, small-scale team.
Mathematics is a field where one exceptionally talented person can achieve higher efficiency than one hundred slightly less capable people. You could think of it as a structural characteristic of mathematics itself.”
Seoha looked at Sujeong and nodded.
Sujeong rose from her seat and approached the whiteboard. A huge, complex diagram was drawn there.
“Hello, Mr. Mayor. I’m Im Sujeong, a mathematician with Team Apex who mainly handles number theory. I’ll briefly brief you on the overview.”
John nodded with an interested expression.
“Oracle is composed of a total of seven layers. The pattern recognition we’re running now is only Stage One.”
Sujeong explained, pointing at each layer.
“Stage Two is causal inference, Stage Three is external variable integration, Stage Four is simulation, Stage Five is optimization algorithms, Stage Six is uncertainty quantification, and the final Stage Seven is…”
As Sujeong trailed off, John looked at Seoha. Seoha finished Sujeong’s line.
“The decision-making system.
The output of each stage becomes the input for the next, and the result that finally emerges through Stage Seven becomes the Oracle’s ‘prediction.’”
“Then you’ll need a digital twin at Stage Four.”
Seoha nodded.
“We still have a long way to go to get there.”
Suri, who had been sitting, suddenly sprang to his feet.
“Seoha! The Stage One test is finished!”
Everyone rushed to the monitor, seemingly curious about the results.
From behind, John watched them with a satisfied expression.
***
“Sujeong noona, about the mathematical model for Stage Two causal inference.”
Even after John left, work continued.
The four gathered in a conference room to continue discussions on the next step.
“Now that the pattern is established, I think we’ve sufficiently secured statistical correlation. Now it seems like it’s noona’s turn to shine, right?”
Sujeong swallowed hard.
‘At what point and under what conditions does the system undergo qualitative change?’
Seoha had decided to adopt an integer constraint and stepped-structure modeling approach here.
Integer constraints always make calculations more difficult.
Problems that could be overcome through differentiation and approximation in continuous models all explode the moment they switch to integers.
In a city-scale system, the difficulty rises exponentially.
“Yeah, got it. I’ll try to complete it by the next meeting.”
Sujeong nodded.
After the meeting, Suri and Sujeong went back. Teo hesitated for a moment as he was about to leave, then approached Seoha.
“Seoha. Sujeong looks like she’s struggling these days. Are you doing this on purpose?”
Anyone could see Sujeong was overexerting herself. Seoha smiled awkwardly and nodded.
“……”
Seoha seemed about to say something but stopped, making a troubled expression.
“Why? What were you going to say?”
“…Noona is right in the middle of explosive growth right now.”
Teo had noticed it too.
The most important moment for a mathematician—
when facing a problem that surpasses one’s limits, a mathematician grows. And sometimes fails.
“Is this really okay?”
Teo had given Sujeong that problem because he knew that as well.
But what Seoha was demanding of her now was not merely on that level.
“This is a secret.”
Seoha approached quietly and spoke in a low voice.
“Sujeong noona is someone incredibly strong in the face of adversity. Though she doesn’t seem to think so herself.”
From the first time they met, Sujeong had been someone who didn’t avoid problems.
“So, shh!”
Seoha moved away.
Left alone, Teo looked up at the ceiling and murmured with a blank face.
“…Don’t tell me Suri and I are going to go through the same thing?”
The room temperature was moderate, but an eerie cold seemed to send a chill through his body.