The hum of the coffee shop was a low-fidelity drone—the clatter of ceramic mugs, the hiss of a steam wand that sounded like a leaking radiator, and the rhythmic, aggressive tapping of Sarah Jennings' fingers on her keyboard.
She looked at me, her eyes narrowed with a sharp, academic hostility. In 1998, a woman in a computer science graduate program was already a survivor of a thousand small wars. She didn't need a stranger critiquing her recursive functions.
"You're nudging my bag, looking at my screen, and telling me my convergence rate is going to stall," she said, her voice a cool, controlled alto. "That's a lot of assumptions for a guy who looks like he's lost on his way to an Econ 101 lecture."
I stayed on one knee, holding the paper I'd picked up. I didn't look away. In my previous life, Sarah had been the only person who could match my pace during the development of the Horizon Spire's neural-net load balancers. She was a woman who respected only two things: raw data and the courage to follow it to its logical conclusion.
"It's not an assumption," I said, pointing to a line of her C++ code. "You're treating the web like a static library. You're using a global ranking vector. But the web isn't a library; it's an organism. By the time your crawler hits the ten-millionth node, your memory overhead will exceed the capacity of a standard server rack. You'll be swapping to the disk, and your search time will drop from milliseconds to minutes."
Sarah paused. She looked at the line of code, then back at me. The hostility didn't vanish, but it was joined by a flicker of genuine curiosity—the kind that only sparks between two high-level engineers.
"And you think a non-linear graph is the answer?" she asked, her tone shifting from defensive to interrogative. "How do you propose to calculate the eigenvector of a matrix that's expanding by ten percent every week without a supercomputer?"
"You don't calculate it globally," I said, standing up and pulling out a chair. I didn't ask for permission to sit. "You distribute the weight. You shard the data across a mesh of low-cost commodity hardware. Use a map-reduce architecture. Let the nodes handle the local clusters and only sync the deltas."
Sarah stared at me. The terms I'd used—shard, mesh, map-reduce—were whispers of the future. In 1998, the industry was obsessed with "big iron"—massive, expensive Sun Microsystems servers. The idea of using a thousand cheap PCs to do the work of one supercomputer was still a fringe theory being toyed with in the backrooms of Google's garage.
"Map-reduce?" she whispered, the words tasting strange in her mouth. "That's... theoretically possible, but the latency on the synchronization would kill the performance."
"Not if you use a proprietary hardware-level encryption bridge to bypass the standard TCP/IP bottlenecks," I said. "I have a guy in the Mission working on the boards right now."
I reached into my bag and pulled out a business card. It was blank, except for a phone number and the name of my shell company: Architectural Revision Systems.
"My name is Nathaniel Han. I'm starting a private research lab. I don't want to build a better search engine, Sarah. I want to build the infrastructure that makes search engines obsolete."
Sarah looked at the card, then at my face. "How do you know my name?"
I didn't blink. "I saw it on the cover of your thesis proposal when I picked up your papers. 'Recursive Node Analysis in Non-Linear Data Sets.' It's a brilliant piece of work. It's also doomed to be buried in a filing cabinet because the people with the money in this valley are too busy chasing the next 'Amazon for Pet Food' to understand what you've actually discovered."
That hit home. I saw the slight slump in her shoulders, the exhaustion of a visionary who had been told 'no' by people half as smart as she was.
"I'm a grad student, Nathaniel," she said, leaning back. "I have a fellowship. I have a path."
"You have a path to a cubicle," I countered. "In five years, you'll be an Associate Architect at a firm that doesn't value you. In ten, you'll be a middle-manager. In twenty, you'll realize you let the most important technological shift in history happen without you."
I pulled out a checkbook—another relic of this era. I wrote a check for five thousand dollars and pushed it across the table.
"That's for your time today. Consider it a consultation fee. If you want to see the hardware I'm talking about, meet me at Valdez Electronics in the Mission tomorrow at 6:00 PM. If you don't show, keep the money. You'll need it for the tuition hikes coming next semester."
I stood up, tucked my chair in, and walked away before she could respond.
I didn't look back. I knew how the "Future Giants" worked. You didn't recruit them with promises of stability; you recruited them with a challenge that they couldn't solve alone.
As I stepped out into the bright Palo Alto sun, my pager buzzed.
It wasn't a text this time. It was a phone number with a 415 area code. I recognized the extension. It was the internal line for the Senior Partner's office at Sterling & Loftus.
The "Ripple Effect" was reaching the shore.
I didn't call back from a payphone. I drove to a crowded mall in San Jose, found a busy food court, and used a prepaid phone card. The noise of the crowd would mask my voice and the background sound.
"This is Han," I said when the line picked up.
"Where the hell are you?" Arthur Sterling's voice was a jagged rasp of panic. "The office is crawling with people, Nathaniel. Not the police. Worse."
"Define worse, Arthur."
"Two men. One in a suit that costs more than my car, the other... he didn't say a word. They're from a firm called 'Strategic Risk Mitigation.' They have a letter from the Treasury Department authorizing a 'technical audit' of our escrow accounts. They specifically asked for the logs of the terminal in my lounge."
I felt a cold sensation in the pit of my stomach. Strategic Risk Mitigation.
In 2025, they were known as the "Janitors." They were a private intelligence firm used by the largest investment banks to "clean up" anomalies—the trades that shouldn't have happened, the people who knew too much, the glitches in the matrix.
If they were already at Sterling & Loftus, it meant my trade hadn't just been flagged by a computer. It had been flagged by a human who understood that a junior engineer making four million dollars in four hours wasn't luck. It was a leak from the future.
"Did you give them the logs?" I asked.
"I'm stalling! I told them the server was undergoing a scheduled maintenance cycle. But they aren't stupid, Nathaniel. They're sitting in the lobby. They aren't leaving until they have a name."
"You give them a name, Arthur, and you're admitting to a felony," I said, my voice dropping to a low, predatory hum. "You let a junior employee use an escrow account for speculative currency trading. That's not just an SEC violation; that's wire fraud. They'll strip your license, seize your assets, and you'll spend your golden years in a federal penitentiary."
I heard Sterling's breath hitch. He was a man who had lived his whole life on the right side of the law, even if he bent it. The reality of a prison cell was the only thing more terrifying to him than the men in the lobby.
"What do I do?" he whimpered.
"You tell them the truth, but a modified version. You tell them that the trade was part of a proprietary 'Stress-Test Algorithm' developed by a third-party consultant. A company called Architectural Revision Systems. Tell them all correspondence was handled through a blind legal proxy in the Caymans."
"Will they believe that?"
"It doesn't matter if they believe it. It creates a legal wall. It moves the target from a person to a corporation. And Julian Vane in the Caymans is paid very well to make sure that corporation is a ghost."
"And what about you?" Sterling asked.
"I'm already gone," I said. "And Arthur? If you see that black sedan again, don't go home. Go to a hotel. Stay in a public place."
I hung up the phone and leaned my forehead against the cool plastic of the booth.
The "Hard Realism" of the world was reasserting itself. I had changed the timeline by exactly one day, and already, the guardians of the old order were mobilizing. In my previous life, the "Janitors" didn't even know I existed until I was forty. Now, they were hunting me before I'd even bought a decent suit.
The butterfly's wings weren't just flapping; they were creating a hurricane.
I spent the next six hours in the San Jose Public Library. I didn't use the computers; I used the physical archives of the Wall Street Journal and The Financial Times from the last six months.
I needed to find the "blind spots"—the events that were supposed to happen but might be shifted by my interference.
The LTCM bailout was scheduled for September 23, 1998. The Federal Reserve would bring together the CEOs of fourteen major banks and force them to cough up $3.6 billion to save the fund.
But if the "Janitors" were active, it meant someone was already looking for the "leak." If I continued to trade the LTCM collapse, I would be walking into a trap. They would be watching the order flow of every major exchange, looking for the specific signature of my "Future Knowledge."
I needed to pivot. I needed an asset that was beneath their notice.
I found it in a small blurb on page 14 of the business section: a local utility company in Northern California, Pacific Grid & Electric, was looking to divest a series of "obsolete" fiber-optic lines buried alongside their gas pipes.
In 1998, fiber-optic was considered a "dead" asset. The "Dark Fiber" bubble had burst early, and everyone thought the world would be powered by copper and satellite.
I knew better. In five years, these "obsolete" lines would become the backbone of the broadband revolution. In ten, they would be the nervous system of the cloud.
The asking price for the entire Northern California loop? Two million dollars.
I could buy it. I had the capital in the Cayman trust. And unlike a currency trade, buying physical infrastructure was slow, boring, and perfectly legal. It wouldn't flag a "Risk Mitigation" AI. It would just look like a stupid investment by a naive shell company.
But to manage that grid, I needed Elias Valdez's hardware.
The sun was setting as I pulled up to Valdez Electronics. The Mission was a shadow-play of orange light and deep, violet silhouettes.
I walked into the shop. The smell of soldering flux was even stronger than the day before.
Elias was sitting at his bench, his eyes bloodshot behind his thick glasses. On the table in front of him sat a small, green circuit board. It was beautiful—a tight, elegant sprawl of components that looked like a miniature city.
"You're late," Elias said, not looking up.
"Traffic," I said. "Did you get the Rogers boards?"
"I got them. And I burned through three of them trying to match your specs. You're asking for a frequency hop that shouldn't be possible with this silicon, Nathaniel."
"Shouldn't be, or isn't?"
Elias finally looked up. He tapped the board. "I bypassed the internal clock. I used an external quartz oscillator and a custom logic gate to force the packet timing. It's... it's fast. I tested it on a closed loop. The latency is under ten microseconds."
In 1998, ten microseconds was magic. It was a speed that wouldn't become industry standard until the 2010s.
"Can you scale it?" I asked.
"If I had a clean-room and ten more of these boards, I could build a relay that would make the Pacific Bell network look like a tin-can-and-string setup. But why? Who needs this kind of speed?"
"I do," a voice said from the doorway.
I turned. Sarah Jennings was standing there. She had changed out of her school clothes into a dark sweater and jeans. She looked at the shop with a mixture of disdain and fascination.
"I tracked your company name, Nathaniel," she said, walking toward us. "Architectural Revision Systems doesn't exist. It's a shell registered in Delaware three days ago. And your 'consultation fee' check cleared at a branch in Grand Cayman."
She stopped in front of me, her eyes boring into mine.
"You're not an engineer for a firm. You're not a 'researcher.' You're someone who just made a lot of money very quickly, and you're trying to build something that the government usually keeps for itself. So I'm going to ask you one more time: Who are you?"
I looked at Elias, then at Sarah. My "Shadow Cabinet." The hardware genius and the software architect. Both of them were nobodies right now, but together, they were the key to the next thirty years.
"I'm the man who knows what's coming," I said.
"That's not an answer," Sarah snapped.
"The Russian default is the beginning of a five-year cycle of instability," I said, my voice quiet but echoing in the small shop. "By 2000, the dot-com bubble will burst. Trillions of dollars in paper wealth will vanish. By 2001, the world will be at war. By 2008, the banks will collapse. And by 2020, the infrastructure of the world—the power, the water, the information—will be controlled by four or five people who don't care if you live or die."
I pointed to Elias's board.
"I'm building a parallel system. A grid that they can't turn off. A network that doesn't rely on their servers. I need the hardware, I need the code, and I need the fiber-optic lines to connect them. I have the money, but I need the two of you to make it work."
The silence in the shop was absolute. I could hear the hum of the neon sign outside.
"You're talking about a revolution," Elias whispered.
"No," I said. "I'm talking about survival. In ten years, you'll thank me. In twenty, the world will try to kill us for this."
Sarah looked at the circuit board, then at me. She didn't look convinced, but I saw the calculation in her eyes. She was a scientist. She wanted to see if the hypothesis was true.
"The fiber-optic lines," she said. "How are you going to get them? PG&E isn't just going to hand them over to a kid with a Cayman checkbook."
"They will if I offer to take over their maintenance liabilities for the next fifty years," I said. "They see it as a debt. I see it as a kingdom."
Before either of them could respond, a heavy thud echoed from the front of the shop.
The glass of the front door shattered inward.
A flash-bang grenade skittered across the floor, its fuse hissing.
"Down!" I screamed, lunging for Sarah and Elias, tackling them behind the heavy steel workbench.
BANG.
The world turned white. My ears erupted in a high-pitched scream. My vision was a blurred mess of spots and shadows.
Through the haze, I saw three figures in tactical gear stepping through the broken door. They didn't have badges. They didn't have uniforms. They had suppressed submachine guns and the clinical, efficient movement of professional killers.
The "Janitors" weren't waiting for the audit. They had decided to skip the paperwork.
The butterfly effect had just turned into a slaughter.
----------------------------------------------
By attempting to recruit Sarah and Elias early, Nathaniel has inadvertently centralized his "risk" in one location. The "Janitors" have been authorized to use extreme measures to recover the "stolen" algorithms they believe Nathaniel is using. As the first bullets hit the steel workbench,Nathaniel realizes that his 2025 knowledge of corporate law is useless in a 1998 gunfight. The future is no longer a guide—it's a target.