[Café on Valencia Street — January 2014, Afternoon]
The notebook sat open between the laptop and a half-empty coffee cup, and the list was embarrassingly short.
Ethan had walked fourteen blocks to reach a café with decent wifi and enough ambient noise to think. The apartment's silence had started pressing in on him — too quiet, too still, too much like a dead man's room. The café was warm. The espresso machine hissed at regular intervals. A Bon Iver track played at tasteful volume. He could work here.
Except the work was going badly.
The notebook — spiral-bound, purchased from a Walgreens on the walk over — contained everything he could remember about the show. He'd organized it into three columns: CONFIDENT, VAGUE, and WRONG?
Under CONFIDENT, the list was short:
· Richard Hendricks creates middle-out compression algorithm
· TechCrunch Disrupt: Pied Piper wins with Weissman Score of 5.2
· Gavin Belson runs Hooli, becomes primary antagonist
· Peter Gregory funds Pied Piper (dies — actor died?)
· Erlich Bachman owns the incubator, gets 10% equity
· Gilfoyle = systems/security. Dinesh = Java. They hate each other (but not really)
· Jared (Donald) = business/operations, weirdly dark backstory
· Big Head fails upward constantly
Under VAGUE:
· Something about Raviga Capital and a woman named Laurie
· Russ Hanneman — billionaire, "three comma club," invested in Pied Piper at some point
· Video chat pivot? Season 3 or 4?
· Jack Barker — "Action Jack" — brought in as CEO, screws everything up
· Pied Piper eventually becomes a decentralized internet thing
· Blood boy — Gavin gets transfusions from a young person? Season 4?
· Jian-Yang copies something. Makes a hot dog app. "Not hot dog"
· The ending involved AI being too dangerous and Richard sabotaging it?
Under WRONG?:
· (This was the column that scared him. Half of VAGUE probably belonged here.)
Eight seasons or six? Six. He was fairly sure. Fifty-something episodes total. But the order of events — which season, which episode, which came before what — was a jumbled mess. He'd consumed the show the way most people consume streaming content: in a hypnotic haze, one episode bleeding into the next, plot points dissolving into general impressions.
He knew Richard was anxious and brilliant. He knew Gilfoyle was deadpan and competent. He knew the broad shape of the story — underdog startup fights corporate giant, pivots repeatedly, succeeds and fails in cycles.
But he couldn't tell you what happened in Season 3, Episode 4. He couldn't recite the terms of Pied Piper's Series A. He couldn't map which characters appeared when, or which conflicts belonged to which season.
Shallow. His knowledge was tragically, dangerously shallow.
Ethan capped his pen and opened the laptop. TechCrunch loaded. Hacker News loaded. He opened a dozen tabs and started reading.
The 2014 tech landscape was alien territory. Not because it was unfamiliar — he'd read about this era in retrospectives and history pieces. But experiencing it firsthand was different from reading about it. The front page of Hacker News featured discussions about MongoDB, Node.js, and whether "the cloud" was a real business model or a fad. A startup called Yo — an app that just sent the word "Yo" — had raised $1.5 million. Someone was arguing that Ruby on Rails was the future of web development.
No PyTorch. The framework wouldn't exist until 2016. No TensorFlow — Google wouldn't open-source it until November 2015. No Hugging Face, no Weights & Biases, no MLflow. The entire machine learning ecosystem he'd relied on daily was either nonexistent or locked behind institutional walls.
"Neural network" appeared on Hacker News exactly once in the current top stories, in a dismissive comment about academic research with "no practical applications." The AI winter hadn't fully thawed. Deep learning was a curiosity, not a revolution. Yann LeCun and Geoffrey Hinton were names that appeared in academic papers, not Bloomberg profiles.
VCs were funding Uber-for-X clones. Social networks. Mobile apps. "Big data" was the buzzword du jour, meaning anything involving a database larger than a spreadsheet. The word "AI" in a pitch deck would get you polite smiles and no check.
This was the world he had to operate in. A world that thought neural networks were interesting toys, that "machine learning" meant recommendation algorithms for Netflix, that the idea of a computer generating human-quality text was science fiction.
And he was going to build a Transformer.
The absurdity of it lodged in his throat like a dry Cheerio. He took a long pull of coffee to wash it down.
---
[Same Café — One Hour Later]
The startup team at the table behind him was loud. Four people — two men, two women — arguing about their product roadmap with the frantic energy of founders who'd just closed a seed round and had no idea what to build next.
"We need to pivot to enterprise," the guy in the Patagonia vest said. "Consumer is dead. Nobody's going to pay for another to-do app."
"It's not a to-do app, it's a productivity ecosystem—"
"It's a to-do app with a chat feature. Enterprise or death."
Ethan half-listened, half-worked. But something was happening. A tugging sensation at the edge of his awareness, like a word on the tip of his tongue. He glanced at the team.
Numbers.
Not literal numbers floating over their heads — nothing that crude. But an impression. A weight. When he looked at the guy in the vest — the one doing most of the talking, the one who'd introduced himself as CEO — something in Ethan's mind registered a value. A four, if he had to assign a scale. Confident. Loud. Average technical ability at best.
The woman next to him — quieter, glasses, hadn't spoken in ten minutes but had been sketching wireframes on a napkin — pinged differently. A six. Solid. Not exceptional, but competent, thoughtful, the kind of engineer who would reliably ship features without burning out.
The second guy: a five. Good enough. Not special.
The second woman, the one who'd been typing on her laptop through the entire argument, occasionally lobbing a correction that the others ignored: she registered as a seven. High. The best person at the table by a clear margin, and the one with the least airtime.
Ethan blinked. The numbers faded when he stopped looking. Returned when he refocused.
Ability number four. He ran a quick mental catalog: the architecture blueprint in his head. The coding acceleration. ChronoCloud. And now this — some kind of talent radar. A way to quantify what took hiring managers months and reference checks and technical interviews to assess, compressed into a single glance.
The possibilities were staggering. And the limitations were immediately obvious. A number told him someone was technically strong. It didn't tell him if they were reliable. If they were kind. If they'd steal IP or poison a team's culture. The best coder in the world was useless if they couldn't work with other humans.
Still. In a world where the most important resource was engineering talent, and where finding that talent was a crapshoot of résumés and whiteboard interviews — this was a weapon.
He tested it on others in the café. The barista: a two. Not an insult — she was making espresso, not training models. A guy reading a machine learning textbook in the corner: a five. The woman across from him coding in what looked like Objective-C: a four.
The numbers were consistent. Same person, same number, whether he looked once or three times. Reliable. Calibrated.
Ethan turned back to his notebook and added a fourth column: ABILITIES.
1. Architectural Intuition — Transformer blueprint, spatial, permanent. Strains with extended focus. 2. Accelerated ML Cognition — coding speed boost, ML-specific only. Triggers in flow state. 3. Temporal Compute (ChronoCloud) — future GPUs, extreme cost. Unexplainable to anyone. 4. Talent Resonance — numerical assessment, technical ability only. Doesn't read character.
Four tools. All powerful. All limited in specific and deliberate ways. Someone — something — had given him exactly what he needed to build an AI company in 2014, and exactly the constraints to make it difficult.
He closed the notebook and stared out the window. A Muni bus rumbled past. A man in a Warriors jersey jaywalked through an intersection. Rain was starting — the light, persistent kind that San Francisco specialized in, more mist than downpour.
Three days. TechCrunch Disrupt was in three days, according to the tech blogs. The event where Richard Hendricks would demo Pied Piper's compression algorithm and score a 5.2 on the Weissman Scale. The event that kicked off everything.
He didn't need to compete at Disrupt. He didn't need to announce himself or pitch or network. He needed to be in that room. Watch the Pied Piper team. See the show characters in person. Assess them — not with his faulty memory of a binge-watched comedy, but with his own eyes.
And his new ability to put a number on what he saw.
Ethan flagged the barista for a second cup. She brought it. He wrapped both hands around the mug and let the warmth soak through his fingers. Outside, the rain thickened. The café grew warmer, fogged at the windows, insulated from the January chill.
He had the blueprint for a technology three years ahead of its time. He had access to hardware that wouldn't exist for half a decade. He had coding ability that bordered on inhuman. He had a radar for talent that could staff a company in weeks instead of months.
And he had twelve thousand dollars, a dead man's apartment, and a head full of memories from a TV show he'd half-watched during a pandemic.
The coffee was good. He needed it to be. The warmth was the only thing grounding him to this table, this city, this body, this absurd second life.
Ethan drained the cup. Set it down.
Then he opened the laptop, navigated to the TechCrunch Disrupt registration page, and typed in the dead man's credit card number.
Author's Note / Promotion: Your Reviews and Power Stones are the best way to show support. They help me know what you're enjoying and bring in new readers! You don't have to. Get instant access to more content by supporting me on Patreon. I have three options so you can pick how far ahead you want to be: 🪙 Silver Tier ($6): Read 10 chapters ahead of the public site. 👑 Gold Tier ($9): Get 15-20 chapters ahead of the public site. 💎 Platinum Tier ($15): The ultimate experience. Get new chapters the second I finish them . No waiting for weekly drops, just pure, instant access. Your support helps me write more . 👉 Find it all at patreon.com/fanficwriter1