Chapter 12: Howard's Engineering Crisis (First Solution)
Howard's text arrived at 10 AM: "you still have that three-word note? I tried the obvious version of it and it's wrong but close."
I read the message twice. The three-word note — gradient-layered crystalline — had been sitting in my notebook for three weeks, annotated but not verified. I had added the fourth word (annealing) the night I wrote it, but the complete solution had remained theoretical.
The Synthesis Core had been processing the problem in background ever since. Not actively, not with deliberate focus, but as one of dozens of low-priority cross-references that ran continuously beneath conscious awareness. Three weeks of background processing had improved the initial insight significantly.
I texted back: "I have it. Coming to your lab."
Howard's engineering lab had the particular chaos of deadline pressure — prototypes in various states of assembly, components spread across multiple benches, a whiteboard covered in stress calculations with several sections crossed out and rewritten.
"The obvious version produces a secondary stress point," Howard said without preamble. "At the attachment interface. The gradient distribution handles the primary fatigue problem beautifully, but then the forces concentrate at the connection point instead."
"Show me."
He walked me through the problem. The component — a gripper mechanism for ISS maintenance arms — needed to withstand repeated full-extension operations without material fatigue. The standard solution was thicker walls, which added mass, which slowed response time, which made astronauts unhappy. My three-word suggestion had offered an alternative: distribute stress across a gradient rather than concentrating it at a single point.
The obvious implementation of this idea created a gradient within the component itself, varying material density from the stress point outward. It worked beautifully for the primary fatigue problem.
It failed at the attachment interface because the gradient termination created a new concentration point.
"It's like solving a leak by moving it," Howard said. "The water still has to go somewhere."
I looked at the stress calculations on the whiteboard. The Synthesis Core hummed at the back of my skull — the specific sensation of an output becoming available. Three weeks of background processing, cross-referencing Howard's engineering principles with Academy City's esper-field materials research, had produced something usable.
"The gradient shouldn't terminate at the attachment point," I said. "It should extend through it. Modify the attachment geometry so the gradient continues into the connection interface."
Howard's expression shifted from frustration to calculation.
"That would require redesigning the attachment bracket."
"Yes. But the redesign is simpler than you'd expect. Use a complementary gradient in the bracket — inverse distribution, so the two gradients interface smoothly rather than creating a transition."
"A complementary gradient." He was writing as I spoke. "So the stress flows through the connection instead of concentrating at it."
"Exactly. And there's a material substitution that makes this easier — the secondary stress point responds better to a different alloy than the primary structure. Titanium-aluminum interface rather than uniform titanium."
Howard stared at the calculations he had written.
"Okay," he said slowly. "That's either brilliant or you just described something from an engineering journal I haven't read yet."
"Could be both."
"You're an esper researcher."
"We work with physical systems."
"That's not an answer."
"No, it's not."
He looked at me for a long moment. The look was not suspicious — Howard did not operate in that register, not about technical competence. It was puzzled. Curious. The expression of someone who had encountered a phenomenon he could not immediately categorize.
"Where did you learn this?" he asked.
"Academy City's materials research overlaps with engineering more than you'd expect. Esper abilities interact with physical substrates, which means understanding material properties is part of the curriculum."
"This isn't curriculum material. This is... this is solution-level thinking. This is what someone does after years of working with a specific problem."
"I've been thinking about your problem for three weeks."
The explanation was true and also insufficient. Three weeks of conscious thought would not have produced this solution. Three weeks of Synthesis Core background processing, cross-referencing multiple disciplines, building on encoded engineering methodology from our first lab conversation — that had produced it.
Howard accepted the explanation because it was plausible and because he needed the solution more than he needed to understand where it came from.
"I'm going to test this tomorrow," he said. "If it works, I owe you a beer. If it doesn't work, I owe you a beer anyway because at least it's interesting."
"I'll take the beer either way."
Evening at the group apartment.
Leonard was working late, something about a deadline for a grant proposal that had crept up on him. Raj came over around 7 PM, and Howard arrived shortly after, still processing the engineering conversation from earlier.
The evening settled into the particular rhythm of friends without a specific agenda: food ordered, television on in the background, conversation that moved between topics without urgency or destination.
Raj talked about his latest dataset — a sequence of observations that might contain evidence of a trans-Neptunian object or might just be orbital noise with unfortunate timing. The distinction required more observations, which required more telescope time, which required convincing the scheduling committee that his search protocol was worth the investment.
"The problem is that I'm looking for something I can't prove exists," he said. "The committee wants evidence that the search is worthwhile, but the evidence is the search."
"That's a bootstrapping problem," Howard said. "You need results to justify the methodology, but you need the methodology to produce results."
"Exactly. And every time I try to explain this, someone asks why I can't just use the existing search protocols."
"Because the existing protocols might be filtering out exactly what you're looking for."
"Yes. But try explaining that to someone who helped design the existing protocols."
I listened to the conversation without contributing to it. The engineering solution was still processing somewhere in the background of my awareness, the Synthesis Core verifying its conclusions against new inputs from Howard's reaction. The verification was taking longer than expected, which meant either the solution had edge cases I hadn't considered or the background processing had produced something that needed more integration time.
Howard told a story about a conference presentation that had gone wrong in three different ways simultaneously — a projector failure, a co-author who arrived late, and a moderator who had mistakenly introduced him as someone else entirely. The story built through escalating absurdity, each disaster compounding the previous one, until the eventual resolution involved a borrowed laptop, a sprinted hallway, and the moderator's public apology.
Raj laughed. I laughed.
Not a polite laugh — an actual laugh, the kind that came from genuine surprise at a turn in the story I had not anticipated. The sound was unfamiliar in my own voice. I had been at Caltech for three weeks and I was not sure I had laughed like that before.
Raj looked pleased. Howard looked mildly alarmed and then pleased.
"See, that's what I'm talking about," Howard said. "You don't react to things the way people expect. Most people laugh at the setup. You waited for the punchline."
"The punchline was better."
"It was, but most people don't wait to find out."
The observation sat in the air for a moment. Howard was not accusing me of anything — he was noting a pattern, the same way Leonard had noted my lack of apparent anxiety. The pattern was accurate. I did wait longer than most people to react. I processed inputs more thoroughly before generating outputs.
It was a consequence of how the Synthesis Core operated. It was also, apparently, something people noticed.
"I'm a methodical thinker," I said. "It takes longer to find things funny."
"As long as you find them funny eventually."
"I do. Just not always at the standard pace."
The evening continued. Raj talked about a telescope calibration session that had gone wrong in three different ways simultaneously — a callback to Howard's story structure that made Howard groan and then laugh. The calibration failure had involved a miscommunicated coordinate system, a software bug that only manifested at specific altitudes, and a mouse that had somehow gotten into the control room.
"A mouse?" Howard said.
"A mouse. It ran across the keyboard during a critical observation window. We lost forty minutes of data because something small and furry pressed the wrong keys."
The absurdity was genuine. Raj told the story with the specific animation he brought to things he found genuinely ridiculous, his hands moving through the narrative, his voice rising at the critical moments.
I laughed again.
The sound was still unfamiliar. But it was becoming less unfamiliar.
At 9:45 PM, I noticed something.
I had been in the apartment for two hours and I had not written a single notebook entry. No encoding assessments, no Synthesis Core outputs, no observations filed for later analysis. Two hours of conversation and company and I had simply been present — not cataloguing, not processing, just there.
The realization was disorienting. The notebook had been my constant companion since arriving at Caltech, the repository for everything I observed and encoded and synthesized. Two hours without an entry was unprecedented.
I considered whether this was a problem.
The Synthesis Core hummed briefly — a background assessment of the situation — and then went quiet. The assessment did not produce an output. It produced something else: an absence of urgency. The two hours were not a problem. They were just two hours.
Some things do not need to be filed.
The thought arrived without the usual processing delay. It was mine in a way that Synthesis Core outputs were not — a personal conclusion rather than a calculated assessment.
Walking home, the night air was cool enough that my hands felt normal against it.
I checked the thermal differential anyway: two degrees above ambient, well within acceptable parameters. The TA was elevated, but the elevation was internal rather than external — the accumulated heat from three weeks of background processing that had produced Howard's engineering solution.
The cost was manageable. The warmth was manageable. The two unrecorded hours were manageable.
I decided not to put them in the notebook.
Some things did not need to be filed. Some moments were just moments — lived, experienced, and left to settle into memory without analysis or categorization.
The decision felt like progress. It also felt like risk. Academy City had trained me to document everything, to treat observation as the foundation of survival. Choosing not to document something was a departure from that training.
But I was not in Academy City anymore. I was in Pasadena, walking home from an evening with friends, and the night was quiet, and my hands were only slightly warm, and somewhere in a physics building three miles away, a calibration notebook sat in a desk drawer with entries that did not quite add up to anything yet.
Howard texted at 11 PM: "it worked."
Two words. No explanation needed.
I read the message, put my phone away, and went to sleep.
Get Early Access to New Chapters
Thank you for reading. For those who want to skip the wait, my Patreon is currently 21 chapters ahead of the public sites.
Schedule: 7 new chapters released every 10 days.
Benefit: Gain a significant lead of 7 to 21 chapters depending on your tier.
Support the project and start reading the next arc now: Patreon.com/IsekaiStories