The Grand
Symposium

The hall of the Institute of Genetics' annual public symposium is unlike most academic venues. Built in the late 2020s as part of the Institute's attempt to rebrand itself after returning to public ownership, the auditorium seats roughly six hundred, featuring warm birchwood panels lining the walls, a semicircular stage that curves gently toward the audience, and a single massive, dark screen behind the podium under soft, amber lighting designed to look approachable.

The crowd is a cross-section of everything the Institute now has to answer to: scientists in lanyards, journalists with open laptops balanced on their knees, and a scattering of generated humans^[1] in the upper rows, sitting close together. A few hold handmade signs, though security asked them politely to lower them for now and they complied, for now, as a moderator steps to the edge of the stage and adjusts the microphone.

The lights dim slightly. A single spotlight locks onto the entrance at stage left. The crowd settles, some still murmuring. Someone in the back coughs. A phone buzzes and is quickly silenced. Then the room goes genuinely quiet, the quiet of six hundred people holding their breath.

A tense ripple moves through the auditorium as the moderator steps back. A brief, uncertain applause rises from the middle rows and spreads outward, lasting barely four seconds before settling. In the front, a journalist with reading glasses hovers her pen over her notepad; in the gallery, the generated humans watch with difficult expressions, whispering in quiet pairs as a researcher in the third row crosses his arms to evaluate.

A journalist types quickly while the moderator, now seated off to the side, makes a note. The applause this time is warmer and more organized, though it remains measured: the scientists clap with restrained enthusiasm, and the press out of professional courtesy.

In the upper gallery, the generated humans exchange quiet, knowing glances; one young woman watches with folded arms, while another beside her claps twice, slowly, and stops. The word "sacrifices" hangs in the air, prompting a researcher in the fourth row to tilt his head, while a journalist underlines and circles the term on her notepad.

The first half of the announcement, the resumption of production, draws an immediate undercurrent of noise. Several scientists lean toward one another. A hand goes up in the fifth row, then lowers again when it becomes clear the floor is not yet open.

A journalist stops typing and stares at the stage. The older academic from before uncrosses his arms. A younger researcher near the aisle says something to herself, working through the implications.

In the upper gallery, the young woman who had kept her arms folded now sits forward. The young man beside her grips the railing in front of him. A quiet but agitated conversation breaks out among three of them in hushed tones.

The screen displays a schematic diagram of a human brain, the occipital lobe highlighted in Institute blue.

The murmuring grows and several hands remain raised. The journalist in the front row lowers hers slowly, but her pen is moving fast across her notepad. Another journalist, standing along the side wall, has stopped using her laptop and is recording on her phone.

In the upper gallery, the young man with close-cropped hair stands up halfway, then sits back down. The young woman beside him places her hand on the railing. A third generated human, previously quiet, lets out a short exhale that is audible in the rows nearby.

A senior scientist in the second row removes his glasses and someone from the upper gallery calls out with a controlled and clear voice:

Security near the side door shifts their weight. The moderator sits very still, watching the stage.

A heavy silence falls over the room. Several scientists in the middle rows exchange quick, tense glances, and a union representative near the aisle underlines a word twice on her notepad, staring down at it. The journalist who had been recording on her phone lowers it slightly, then slowly raises it again.

In the upper gallery, the young man with close-cropped hair sits back down fully now, but his expression has darkened. Another generated human on the same row, a young woman with short dark hair, speaks up with a measured voice:

Reactions ripple through the hall. Some shift uncomfortably and a researcher in the third row looks at the ceiling briefly. Another voice comes from the main floor, a middle-aged scientist near the center:

Some of the tension in the room releases. The scientist in the third row who had asked the question nods slowly. The journalist in the front row has stopped underlining and is now writing longer lines.

In the upper gallery, the young woman who has been watching the stage with folded arms finally unfolds them. A younger researcher from the fourth row who has been quiet since the beginning of the presentation raises their hand:

In the gallery, the young man with close-cropped hair leans forward again.

Six hands are being raised up simultaneously. Two people begin speaking before catching themselves. The postdoctoral researcher who asked the previous question sits back in her chair with disbelief. The journalist in the front row stops writing entirely and simply stares at the stage. In the upper gallery, the young woman folds her arms again, tighter than before. The young man with close-cropped hair stands up fully this time.

Another generated human, previously silent throughout the entire presentation, speaks with quiet fury:

A researcher in the second row, the senior scientist who had removed his glasses earlier, puts them back on and raises his hand with the deliberate calm of someone choosing words very carefully:

The scatter plot sits on the screen behind the stage.

The room falls into a long silence and then from the upper gallery, the young man with close-cropped hair speaks, and his voice is no longer restrained:

A murmur of agreement moves through the hall. A woman near the center aisle raises her hand. She has the lanyard of a union representative, the Institute's own scientific workers' union, visible against her jacket:

The moderator shifts in their seat and glances toward the side door. The scatter plot remains on screen.

The union representative lowers her raised hand and writes something in her notebook with short, firm strokes. The senior scientist in the second row removes his glasses again. He sets them on his knee this time rather than his face. Three journalists are now recording openly and one has stood up and moved toward the aisle for a better angle. The generated humans are quiet and several of them look at one another. The young woman with folded arms says something very low to the person beside her. The young man with close-cropped hair sits back down slowly.

Then a voice from the gallery, further back, quieter, almost conversational:

The scatter plot is still on the screen. All eyes are away from it.

A younger scientist near the back lets out a single, involuntary laugh, then covers his mouth. The three journalists recording have gone very still. The union representative has stopped writing and now she is simply listening now. In the upper gallery the generated humans are silent. The young woman with folded arms stares at the stage even more carefully. The senior scientist in the second row puts his glasses back on. The entire room leans forward at the unfinished sentence.

From the gallery, the quiet voice that had spoken before finishes it, almost gently:

After a long silence, the union representative stands up. She stands without raising her hand.

The postdoctoral researcher has her laptop open and is typing rapidly into a messaging application. The senior scientist in the second row stands as well, with the unhurried gravity of someone who has made a decision:

From the gallery, the young woman with folded arms speaks clearly without heat:

The scatter plot is still on the screen.

A journalist in the third row raises her hand and keeps it raised until the room completely quiets down:

A researcher near the center, who has been quiet, speaks with a flat and precise voice without raising his hand:

In the upper gallery the young man with close-cropped hair stands again with a measured, almost academic voice:

A ripple of grim, appreciative murmuring moves through both the gallery and the main floor. The union representative remains standing and the senior scientist speaks again:

The moderator is now standing fully, hands clasped, watching the stage with visible unease.

The journalist in the front row puts down her pen, the postdoctoral researcher closes her laptop and in the upper gallery, the young woman with folded arms unfolds them once again:

Sustained, unambiguous applause comes from the gallery. Several people on the main floor join it. The union representative speaks over the dying applause, still standing:

The senior scientist stands fully now and addresses the room rather than the stage:

The young man with close-cropped hair speaks, almost quietly:

The scatter plot remains on the screen.

The journalist in the front row covers her mouth briefly, then writes something down. The postdoctoral researcher looks at the senior scientist beside her. The young man with close-cropped hair speaks again:

The young woman who has been the speaks next:

A researcher in the fifth row begins to clap slowly. The union representative sits down for the first time, looking at her notepad. Then she looks at the stage with an expression that is almost sympathetic:

The moderator has stopped looking at their phone. They are watching the stage.

The room goes very quiet. The journalist in the front row writes one word on her notepad and underlines it twice. In the upper gallery, the young man with close-cropped hair speaks with a careful voice:

A woman in the fourth row, who has been quiet throughout the presentation, raises her hand. She has the lanyard of a medical ethics observer:

The room shifts again, and this time the shift is more complicated. The journalist in the front row has stopped writing and is now staring at the stage with her pen suspended, thinking. The postdoctoral researcher looks up from her hands. In the gallery the generated humans are silent. The ethics observer then speaks slowly:

The senior scientist looks up from the floor with a tired voice:

From the gallery, a young woman speaks:

The moderator does not move.

The lowered voice does not help as the microphone catches everything. The ethics observer sets down her pen. The journalist in the front row stands up without asking permission.

The postdoctoral researcher has gone very pale, looking at the senior scientist. A generated human further along the row, who has been silent throughout the entire presentation, says one thing, barely above a whisper:

The union representative closes her notepad. The senior scientist speaks with a finality attached to his voice:

The moderator walks toward the stage.

The moderator, who had been walking toward the stage, stops mid-stride. A younger researcher near the back begins laughing and the person beside him puts their face in their hands.

The senior scientist says nothing and simply nods once. In the upper gallery the young man with close-cropped hair looks at the young woman beside him and then looks back. After everything, after the throttling and the alignment and the defects, this moment produces something almost like disbelief on her face:

Scattered, exhausted applauses occur. The moderator stands exactly where they stopped, equidistant between their chair and the stage, deciding.

The moderator sits back down. There is nothing in their training for this.

A long silence falls over the hall. Then, from the main floor, a single voice, a junior researcher who looks approximately twenty-three years old and has been furiously taking notes since the beginning:

The journalist in the front row sets her pen down on her notepad and stares at the ceiling for a moment. The postdoctoral researcher has both hands pressed on her knees. The senior scientist takes off his glasses, folds them, and puts them in his pocket. In the upper gallery the young woman stands slowly:

The room is still. The moderator walks to the stage.

The moderator stops walking.

The assembly script appears on the screen, white monospace text on a dark background, clinical and incongruous above six hundred people. The junior researcher who had confirmed the registers reads it.

The postdoctoral researcher also reads it. She looks at the senior scientist. He has his glasses back on. A long, extraordinary silence fills the hall and the young man with close-cropped hair in the gallery speaks, and his voice is very quiet, almost reflective:

The young woman beside him looks at the script, then at the stage. She then says exhaustedly:

The moderator reaches the podium.

The bus is half-full on a grey afternoon. Alberta is sitting alone near the middle, a small canvas tote on her lap, looking out at the street. She has long wavy brown hair, a teal top with small buttons at the collar, and an open lavender cardigan. When Xue takes the seat beside her, she glances over almost immediately, as if she had already noticed the approach.

The bus stop serves three different lines, its low shelter of scratched plexiglass, uncomfortable four-person bench, and chronically out-of-date timetable rarely empty under the flat afternoon light that comes before rain. Alberta sits near the middle of the shelter, phone in both hands, shoulders slightly forward, without looking up.

Alberta pockets the phone and walks to the other end of the shelter without another word.

Alberta is at the far end of the station, almost invisible, phone out again.

The bus arrives. The doors open.

The podcast is recorded in a small studio rented by the hour, its walls lined with foam panels. Two chairs face each other across a low table, where a glass of water sits untouched beside the psychologist. To the side, a laptop screen displays blinking recording levels beneath the glowing red light.

The host leans forward slightly, elbows on knees.

The host glances down at their notes briefly, then back up.

The host sits back.

The host picks up their pen, then sets it back down without writing anything.

A short pause.

That sounds less like a personality structure and more like something closer to a compulsion.

The host lets a beat pass before responding.

The host types something on the laptop, scrolls briefly, leans closer to the screen.

The host looks up from the screen.

But they are explicit in the abstract that without access to the actual instruction set, without knowing what the chip is actually telling the brain to do at the lowest level, they can only observe the outputs. They cannot see the mechanism. It is like watching someone's behavior change after taking an undisclosed medication and trying to reverse-engineer the compound from the symptoms alone.

A short pause.

Which brings us back to your call for the Institute to comment. Because the Institute is the only party that holds the ISA, and they have stated publicly they will not release it.

The host types again, scrolling through results, expression shifting slightly as they read.

The host scrolls further.

There is a follow-up note in a community forum linked from the Frontiers paper, it looks like an open research thread, and they mention that the published instructions from the Institute appear to be deliberately minimal, the equivalent of releasing a dictionary with only twenty words in it. Enough to prove the language exists. Not enough to read anything written in it.

The host looks up.

Which some contributors to this thread are interpreting as a deliberate strategy rather than a partial disclosure made in good faith.

A short pause.

Does that track with what your patients have described when they talk about becoming aware of their own limits?

The host nods, setting the laptop slightly to one side.

A short pause.

I want to ask you something more personal, if that is alright. You mentioned your patients become aware of their limits over time. Some of them, you said, can override temporarily. In your clinical experience, what does that override cost them? Because from the outside, from the perspective of someone in a relationship with a generated human who is pushing past their limit to stay present, that might look like effort, like care. But you are sitting with these patients after. What are you actually seeing?

The host is quiet for a moment, turning the pen over in their hands.

A short pause.

Which raises a question I suspect you have already sat with. If the energy cost of overriding remains constant, or close to constant, indefinitely, then asking a generated human to maintain a secure attachment style is not a therapeutic goal. It is a chronic condition management goal. You are not helping them heal. You are helping them sustain an effort that will never become easier.

A short pause.

Is that a fair characterisation of where your research is pointing?

The host sets the pen down entirely.

A short pause.

For listeners, I want to note that we did reach out to the Institute of Genetics for comment before this recording. We received a statement saying, and I am quoting directly, "the Institute does not comment on ongoing research protocols." That was the entirety of the response.

Another pause, shorter.

We should also say, because this matters, that Siedlischte has been organising. That there are generated humans who are aware of exactly what you have described today, not in clinical terms, but from the inside. And that the reverse-engineering community you mentioned is partly made up of generated humans working on their own biology without institutional support, without access to the ISA, and without, apparently, any expectation that the Institute will cooperate.

The host looks at the guest.

Does that community give you any clinical hope, or does the scale of what was done make hope a difficult word to use right now?

The host nods slowly.

A short pause.

There is something almost deliberately isolating about that situation. The Institute holds the ISA. The peer review infrastructure is not equipped to handle the research that would challenge the Institute. And the community doing the most urgent work is the one with the least institutional support to get that work taken seriously.

A short pause.

I do not want to end on a conspiratorial note, because I think the more accurate word is systemic rather than deliberate. But the effect is the same regardless of intent.

The host glances at the recording software.

We are coming up on time. I want to give you the last word. If a generated human is listening to this right now, and statistically, some are, what do you want them to know?

The host lets the words settle for a moment before speaking.

The host reaches toward the laptop.

This has been an extraordinary conversation and I suspect it is one we will be returning to as the research develops. For listeners, links to the Frontiers paper and the open research thread will be in the episode notes. If you are a researcher in neuroscience, computer architecture, genetics, or clinical psychology and you are interested in contributing to this work, the contact details for the institute are there as well.

A short pause.

And for the Institute of Genetics, should anyone there be listening, the invitation to come on and respond remains open.

The red light goes off. The recording software shows the waveform flattening to silence.

The psychologist finally picks up the glass of water.

The IT lab of the University of Siedlischte is only three months old but already looks thoroughly lived-in, occupying the ground floor of a hastily repurposed former warehouse. The ventilation hums at a low, persistent frequency that everyone has long stopped noticing, while near the door, a pinned monitor displays the "CORTEX PROJECT - Open Reverse Engineering Initiative" website, its real-time donation counter sitting at 4,847 euros, enough to keep the servers online for another two months. On the central table, an empty chip housing, surgically removed from a volunteer, sits in a clear specimen container.

Seven people are in the lab this afternoon, some here since morning. Mira, 24, sits closest to the main analysis workstation under three open monitor windows, headphones resting around her neck. Beside her, Felix, a nineteen-year-old generated human, studies a printed copy of the partial ISA taped to the wall and annotated in three colors, while Dani and Ryo work back-to-back on separate machines at the far end. Near the door, Priya manages the incoming correspondence from researchers reacting to the podcast, while the red cursor blinks on Koko's screen.

Felix spins his chair around first, eyes on Koko's screen.

Mira pulls one headphone cup aside, leaning over without getting up.

Ryo stands up from the far end and walks over, coffee in hand.

Dani calls from the far end without turning around.

Priya looks up from the correspondence monitor.

The chip housing in the specimen container sits in the middle of the table. Seven people are now looking at Koko's screen.

A beat of silence. Then everyone speaks at once and then stops. Felix goes first.

Mira is already typing.

Ryo sets his coffee down.

Felix stands up and goes to his annotated wall sheet, uncapping a marker.

Dani has crossed the room and is standing behind Koko now.

Priya is typing fast.

Ryo picks his coffee back up and stares at the wall for a moment.

Mira stops typing.

Dani pulls up a chair beside Koko.

Felix writes "0x60 = NOP" and "0x90 = INT" on the wall sheet in red, draws a box around both.

Priya looks up from the keyboard.

Mira opens a shared document and starts building a table.

Dani pulls their chair closer to Koko.

Ryo sits back down at his machine.

Priya posts the update to the website and turns around.

Mira looks at the spreadsheet, then at the chip housing in its container.

The room settles into a different kind of quiet, the quiet of people who have just understood the shape of a very large amount of work and have decided to do it anyway.

The lab noise continues behind Koko, the sound of Felix sorting components, Dani reading the Ljubljana papers aloud to Ryo in fragments. Mira glances over from the harness script, catches something in Koko's expression.

Koko doesn't answer immediately. Mira rolls her chair over.

She reads the screen for a moment.

Felix looks up from the breadboard.

The component sorting stops. Dani puts down the Ljubljana paper.

Ryo speaks from across the room.

The room goes very quiet.

Felix has put down the soldering iron.

Dani stands up.

Ryo walks over slowly, reading over Koko's shoulder.

A short pause.

Felix looks at the chip housing in its container.

Dani sits back down slowly.

Nobody speaks for a moment. Priya has not posted anything. The donation counter on the monitor behind her ticks up one increment. Someone somewhere has just contributed.

Mira opens a new document, typing fast.

Dani paces slowly behind the table.

Ryo sits on the edge of the table.

Felix writes each type on the wall sheet with a column for known registers and a column marked unknown.

Mira stops typing for a moment.

Dani stops pacing.

The sentence lands heavily. Felix puts the cap back on the marker.

Priya speaks from the door.

Ryo sends a message to the TU group without being asked.

Mira is already opening tabs.

Dani sits back down with the Ljubljana papers.

Ryo pulls up the IBM Z architecture reference, scrolling.

Felix adds three new columns to the wall sheet: SPARC, MIPS, IBM Z.

Mira pauses on a tab.

Priya is writing a list on a notepad.

Dani looks at Koko.

The room is quiet for a moment. Ryo closes the IBM Z tab and opens a fresh terminal window.

A short pause. Then Dani laughs.

Mira looks genuinely thoughtful rather than dismissive.

Felix lowers his phone.

Priya adds it to the sourcing list without being asked.

Mira looks at the chip housing in its container, then at Koko.

The MCP4725 DAC^[16] sits half-assembled on the breadboard beside Koko, jumper wires trailing. The chip housing is still in its container. The donation counter ticks up again.

Mira sets up the harness script for the next test and looks at Koko.

Ryo writes a quick test script and pushes it to the shared repository.

Dani pulls up a multimeter from the equipment drawer and sets it on the table beside Koko.

Mira checks the I2C connection on the breadboard.

The observation sits in the air for a moment.

Priya opens the test log on the shared document, timestamp ready.

Ryo hits enter on the script from his machine.

The multimeter on the table begins showing voltage increments. Clean, regular, stepping upward. Everyone watches a different screen. Mira watches the harness log. Felix watches the multimeter. Dani watches the chip housing. Ryo watches the I2C traffic monitor. Priya has her hand on the timestamp key.

0.2 millivolts. 0.4. 0.8.

The harness log shows nothing but the standard timeout cycling.

1.2 millivolts. 1.8. 2.1.

Still nothing. Ryo exhales quietly.

2.6 millivolts. 3.0.

Mira sits forward.

Felix does not look up from the multimeter.

Dani reads the multimeter.

Nobody speaks for the remaining steps. The sweep completes. The terminal shows the final 0xE4 error as expected. Ryo pulls up the full log.

Priya timestamps the log entry. Felix straightens up slowly.

Mira looks at Koko.

Ryo shakes his head slowly, carefully.

Mira is already modifying the sweep script.

Dani leans over the datasheet.

Priya updates the shared log with a new entry header.

Ryo looks at the chip housing.

Ryo pushes the script.

The multimeter needle moves in increments too small to see individually. The I2C traffic monitor shows clean clock pulses.

3.1 millivolts. The harness log shows 4987 milliseconds, same as before. 3.4 millivolts. 4991. A slight increase, the chip is less interested. 3.6 millivolts. Back to 4987.

3.8 millivolts. 4971. Felix makes a sound. 3.9 millivolts. 4943. 4.0 millivolts. 4901.

4.1 millivolts. 4820. 4.2 millivolts. 4634.

4.3 millivolts. 4401. 4.35 millivolts. 3887. 4.4 millivolts. 2203.

Everyone inhales.

4.42 millivolts.

The terminal clears. A new line appears.

Priya's hand comes down on the timestamp key. A new error code.

Mira pulls the harness script up immediately.

Ryo adds three lines to the script.

Felix is already at the wall sheet with the marker.

Dani opens a fresh assembly file.

Ryo has a terminal window open, cursor waiting.

Priya is watching the donor forum. The Ljubljana contributor and someone from the TU bioelectronics group are both in the thread, reading in real time.

The lab goes quiet as Koko begins to type.

Ryo has his finger over the enter key on the log monitor.

Nobody speaks for three full seconds. Ryo breaks first.

Priya is typing rapidly.

The analog carrier hums at 4.42 millivolts. Felix caps the marker.

Mira pulls up a new file immediately.

Dani opens the sine wave approximation from the Ljubljana paper.

Priya watches the forum. The Ljubljana contributor and now someone from the TU bioelectronics group are both in the thread, reading in real time. Dani looks at the chip housing.

Mira sits back.

Ryo looks at the MCP4725 half-assembled on the breadboard.

Felix writes on the wall sheet: "a-type = READ ONLY from software. Physical input only via HAL_ABUS."

Priya relays to the forum without being asked. The TU bioelectronics group representative replies within thirty seconds.

The room absorbs that. Felix looks at Koko.

An hour later. The breadboard has been tidied, the jumper wires routed cleanly along the edge. The MCP4725 sits connected between the Raspberry Pi and the chip's interface pins, a small level shifter on a daughter board between them. Felix has taped the schematic to the wall beside the register type columns.

Ryo runs the DAC calibration script. The multimeter reads 4.42 millivolts steady on VOUT.

Mira reads it.

Dani checks the level shifter connections one more time.

Ryo is already at the Pi.

Dani exhales.

The room goes completely silent. Priya's hand hits the timestamp key. Felix turns to the wall sheet and writes in large letters: 65Hz - FULL_EXEC. Mira stares at the terminal.

Dani looks at the chip housing in its container.

Nobody speaks for a moment. The donation counter ticks upward. Then again. Priya has not posted anything yet. The Ljubljana contributor and the TU group are both watching the commit log in real time. Felix looks at Koko.

Ryo looks up from the Pi.

Mira shakes her head slowly, thinking out loud.

Dani leans forward.

Felix lowers his phone.

Priya looks at the lab budget spreadsheet.

Dani looks at the chip housing.

Ryo sends a message to the TU group without being asked. Felix looks around the room.

Ryo returns to the terminal. Mira reopens the harness script. The opcode spreadsheet is still there from before, half-populated, waiting.

Felix writes "SPARC - tomorrow" on the wall sheet below the 65Hz note.

Dani carefully updates the deviation table with the new findings: IPC confirmed, HAL_ABUS optimal window 65Hz, a-type registers physical input only, channel 0x003 active.

Priya posts the update to the forum. Within two minutes there are eleven responses. The donation counter moves.

The chip housing sits in its container on the central table, closer to Koko than to anyone else. The room sounds like work again: keystrokes, a chair shifting, Ryo reading something quietly to himself.

The SPARCstation 20 arrives on a Tuesday morning in a cardboard box too large for it, packed with a decade of someone's archived printouts used as padding. Koko sets it on the central table next to the specimen container with the chip housing. Felix tears the box open: the machine is the color of old teeth, that institutional beige of 1990s Sun Microsystems hardware, with its case intact, fan grille dusty but uncracked, and a faded sticker on the front reading "EIGENTUM DES INSTITUTS FÜR PHYSIK".

Mira plugs it in and the machine POSTs, its fan spinning up to a pitch slightly higher than expected as Solaris boots to a login prompt with a blinking cursor. Ryo pulls up a chair while Dani finds a spare monitor cable and Priya opens the SPARCstation 20 hardware reference, as Felix already examines the capacitors through the open case panel with a magnifying loupe.

Koko finds the flat-head screwdriver in the equipment drawer, the good one with the red handle that everyone uses and nobody puts back. The SPARCstation 20 opens differently from modern hardware. Four Phillips screws on the back, then the top panel slides forward and lifts off as a single piece. It takes a moment to figure out the direction.

Inside the machine, the motherboard is clean by the standards of a thirty-year-old computer, no visible corrosion, no catastrophic leakage. The layout is dense and logical in the way 1990s engineering often was, everything serving a purpose, nothing wasted. Two SIMM slots, both populated. A SCSI ribbon cable running to a hard drive that is probably the original. The CPU module, a hyperSPARC, sits under a heatsink held in place by a spring clip.

Felix moves the loupe slowly across the board, starting at the power regulation cluster.

Ryo is already searching the component database.

Priya adds it to the order queue without being asked. The machine sits open on the table, its thirty-year interior visible to the afternoon light coming through the lab's single large window.

They work through the board methodically, Koko placing the ESR leads, everyone else watching the meter and logging the readings. Ryo writes each result in the shared document in real time.

Priya updates the component order. Seven capacitors, values and ratings from the datasheet. Then Koko moves to the NVRAM chip. Finds the bent pin. Looks at it for a moment. Finds the smallest flat-head in the drawer, the one with the yellow handle. Very carefully, applies pressure to the pin.

A sound that is not loud but is very final. The pin does not bend. It separates from the body of the chip cleanly, as if it had been waiting for permission.

The broken pin sits on the motherboard, a small silver fragment next to the chip body. Priya adds the M48T08 to the component order and marks it urgent.

They bring protective equipment and a soldering iron to the bench. Dani clears the space and lays down an antistatic mat. The iron, a temperature-controlled station, the tip clean from the last session, heats to 320 degrees Celsius. Mira prepares the replacement capacitors on a strip of antistatic foam, each with its value written in marker on a small piece of tape beside it. Priya photographs the board before anything is touched, three angles, close-up on each capacitor position.

The first capacitor, the dead one in the power regulation cluster, comes off cleanly. The pad is intact. Good. The second takes longer, flux-bonded, shifting slightly before releasing. The pad holds. The CPU cluster capacitor reveals a faint shadow of corrosion on the pad beneath, not structural damage but the beginning of one.

They work through all seven. The fume extractor hums steadily. The lab smells of flux and hot metal. Ryo documents each completed joint in the shared log.

The board sits on the antistatic mat, seven fresh solder joints slightly shinier than the old ones, the new work identifiable from the old. Two days until Thursday.

Ryo pulls up a chair to the main workstation and opens a browser. Mira sits beside him.

Koko opens archive.org. The results come back immediately. Two relevant entries at the top.

Felix looks up from the hardware reference at the QNX suggestion.

Everyone looks at Priya. Priya manages the lab's hardware inventory.

Koko retrieves the Pentium 4 from the storage room, connects it to a spare monitor, and presses the power button.

The fans spin up at full speed immediately. Not the gradual spool-up of a normal POST. All of them, full speed, the sound of a machine trying very hard. The monitor flickers. A solid red screen appears, filling the entire display, white text in a font that is clearly not a standard BIOS output but a manufacturer's embedded diagnostic firmware.

The side of the case swings open on a hinge stiff from the dent warping the frame.

The heatsink is present but sitting at a visible angle, one of its four push-pin clips broken, the others holding it loosely. The thermal paste between the heatsink base and the CPU lid has dried to a grey powder and cracked in several places, the gap visible to the naked eye. The GPU slot is worse. The card is still in the slot, a mid-range card from around 2005, but the connector is visibly cracked along the bottom edge, one corner pushed downward at an angle the slot was not designed to accommodate. The RAM slots reveal a bent contact pin in slot A, and slots C and D hold no DIMMs at all.

Dani examines slot A under the magnifying loupe, then very carefully uses a mechanical pencil with the lead retracted as a probe, applying pressure in small increments against the bent contact pin, checking the angle after each pass. Forty seconds of absolute stillness in the lab.

Priya places the order and updates the budget log. Felix looks at the two open machines, the SPARCstation with its fresh capacitors and broken NVRAM, the Pentium 4 with its dented panel and rebuilt RAM slot.

Then, without warning, a small explosion from the SPARCstation occurs at the end of the bench. The sharp crack immediately comes with an acrid and chemical smell. A thin curl of grey smoke rises from inside the open case, from the power supply unit at the back right corner. Felix moves first, reaching past Koko and pulling the power cable from the wall socket in one motion. The fans spin down. The smoke thins and stops.

Priya places the PSU order with the UK seller and updates the budget log.

Nobody speaks for a moment. The wall sheet is visible from where everyone is sitting.

On Thursday, the NVRAM chip arrives in a small antistatic bag inside a padded envelope from a components distributor in the Netherlands. The PSU from the UK seller reaches the university mailroom on Friday. On Saturday morning, Ryo and Mira take the U-Bahn to the Altona electronics market and return with a Socket 478 cooler, two sticks of DDR400 RAM, a replacement PSU for the Pentium 4, and a bag of assorted thermal paste syringes a vendor threw in for free.

Later, on Monday morning, both machines are open on the bench. The reassembly takes most of the morning. PSU first, then CPU module reseated on the SPARCstation after full isopropyl cleaning, then a second ESR pass on all seven new capacitors confirming they survived the venting event, then the M48T08 seated carefully into its socket. On the Pentium 4: new PSU, new cooler with fresh thermal paste, two DDR400 sticks in slots B and D, the Radeon 9250 slotted into the graphics port.

After several hours of work, both machines are ready.

Koko pulls the QNX 6.4.1 live CD from its sleeve. Burned two days ago from an archive image, verified, labeled in Priya's handwriting: QNX 6.4.1 LIVE - CORTEX PROJECT. Felix opens the Pentium 4's optical drive bay. Koko presses the power button.

No red screen this time. The fans spin up at a normal pace. A single clean POST beep. The BIOS splash appears, then the boot sequence begins reading from the optical drive.

A loading bar appears, QNX's distinctive boot progress, minimal, grey on black, moving steadily. The desktop appears: QNX's Photon microGUI, lightweight, loading in seconds on hardware this old.

Meanwhile Ryo moves to the SPARCstation and powers it on. The Sun logo appears on screen, white on black. The POST runs and then OpenBoot's prompt appears.

Mira exhales. Felix reads the memory line: 64 megabytes, both SIMMs reading correctly. Ryo types at the OpenBoot prompt.

The text-based installer. Koko takes the keyboard.

A short pause. Everyone reads the summary. Mira nods. Dani says press it. Koko presses F2.

The SPARCstation reboots from its own drive for the first time under their ownership. The Sun logo appears. OpenBoot hands control to the Solaris kernel. Boot messages scroll past, devices attaching, file systems mounting, services starting.

The room reorganizes itself around the central table. Priya closes the forum tab without being asked. Dani moves two laptops off the table. Ryo wheels the DAC interface from its corner position to the table beside the SPARCstation. Felix spreads an antistatic mat across the cleared center of the table. Then everyone looks at the specimen container.

Felix pulls on nitrile gloves. He hands a pair to Koko without speaking.

Dani checks each wire connection against the diagram taped to the inside of the case. All 96 SBus lines accounted for. The DAC is running. The multimeter reads 4.42 millivolts on VOUT. 65 hertz on the frequency generator.

Koko lifts the specimen container and sets it on the antistatic mat. Opens the lid. The chip housing is smaller than it looks in photographs, roughly the size of a large thumbnail, matte grey, its pin array visible on the underside. Priya photographs it before anything else is touched. Koko places it in the interface cradle Ryo fabricated from a ZIF socket base and two strips of brass shimstock. The cradle holds it without pressure, contact made through the shimstock touching the perimeter pins.

Koko connects the final wire, the analog carrier line from the DAC output to the chip's HAL_ABUS pin, a single fine-gauge wire terminated in a micro-clip. The clip seats. The multimeter on the carrier line reads 4.42 millivolts.

The root prompt blinks on cortex-sparc. The donation counter ticks up on Priya's screen. Someone is watching the commit log right now.

The oscilloscope shows a response on HAL_ABUS. A brief perturbation in the carrier signal.

Koko writes a second test with three SAVE and three RESTORE calls. The oscilloscope shows six perturbations, three rising on SAVE, three in the opposite direction on RESTORE. But the spacing between them is irregular. Ryo stares at the timing display.

Ryo turns and looks at the Pentium 4.

Koko takes the chair at the Pentium 4. Ryo pulls a second chair beside it. They write the measurement harness together, Ryo dictating structure, Koko typing, both arguing over specifics.

Dani holds out the comparator output wire. Koko takes it and connects the comparator output to LPT STATUS pin 13. A small brass hook terminal, a single clip, seated.

Koko starts the measurement harness on QNX. Ryo runs the SAVE sequence on the SPARCstation simultaneously.

The room reads the log in silence.

The forum posts. Priya watches the replies come in. Then one from the Ljubljana contributor.

Nobody moves for a moment.

Felix uncaps the red marker and writes beneath 65Hz - FULL_EXEC:

106μs - SPARC WINDOW = LFP RESPONSE

The chip housing sits in its cradle, the carrier humming at 4.42 millivolts, 65 hertz, connected to both machines. The lab is three months old and now it feels even older.

The paper takes eleven days of intense writing and collaborative debates. Initially divided on where to submit, the argument is resolved when Priya secures TU bioelectronics and Ljubljana researchers as co-authors, shifting the focus to a joint submission to Frontiers in Neuroscience. Koko drafts the technical section, Ryo refines the methodology, and Mira handles the results, while Dani fights in tracked changes for three days to keep the word "deliberate" in the description of the Institute's design choices. The TU group provides the biological correlation linking the 106-microsecond window to LFP data, and after five drafts from Felix, the non-technical summary is finalized.

Two weeks after their first debate, the paper is finally ready. The lab gathers at the central table, surrounded by the specimen container and a wall sheet covered in data, as Priya opens the submission portal to enter the final details.

She submits.

Three days after submission, Frontiers acknowledges receipt and assigns reviewers. The same day, Priya pushes the repository to the lab's self-hosted GitLab instance.

Priya posts the repository link to the forum. Within four hours, eleven people have cloned it. Within a day, the first external issue is opened, filed by a username that reads only "generated_22", the same person who donated everything they had saved that month.

Koko reads it. The chip housing sits in its container on the central table. The wall sheet reads 65Hz - FULL_EXEC and 106μs - SPARC WINDOW = LFP RESPONSE. Koko opens the issue and types the first reply.

It's late evening in Xue's room, where the Cortex Project forum is open in one browser tab. A thread titled "Institute just dropped something" has already gathered forty-seven replies in two hours. Clicking the link redirects him to YouTube, where the official Institute of Genetics channel, dormant for eleven months, uploaded a video forty minutes ago under a thumbnail showing a stage, blue lighting, and a lectern with a double helix emblem.

Xue turns the volume up. Pulls the hoodie tighter. The screen fills with the stage.

The venue is smaller than the original Symposium hall, more controlled, the audience visible but selected. Researchers in Institute lanyards, a row of journalists with press credentials, a camera crew that is clearly professional, are all present. The lighting is warmer than the institutional blue. The stage has a curved translucent backdrop that shifts color slowly. There is a glass of water on the lectern with empty notes. The presenter, Dr. Rainer, walks on.

The presenter lets the analogy sit for a moment. In the selected audience, a journalist in the third row writes something down. A researcher near the aisle crosses his arms, evaluating. A woman near the front with a press lanyard glances at the person beside her. The translucent backdrop behind the presenter shifts from a cool blue to a slightly warmer tone.

The journalist in the third row has stopped writing. A researcher near the back of the room tilts his head, the gesture of someone filing something away. Two people in the front row exchange a glance that the camera catches for half a second before cutting away. The press row is very still.

A woman in a grey blazer near the center of the press row writes a single word and underlines it. A researcher two rows back shifts in his seat. Near the aisle, a man in an Institute lanyard nods once.

A researcher in the second row looks down at his hands briefly, then back up. The woman in the grey blazer has stopped writing. The camera stays on the presenter for two full seconds, then cuts to the wide shot. The man in the Institute lanyard who nodded before is no longer nodding. A journalist near the end of the press row has his phone angled slightly upward, recording something on the laptop beside him. The camera catches it for less than a second.

Xue's cursor moves to the pause button and stops there without clicking.

The camera holds on the presenter. In the room, a journalist in the press row sets her pen down on her notepad. A researcher near the center of the audience looks at the ceiling for a moment, then back at the stage. The man in the Institute lanyard near the aisle is very still. Near the back of the room, the woman with the researcher's badge is now sitting fully upright, eyes on the stage.

The camera cuts to the wide shot. Six rows of people, the press row across the front, the stage behind them with its drifting backdrop and its single presenter and Institute emblem. Nobody in the room says anything.

The presenter steps back from the lectern. The applause from the selected audience is measured and brief. The production lighting fades the backdrop to the Institute's deep blue. The double helix emblem on the lectern catches the light one final time. The camera holds on the empty stage for three seconds after the presenter exits. Then slowly fades to black.

The Institute of Genetics official channel endcard appears and the video ends.

The screen returns to the YouTube page. The recommended column on the right is already full of responses, reaction videos loading in real time. A legal analyst, a tech ethics channel and someone recording from their phone with the caption "I cannot believe they said this", and a generated human, nineteen years old by the look of them, all appear in the feed.

Xue opens the comment section.

Xue reads without scrolling for a while. Then scrolls and stops at the one with the highest number of likes.

Xue closes the laptop halfway and the room goes darker. The forum tab is still open in the background and the thread titled "Institute just dropped something" now has two hundred and fourteen replies.

Alberta has not messaged in four days.

Xue scrolls past three reaction videos and a legal analysis thread. Then something in the sidebar, a link shared in the Cortex Project forum thread. The domain is fsf.org. The page loads in under a second, minimal, no JavaScript, the way FSF pages always load.

Xue reads it to the end and then opens a new tab, navigates to the Open Cortex letter, moves the cursor to the signature field, types a name and then submits.

The protest has been building since the Little Symposium video, growing from forum threads and petitions over three weeks into a crowd of several hundred gathered on a Saturday afternoon in central Hamburg, on the pavement between the Hauptbahnhof and the Binnenalster. Generated humans make up the larger part of the gathering, joined by university students, researchers who signed the Open Cortex letter in their own names, and a few journalists tracking the fallout.

Handwritten banners rise above the crowd: "OPEN CORTEX," "PUBLIC MONEY PUBLIC CODE," "MY BRAIN MY CODE," and "PRINTED DOES NOT MEAN PROPERTY." One simple cardboard sign in careful block letters reads: "I AM NOT A USAGE LIMIT." Near the left edge, a few rows back from the front, Alberta checks her phone, spots Xue approaching, and quietly slips the device into her pocket.

At the front of the crowd, a woman with a KDE contributor badge steps up to the PA microphone.

Sustained applause. She steps back. A man in a freedesktop.org shirt moves forward, reading from his phone.

Someone near the front raises a fist. Several people around them follow. A generated human, perhaps twenty, steps up without introduction.

The crowd is very still. Then the applause comes, longer than anything before it.

A woman near the barrier, older, steps to the microphone.

A pause in the crowd that is very full. A younger man steps up beside her, translating into German for a section of the crowd gathered on the left. Then a group moves forward together, perhaps fifteen people, wearing green and carrying a banner that reads SYSTEMS COLLAPSE TOGETHER, a Fridays for Future regional flag visible among them.

Applause from across the crowd, genuine and wide. Behind the Fridays for Future group, a small cluster of people hold a hand-painted sign: OPEN SOURCE THE EARTH, with a small chip drawn next to a seedling.

A generated human near the PA, perhaps eighteen, steps up without having planned to, visibly nervous.

The crowd moves slightly closer around them.

A group in matching grey hoodies moves to the front, the GNOME Foundation logo on their sleeves.

From the left side of the crowd, a banner advances: CHAOS COMPUTER CLUB HAMBURG. Three people with CCC lanyards reach the front. The one with the microphone speaks without notes.

Extended applause. Someone near the PA shouts something in agreement. A quieter group moves forward next, carrying a banner: DIGITALCOURAGE E.V.

From the back of the crowd, a ripple of movement as a large contingent arrives, Extinction Rebellion flags, red circles, and among them the Fridays for Future Hamburg group merging with another from Bremen, the two groups joining their signs together.

A generated human in the crowd, somewhere in the middle, starts a chant that spreads outward in both directions.

It runs through the crowd twice, then three times, then settles into a low sustained repetition that the PA does not need to amplify.

A woman steps to the microphone, switching between German and English mid-sentence.

She pauses.

The crowd is very quiet for a moment. Then a Polish-language sign rises near the back, held by Tomasz from Siedlischte, visible now in the crowd: WOLNOŚĆ NIE JEST PATENTEM. Freedom is not a patent. Another goes up beside it, hand-lettered in Romanian: LIBERTATEA NU SE LICENȚIAZĂ. Freedom does not require a license.

The chant from before returns, softer, and in more languages.

The complaint is filed on a Tuesday morning, forty-three days after the Little Symposium video, thirty-one days after the protest, and eighteen days after the Cortex Project paper received its first peer review comments. Received by the Verwaltungsgericht Hamburg at 09:14, the 247-page document is assigned to a chamber by 09:47 under case number VG Hamburg 7 K 2891/26, leaked to journalists by 11:30, posted to the project forum by 14:00, and spread everywhere by evening.

The document continues for a further 189 pages, comprising annexes, supporting declarations, technical expert reports from the Cortex Project team, and the certified transcript of the clinical psychologist's public broadcast. The Institute of Genetics has 21 days to file a response.

Three weeks later. The administrative court schedules a preliminary hearing on the interim injunction application. The chamber is small, wood-paneled, two long tables facing each other, a raised bench for the presiding judge. A court reporter. A clerk. No public gallery. Press wait in the corridor outside.

The judge is Dr. Petra Vasilescu, appointed to the administrative chamber eight years ago, known in Hamburg legal circles for asking questions that lawyers find difficult. Both parties are seated.

A pause at the defendant's table.

She writes something.

Another pause at the defendant's table. Dr. Krell writes something and slides it to Dr. Bormann. He reads it.

The corridor outside. The press are still there. The scientific workers' union representative, attending as an observer, steps outside during the recess and says nothing to anyone. She looks at her phone. She types something. Inside the chamber, the defendant's table is in quiet, intense conversation. The plaintiffs' table is very still. Dr. Scholz reviews a document. The generated human plaintiff sitting beside her, one of the forty-seven named in Annex A, looks at the wood-paneled wall.

The recess ends.

A sound from the plaintiffs' table, quickly contained.

She closes her folder.

She stands, everyone stands and the hearing adjourns.

The full preliminary hearing takes place over three days in November. The court hears expert testimony from the Cortex Project team, from the clinical psychologist, from a constitutional law professor at the University of Hamburg, and from two technical experts retained by the defendant, both of whom concede under cross-examination that the IPC instruction's function cannot be fully characterized without the complete ISA. On the fourth day, Judge Vasilescu reads her ruling.

The ruling is published at 11:43 on a Wednesday morning. By 11:51, the Cortex Project forum thread "They blinked" has four hundred new replies. By 12:15, the Institute of Genetics issues a press release.

The press release lands in the Cortex Project forum at 12:19. For approximately ninety seconds, nobody responds. Then seventeen people type at the same time.

The traditional memory injection protocol is not new. It predates the chip by several years. It works by introducing synthetically encoded RNA sequences into hippocampal tissue during the accelerated maturation process, consolidating specific memory structures without external hardware. It is less precise than the chip. It is also entirely biological, entirely undetectable by any current scanning method.

The hot-swap extraction kit is real. It ships six weeks later in a white box with the Institute's double helix emblem on the lid. The documentation is thorough and accurate. Several generated humans in the Siedlischte community receive theirs on the same day and open them together in the IT lab, laying the components on the antistatic mat, reading the instructions with a lot of attention.

The ISA sale is announced six weeks after the press release, before the court's sixty-day disclosure deadline expires. The announcement comes not from the Institute but from a financial news outlet: the Institute of Genetics has retained an investment bank to conduct a private auction of the complete ISA, all associated patents, and all documentation, including the HAL_ABUS specification that has never been publicly released. Four parties are reported to be in contention.

The court, informed of the ISA sale, schedules an emergency hearing. Judge Vasilescu's question, submitted in writing to both parties before the hearing date, is brief. It reads: Does the court's disclosure order apply to the current holder of the ISA, whoever that may be at the time of compliance?

In the IT lab at the University of Siedlischte, the wall sheet still reads 65Hz - FULL_EXEC and 106μs - SPARC WINDOW = LFP RESPONSE. Koko adds a third line, lower down, in a different color marker borrowed from Dani.

WHOEVER BUYS IT STILL HAS TO ANSWER TO THE COURT.

The news breaks on a Thursday at 07:23, before most of the lab has arrived. Priya is first in, as always, and she is the one who finds it: a Reuters financial wire story, picked up by The Register twenty minutes later, confirmed by Intel's investor relations page at 07:41.

By the time Felix arrives at 08:15, Priya has printed the Reuters article and taped it to the wall sheet below Koko's handwritten line about the court order. By the time Mira and Ryo arrive together at 08:30, the forum thread has been running for two hours and has three distinct camps: people who are relieved it is not Microsoft, people who think Intel is worse, and a small but growing group of hardware engineers who have been reading the acquisition terms and are asking a very specific question about what exactly Intel bought. Dani arrives last, at 08:47, reads the article on the wall, reads the forum thread on their phone, and sits down without speaking as the wall sheet now reads:

65Hz - FULL_EXEC
106μs - SPARC WINDOW = LFP RESPONSE
WHOEVER BUYS IT STILL HAS TO ANSWER TO THE COURT.
Reuters, 07:23: INTEL ACQUIRES NEURAL INTERFACE ISA IN €340M DEAL

The chip housing is still in its specimen container on the central table. The red cursor blinks on Koko's screen.

Ryo writes on the wall sheet below the Reuters line:

PROJECT OPENCORTEX - Free ISA, clean-room, coreboot target

Ryo adds to the wall sheet.

v0: NASM macro layer over x86-64 SDM
v1: LLVM backend, OpenCortex target
v2: coreboot payload, full stack

The clean-room declaration is committed at 09:14, dated, signed by all present, pushed to the repository before anything else touches it. Team A is the research team, everyone who touched the chip, the oscilloscope, and the DAC. Team B is the implementation team: Ljubljana, TU bioelectronics, anyone new from the forum. Koko opens the first source file. The local LLM instance, a quantized Llama model running on the lab's fastest AntiX workstation, loads its context: the Cortex Project repository, the coreboot payload API documentation, and the OpenCortex behavioral specification draft Priya committed eleven minutes ago.

The file is opencortex_payload.c. The cursor blinks on line one.

Koko moves to the next function. The LLM autocomplete flickers.

The LLM autocomplete offers the main function.

The first commit goes in at 10:47. Fourteen minutes later, the Ljubljana contributor opens the first pull request from Team B. It is eleven lines, a correction to the IPC parameter struct padding based on their biological signal timing data. The wall sheet gets one more line, added by Priya in her smallest handwriting:

opencortex: initial commit 10:47

Ryo has been running the payload in QEMU while Koko and Felix worked on the IPC stub. He runs a test binary that skips the HAL probe and goes straight to an IPC call.

Koko opens oc_ud_handler.c.

The LLM offers the IDT installation function.

Koko accepts. Commits. Pushes. Second QEMU run, with the handler installed and the NOP test.

A pause in the room.

He commits the response as RECONCILIATION.md in the repository root.

Koko adds it. The LLM autocomplete, mostly silent for the last hour, offers one final suggestion as Koko closes the preamble section.

Koko looks at it for a moment and then accepts it without modification. The commit message reads: spec: add reconciliation policy and stability framework

On a quiet Saturday in Hamburg, Xue stands outside the U-Bahn exit at Jungfernstieg at eleven in the morning, hands in his jacket pockets, watching pigeons peck at a dropped pretzel. Alberta arrives four minutes late, wearing a slightly too-large dark green cardigan instead of her usual lavender one, eating an apple from her tote bag as she scans the plaza for a rubbish bin.

Alberta walks together with Xue at a leisurely pace. She carries the tote bag on her right shoulder. They cross into Neustadt. The streets get slightly narrower, the architecture a little older, the Saturday morning foot traffic thinning out as they get further away from the city center. Then, on the side of a building at the corner of a block undergoing renovation, mounted on scaffolding high enough to be visible from three streets away, a billboard appears:

The two of them stop walking at approximately the same moment.

Directly beneath the billboard there is a narrow retail space with a hand-painted sign above the door reading SERVERHAUS HAMBURG - KOMPONENTEN & GEBRAUCHTE HARDWARE. The window is dense with products: rackmount units, cables coiled on hooks, a shelf of used SSDs in antistatic bags, a Raspberry Pi display stand. And in the front right corner of the window, propped against a used server chassis with a dent in its left panel, there is an Intel Neural Compute Architecture promotional card with the same blue as the billboard and aggressive typeface.

A group of three passes on the pavement behind them, two in conversation, the third glancing at the shop window and halting the group. The first two look at the window, and among them, a generated human in a yellow jacket from the group reads the Intel promotional card in the window.

The shop is closed. Saturday morning hours, a handwritten sign on the door: 11:00 - 16:00. It is 11:24. The generated human in the yellow jacket reaches the door first, tries to pull handle without really meaning to, and it opens.

Everyone looks at each other. The door is pushed open six centimetres. The smell of dust and old electronics comes out first. Then someone at the back of the group steps forward and pushes the door fully open. Glass crunches faintly under the door's arc. Then they all go in one by one.

The shop is narrow and goes further back than the window suggested. The front section is intact, but three metres in, a shelf unit on the left has partially collapsed, components scattered across the floor, bubble wrap and antistatic bags split open. A monitor on the service counter is dark but intact. The NCA units in the back sit in open shipping crates.

The generated human in the yellow jacket crouches near the collapsed shelf, looking at something on the floor without touching it.

Lena crosses the room immediately. Two others follow. Alberta comes last, standing at the edge of the group forming around the server drawer.

The CPU module is visible through the chassis venting, a flat matte-grey package seated in a proprietary socket, considerably larger than a standard desktop processor but much smaller than a full server chip. Around its perimeter, a ring of fine-pitch connectors that do not correspond to any standard socket format. Along one edge, barely legible on the substrate in small characters: NCA / HAL-COMPATIBLE.

Lena takes out her phone and photographs it.

Lena holds her phone light over it. The module sits in Xue's palm, lighter than expected. The top surface is a polished metal plate, an integrated heat spreader^[36], but the alloy is wrong for a standard IHS. Too light. A slightly warm silver rather than copper-tinted. And completely blank. No Intel logo. No product marking. No serial number visible on the top face. Everyone is looking at it.

On the underside, the fine contact array with the pins. And along one edge, in characters small enough to require Lena's phone light to read:

NCA · HAL_ABUS READY · 65Hz · 4.42mV · CORTEX-COMPATIBLE

Nobody knows. Xue is still holding the module, the blank edge with its four analog pads facing upward in the phone light. The street outside passes normally. The ULTRA-RESISTANT billboard is still visible through the shop window, framed by the door left open.

The group reorganizes itself. The quiet person takes the shipping crates. The thermal person photographs the chassis spec plates one by one. Two other two document the socket from every angle, using each other's phone lights. Alberta takes the NCA module from Xue carefully and holds it steady while Lena photographs the underside contact array in sections, overlapping frames.

Lena lowers her phone and looks at the group. A few nods, a shrug from the thermal person, the quiet one already uploading photos.

The group waits for the upload to finish. Someone finds a relatively clean spot on the counter to lean against. The thermal person reads the chassis spec plate one more time, quietly, to themselves. Alberta carefully reinserts the NCA module into its socket with a soft click, pushes the server drawer back in and then looks again at Xue.

Lena's upload completes. She pockets her phone and then they all exit the shop and make sure the door is closed.

The group walks from Neustadt to Altona. Jonas has already pulled up the Cortex Project forum on his phone and is reading the first replies to the upload.

Alberta is beside Xue and she has moved the tote bag to her left shoulder. The street opens as they cross into Altona-Altstadt, the architecture becoming older and lower.

Alberta does not answer immediately. They walk for half a block.

They turn a corner and the street opens into a wider residential stretch, trees lining the pavement, the traffic quieter. Wohlerspark is a few minutes ahead, visible as a gap in the buildings where the green starts.

Everyone looks back the way they came.

She goes through the gate. The others follow one by one.

Wohlerspark used to be a cemetery, its last burial in 1945, converted to public green space in 1979. The linden alleys are old and high. The weathered gravestones remain along the edges, pale and slightly tilted.

Everyone raises their hand. Including Xue. Including Alberta, after a brief pause.

They reach the central clearing. The linden trees close overhead.

She sits down on the grass beside where Jonas is sitting. Lena leans back on the bench. Maya has her eyes closed, face tilted toward the sky through the linden trees. The forum replies keep coming. Jonas turns his phone face down on the grass in the afternoon.

Intel Deutschland GmbH, Munich. Monday, 07:43. The third-floor Schwabing office was designed to feel open and collaborative but currently feels like a pressure vessel, its floor-to-ceiling windows catching the morning light at the wrong angle for everyone sitting at screens. Nine people are in the room, four called in before six, gathered around a central table covered in printouts of a Hacker News thread with four thousand comments and growing, which has sat on the front page for thirty-one hours: "INTEL NCA IS A NEURAL INTERFACE CHIP BUILT ON A PROPRIETARY BRAIN ISA CURRENTLY SUBJECT TO A GERMAN COURT DISCLOSURE ORDER."

The head of European communications, Petra Haugen, stands at the window with her phone, having been on calls since five forty-five. She ends the current one and turns to the room.

A short pause. Someone at the back of the room makes a sound.

A man in a grey pullover, Stefan Brandt, senior regulatory affairs, stands up without being asked.

A younger woman near the door, Karen Fischer, communications, looks up from her laptop.

Haugen picks up her phone and then then puts it down.

The news read in the room:

The consortium announcement comes on a Thursday, seventeen days after the emergency injunction halted NCA deployment across the EU. It arrives as a joint press release, six corporate logos (Intel, Microsoft, Google DeepMind, IBM Research, SAP, Qualcomm) arranged in a row above the headline.

The consortium announcement lands on a Thursday. Stronk is incorporated on Friday. The founding document, filed with the Hamburg commercial register at 16:47, lists the entity type as eingetragene Genossenschaft^[45] (a registered cooperative) and gives the registered name as STRONK eG. The articles of association, posted publicly to a GitLab repository at 16:48, one minute after filing, begin with the following preamble:

The membership fee of one euro is paid, in the first twenty-four hours, by forty-one thousand, two hundred and fourteen people. By the end of the first week, two hundred and nineteen thousand members joined in the cooperative and two hundred and nineteen thousand euros have been paid. The GitLab repository for the articles of association has been forked six hundred times. Someone has added a CONTRIBUTING.md. Someone else has added a CODE_OF_CONDUCT.md. A third person has added an issue titled "should we have a logo" which has forty-seven comments.

The logo question is resolved on day four when a generated human graphic designer from Brno submits a pull request containing a simple geometric mark: a raised fist, but the fist is made of circuit traces. The PR description reads: "it is extremely on the nose. that is the point." It is merged within an hour.

In the IT lab at the University of Siedlischte, Koko is reading the Stronk articles of association on screen. The wall sheet behind them reads 65Hz - FULL_EXEC and 106μs - SPARC WINDOW = LFP RESPONSE and WHOEVER BUYS IT STILL HAS TO ANSWER TO THE COURT.

Felix reads over Koko's shoulder.

A pause in the lab.

One by one, everyone in the lab pays one euro. The counter moves seven times.

The legal question is real and nobody is pretending otherwise. Stronk has money but it does not yet have a mechanism to acquire the ISA as Intel has not offered to sell, the consortium is not yet incorporated and the court proceedings are ongoing. The CJEU^[47] referral has been accepted and will take between eighteen months and three years to resolve.

What Stronk has, besides money and a logo and a legally filed preamble that uses the word "consortiuming", is a lawyer. Her name is Dr. Amara Diallo, Brussels-based, specialising in EU competition law and intellectual property. She joined Stronk as a member on day one for one euro and posted a thread to the forum on day three.

Eight months after Stronk's incorporation, the Neural Compute Architecture Consortium formally grants membership to Stronk eG. The decision takes four board votes, two legal opinions, one emergency meeting called by Microsoft's representative, and a letter from the European Commission's Directorate-General for Competition noting that the consortium's membership criteria were under review as part of an ongoing Article 101 TFEU examination.

It's evening in Xue's room. The Microsoft BUILD^[50] 2036 keynote stream is open in one tab. It began forty minutes ago. The arena in Redmond is full. Brad Mercer, Microsoft's CEO since 2031, is walking across a stage the size of a tennis court, microphone in one hand, the other jabbing the air, jacket off, sleeves rolled up, the collar of his shirt visibly darker than it was at the start.

A ripple with few laughs and uncertain looks occurs through the audience.

The arena applauds. Xue's cursor hovers over the comment section.

Xue opens four more tabs. The other consortium CEOs have all made statements in the last ninety minutes, timed to the BUILD keynote.

Xue opens the chat.