The first hell box was created in 2032 as an art project. This was in the tail end of the second flowering of AI technology— the false one that started in the 2010s, back when they thought mathematically modeling a neuron was important but didn't yet know how to do it— and the very start of practical microfluidics technology. Delgado's box had no gates. Microfluidics research at the time focused on the idea a microfluidic cube's 3D structure meant circuits could be packed and routed more efficiently than in 2D photolithography, but that wasn't the interesting part to Delgado. Delgado wanted to simulate nerves, and she concluded since biological brains were 3D her simulation should also be 3D. There wasn't any scientific thinking to the decision. She just thought it was poetic.
Delgado's box was a 4-centimeter cube of mineral oil. Its functional resolution was lower than it could have been at the time, due to the medium and the use of already-outdated Gantz agitation to kickstart the flow. Delgado had no budget, was not a microprocessor engineer by trade and had no background in contemporary machine learning. All she had done was develop a model of a nerve cell and write down the dynamics of that cell when it was signaling pain, then reconstruct those dynamics in microfluidics. She packed the fluid parcels simulating her nerves as tightly as she could in the cube, and once the loop had started it was stable. As long as she kept pumping in kinetic energy, every millimeter of the box would move in a way replicating a nerve in continuous pain, just with fluid flow instead of electric and chemical effects. It wasn't alive. It couldn't think. It didn't do anything. All it did was hurt. She called it a hell box. She filmed it running for four minutes and posted it on the Internet along with the instructions for her setup. It got very little attention at the time.
Given the specifications she posted, you can calculate the box would have kept running on inertia for about two minutes after she turned it off, depending on the mineral oil's purity.
Microfluidics engineers remembered Delgado's box. For the next few hundred years, it was a tradition when testing out a new flow-circuit medium to replicate Delgado's box at whatever maximum resolution your tech could handle. It was like a hello world program, a red triangle. A proof of concept. A joke. by 2100 they had proper microfluidic neuron simulation, but the hell box was just so much simpler, so unambiguous, so visceral. You obviously can't fill a square kilometer with proper brain and have it do anything coherent. But an FCE can implement a packing with Delgado's model in half an hour, and confirm a homogenous execution in half that or less. So as a functional test, an installation, an amateur learning project. It remained common.
So when the Sirius experiment entered its shakedown phase, there was no hell box configuration on the test schedule. But there didn't need to be. All it would have taken is some unmonitored engineers running a thrown-together program during downtime hours, and since the event did happen during Kenes Station's sleep shift, this must have been what happened. The experiment's initial goal, inducing structured flow in a star's convection zone, had been achieved ten years prior and the project had just achieved its first successes with star-scale microfluidic circuits. Sirius is a fairly stable star and very energetic, so there was no need for any external source of energy. Once a circuit was induced it stayed stable on its own until the controller forcibly induced a new one. Self-powering, self-reinforcing. A computer twice the size of the Sun, or it would have been, with another year of work.
But instead someone turned it into a hell box. Three to five cubic gigameters of carbon and oxygen in motion, every single micrometer in pain. They probably only meant to leave it on for a minute or so. But then the spiders started crawling out, legs the width of Saturn, and then the bigger worse things. Most of the controller grid probably stayed functional for most of an hour, but it didn't matter, because Kenes station was destroyed in the first wave of radiation and chitin and there was no one there to operate. By the time anyone understood what was happening it was twenty to thirty hours later, an Emergence had already torn apart Sirius B and Earth evacuation was inevitable due to the gamma ray burst, not that anyone fifty years into the billion-year lifespan of this particular hell box would consider a gamma burst from a collapsing white dwarf the worst thing that can happen to a planetary neighborhood. That's one of the things we've learned, along with the lesson, first and foremost, that there's only so much pain you can get away with putting in a single place.
