For 30 advance/early chapters : p atreon.com/AutumnXd
It was dinnertime, and most of the base was in the cafeteria. The moment Ryan walked in, he became the center of everyone's attention.
The only ones who stayed relatively composed were the people from Reeves's five labs. They worked in Storm Bay every day now, and Ryan had selectively let them in on a little information beforehand. They'd long since made their peace with Crimson Typhoon.
"Ryan, is the new project really building a mech that big? I thought Scrapper was already enormous."
Ryan had just sat down with his dinner tray when a researcher beside him spoke up. Others were seated along the long table, and they all pricked up their ears. They wanted the answer to that question too.
Ryan glanced at them. Like Kyle, they were firefighting mech researchers, except this group handled the structural design side.
"The project's been publicly announced. You think it could be fake?"
At Ryan's confirmation, everyone unconsciously stopped chewing.
"A mech that big. How do you even design the structure to satisfy balance and stability."
The group considered it. Their firefighting mech alone had a structural design as tangled and complex as a knot. Scale a mech up tenfold, and how many researchers would they have to sacrifice to the gods to keep the thing from falling apart?
Never mind that the big mech also had to satisfy a seven-thousand-meter deep-submergence requirement. Beyond balance and stability, the overall compression resistance and seal integrity had to reach a level of perfection. Plus the life-support system, the power system.
The thought made them even more curious about the big mech's true appearance. If it could actually be built, it would be an industrial marvel in the history of mankind.
The longer they thought about it, the more they set down their chopsticks and drifted, unprompted, into discussing the structural design of a heavy mech. Some started sketching on the table with their fingers, simulating the mech's design. The occupational disease had clearly taken hold. Now and then they asked Ryan whether their conclusions were correct.
This was the everyday descent into obsession at the base.
Watching the scene, Ryan rapidly shoveled the rice out of his bowl and decided to get away from these obsessed people. All they knew was research, all day. They were boarding an express flight to a mental breakdown.
Back in his dorm, Ryan had just finished a shower when he remembered something.
He'd filmed the Storm arm during testing.
Now that the Crimson Typhoon project was public, it was time to release the footage. The timing would maximize the heat.
When he made a move, he moved like a seasoned blade. The craft of harvesting attention was use-it-or-lose-it. He couldn't afford to stop.
He opened his social media account, pulled up the previously filmed video with practiced ease, and gave it another look. Nothing needed to be censored. The entire video contained nothing but a massive mechanical arm and its outer armor. No extraneous objects.
Upload.
Post.
Then Ryan set the phone down, sat at his computer, and continued training the neural network, studying the artificial intelligence's base architecture along the way.
The deep neural network algorithm the system provided was, in functional terms, considerably more powerful than other neural networks of the same class.
In image recognition, it had already demonstrated astonishing efficiency and accuracy.
Earlier, Ryan had randomly grabbed a sample of videos from TikTok and shown them to the system's neural network. Without any instructions, the network had taught itself to recognize all kinds of objects: the animals that appeared in the videos, the humans, the furniture, and so on.
By now it could perfectly identify more than twenty kinds of animals commonly found online, whether in motion or static. By the latest results, the error rate had dropped to zero.
Ryan had also tried manually cropping out the midsection from several cat photos. Faced with these deeply confusing images, the system's neural network still accurately identified the subject as the midsection of a cat, not a leopard or a tiger. Images like that, sometimes even a human couldn't identify.
Various large companies in the industry had run similar experiments, but this was still an astonishing result.
The videos Ryan had shown the network numbered only around a hundred thousand, and the ones featuring cats or other animals were even fewer. Which was to say, the system's neural network had performed only a minuscule amount of training and could already perfectly identify a given object.
Other neural networks, to achieve anything close, needed to build an enormous database and an enormous network, then run at least hundreds of millions of training iterations.
The images and videos people uploaded to cloud storage, and the major livestreaming sites, all had a deep neural network hidden behind them, conducting scan-style "learning," monitoring video feeds and comment streams in real time to flag violating content.
But even after that much learning, most of the industry's neural network image recognition was still underwhelming, a long way from the system network's hundred-percent accuracy.
Of course, the system's neural network wasn't without a drawback. Or, perhaps, it didn't quite count as a drawback.
The longer it ran, the more server resources it demanded. His computer was already struggling to contain it.
"Time to get it a server." Ryan opened a bottle of cola, looked up server specifications online, and quickly settled on one of them.
He sent the link and the requirements to Patricia, then went hands-off, the carefree manager.
At that moment.
Danny Price, having returned to campus from home, was lying on his dorm bed, scrolling through the news on his phone.
In the room, his roommate Trey was practicing his fadeaway jumper without a ball, jumping up now and then to launch an invisible shot into the air.
"Hey, Price. You hear about that big mech thing lately?" Trey asked, taking a breather.
Danny lifted his head and glanced at him. "Of course I heard. I'm literally reading about it right now."
"Everybody online's saying a mech that big can't be built. Got me second-guessing it, honestly. You think a mech that size is even realistic?"
When Danny had first heard about the project, as a mech fan, he'd been thrilled. But the more he read the online chatter, the more he started to feel like maybe the project wasn't credible.
He looked at all the formulas and theorems people were listing, pronouncing the project dead on paper. One person saying it was fine. But a hundred people saying it carried some weight. Repeat a thing often enough and it starts to sound true.
And it wasn't just domestic. It was the same everywhere. Plenty of foreign university professors had stepped up, calculating the difficulty of building a mech this large through various methods, and they all reached the same conclusion: nearly impossible.
Trey started up the fadeaway motion again. "Course it's not realistic. Use your head. How could anybody build a mech that big right now?"
But his words went a long time without a response from Danny. He stopped the jump shot and looked over, puzzled.
Danny was staring at his phone, slack-jawed.
"Price? What are you looking at?"
Danny turned the phone around to show him the screen. "Who said a big mech can't be built? Look at this."
On the screen, a massive red mechanical arm was smoothly going through a fist-clenching motion.
A glance at the account that posted the video. The Mech Expert.
In an instant, Trey was slack-jawed too.