FLICKER — NaNoWriMo 2016
NOVEMBER 10TH, 2016 — DAY 10
Short again. But I’m getting over caring about it. I was going to try and push to at least not open it more but I’m flaking hard and writing shit(ter).
“Do what you want, kid. This whole thing’s a fucking mess.” Nedaara whipped a leg over her bike and paced off to join the group.
“This is an opportunity!” Keiko called after her. Nedaara could only wave her off.
Moon stepped up to Keiko. “Where do you want me?”
CHAPTER 10
Faces of the boilersuits were flattened into point-masks, bodies optically sized, weighed, pushed up to Neon’s enclave on anchor. It took a few quantum oscillations for them to be poured over. A flick and they feed into groups based on gender, race. Virtutars fetched records from elsewhere on anchor, mapped the cybernetic data atop them to build out initial profiles. A flick or two and the profiles entered viewer models, tested against algorithims of Neon’s user-base constructed from a decade of A-B testing. The field of a dozen profiles thinned, pruned by the preferences of a base predictable given its size.
The virts went deeper. Profiles whipped through subset models, tested against Indian females, MTs, those within a mile of a dispatch. Any configuration bubbled into resolution, fed the profile, then burst back into the enclave as the virt moved on. All rendered simply within the exchange of quanta within servers in Little Tokyo, in Richmond up in Frisco, across in Shenzhen, a lattice of light parcels flicking through fibres and jumping server synapses. In places they’d look like laser, crossing a synapse bearing a significant load. In others, through fibre bundles as thick as a tree trunk, they’d glow cold.
The MTs folded their sense data onto the scans, built out the profiles with impressions formed by their own virts. Silent judgements — gleaned in a few flicks of quanta through their taps — fed through the same channels, pulling profiles out from dismissal, discounting others from approval. Profiles were mapped against narratives, tested for coherence and believability before scored for potency in concert with the models. All from the back of the freighter, pinging anchor through satellite relays embeded in the cabin. The boilersuits’ profiles run over and over in varied permutations to yield a field of four.
Kim Seong-min performed well as LEADER. The narrative filled in her victory motivation with PERCEIVED INFERIORITY from RELATIONSHIP WITH ESTRANGED FATHER. María Rocha could be cast as LOYAL PROTECTOR given ELDEST SISTER of YOUNGER MALE SIBLINGS. Park Eun-jung as DEXTROUS SHOWMAN, ATHLETIC from his past as STUNT RIDER. Willis Loehr would be MECHANIC, offer the team TECHNICAL PROWESS once understood to be a MECHATRONIC ENGINEER. This team composition rose to the top from the rattle of the algorithim through anchor. It was intentional artifice, specified in consultation with the patterns of Neon’s base. The MTs sitting in the back of the freighter did all of this before Moon and Carter were at the starting line.
Nedaara joined the boilersuits lined up along the section of the freeway by the line. They’d laughed at Carter’s gear before, but they couldn’t help but feel the throb of its engine in their chest, right where it felt best. Moon’s ticked over with the murmur of a mosquito in comparison, the occasional sputtered burp from the exhaust.
Eu was to follow. She ran I/O hot. Her cerebellic implant glowed with shimmering quanta as she straddled one of the boilersuits’ bikes positioned behind Moon and Carter. The heat bouncing off the tarmac, the raw scent of burning fuel from the bikes before her, the silence that had started to fall over the watching boilersuits, all funnelled up through her virt and packeted. She was the guinea pig. It was this packet that would later be pushed to anchor for Neon’s virts to chew through. They’d test it with some fleshy subscribers as well as the company’s robust models. Her experience from the closest possible proximity, her experience was what would be shared.
History had voiced concern about the rise of synthetic intelligence, the manifestation of a fear of building that which humanity couldn’t control. The fear was not unfounded. Every cell in the stream that flowed through Los Angeles ran atop a neural network inhabiting a small segment on anchor, one that exhibited the trivial capacity for system learning. When a human learns, they learn through mistake. Our fundamental failing is that the lessons from the biggest mistakes cannot be taught. The biggest mistakes are those that kill the ones making them. For a system, any one cell is disposable. Any one cell can wreck and the others will learn the lesson that led to its demise. A cell that careens off the Santa Monica Freeway, plummets down killing all inside will send telemetrics of velocity, g-force, brake temperature, all the constituants of what can inaccurately be called its “experience” to the neural network before it goes dark. The fleet improves regardless of the fate of any one cell. No cell will make that mistake again. For those that were lost inside, they aren’t able to teach the lesson from their own fatal mistake of stepping up to the curb and calling their final cell.
The fear is smart. And a lot of people feared the machines would beat us. But they didn’t count on us becoming machines. Even if the way we are equipped to receive data is all together far messier. Piercing the myelin sheath that surrounds an axon allows the electrical impulses that run through our nervous system to be tapped. Optical interfaces in the tap allow pre-conscious sense data to be packeted, pushed, pulled and received as if that sense data was one’s own. Just as any new cell possess the wisdom of two decades of fleet learning, a newborn human could conceivably have taken some of the first steps on Martian soil in 2027, fought Korean loyalists for the United Korean Peninsula in 2024, and tasted part of the fifteen-metre-long tentacle of a colossal squid all before their first meal of breastmilk. But that would be ridiculous.
We built ourselves again online. Virtutars that were our algorithmic approximations, able to spool through packets without us, learnt and pushed down what they knew we’d want.
Logs:
S: 16–1110_1029, 12799WE: 16–1110_1057, 13056W | 257WS: 16–1110_1102, 13056WE: 16–1110_1156, 13322W | 523WS: 16–1110_2105, 13322WE: 16–1110_2203, 13813W | 1007WTOTAL: 2.5H | 1007W
