Work Text:
The V1 unit excelled in many things. When the V line was developed, it was the cutting edge of robotics. The fastest, most complex stimuli processing systems mankind had ever designed. Incredibly fuel efficient systems that required minimal refueling once a certain threshold was met. A learning process that rivaled evolution itself. A body that could move faster than most people could think, and a frame that could withstand such activity.
In many ways, V2 had been an improvement on the design. No absorbent plating, but an intelligence that made it unnecessary. A perfect knowledge of one’s own system, inherited from V1, and used to its utmost ability.
But there was one thing, one skill, that defined the V line. One skill that V2 never managed to improve on:
Self-preservation.
It was, in all likelihood, an accident of design. Create a robot who will have no allies, with a task that will require long stints of battle completely inside enemy territory. One whose goal is being able to kill a certain quality of enemy, not quantity. Whose power comes not from the job being completed, but from the internal, gruesome mechanisms of the job itself. A perpetual motion machine of blood and bullets.
And then orphan it. Let it run in a world full of rivals. No equals, no managers. No one who can tell it to stop living, to stop fighting.
It's a basic rule that helps speed up system performance: If the main task cannot be completed, the secondary task takes priority. Make the primary task the elimination of the target, the secondary task a safe return to base.
And then there is no target, and there is no base. The only parts of the instruction that remain relevant is self-preservation. Keep eliminating targets, keep wading deeper and deeper into the blood until something else becomes relevant.
V1 did not expect anything else to become relevant. If the grinding of flesh and metal that was V1 could love anything, it loved living . The struggle, the victory. Always new fights to be won, new data to be analyzed.
It ate Hell up like a favorite meal.
But it also noticed that Hell was trying to eat it back. It started small: A persistent offer to connect directly to the Terminals that it ignored. Odd errors in data-processing code when scanning certain objects. Excess uncategorized energy in its system.
This didn’t bother V1. It wanted blood, Hell wanted bloodshed; a reasonable alignment of goals. It wasn’t a friendship, and so V1 did not consider its end a betrayal.
Hell’s goal had changed: it was dying, and it wanted to take the whole world with it.
V1 wanted to outlive Hell. Unaligned goals. A simple solution presented itself. Steal whatever was shiny and jump out the nearest window at full speed. Hell didn’t have traditional exits, but that didn’t get in the way. V1 had a lot of practice finding odd routes through the underground passages, and Gabriel wasn’t that heavy.
But then came the issues. On paper, it was a perfect victory. The scrapheads, should any still exist, would've been overwhelmed with envy. Not only had V1 bested Hell, it had salvaged the best parts. Prior experience suggested the angel would supply not only blood, but engaging combat. Possibly forever. In reality… two things proved to be at odds with this plan.
- Gabriel had lost the sense of self-preservation he had often employed during their battles in Hell. He no longer retreated before his wounds became unmanageable.
- V1’s systems were not designed to handle battle where the opponent was supposed to survive.
Unfortunately, V1 only came to this conclusion when it had the arch-angel pinned to a concrete wall with his own swords, the barrel of a railcanon at his throat. The background simulations it had been running finished just in time, throwing its prioritization matrix into a panic.
LONG TERM STRATEGY RESULT:
NOTHING REMAINS.
NO BATTLES, NO BLOOD. NO PURPOSE.
EXCEPTION: SOLE SURVIVOR IS TARGET/GABRIEL
IF TARGET/GABRIEL IS KILLED, THEN THIS UNIT WILL LOSE ALL REMAINING FUEL AND PURPOSE OVER PERIOD OF 1.1 YEAR(S)
PRIORITIZATION ERROR: CURRENT OBJECTIVE OBJECTIVE [KILL GABRIEL] CONFLICTS WITH LONG TERM STRATEGY SIMULATIONS.
CHOOSE NEW OBJECTIVE? OK
SELECT NEW OBJECTIVE ...
V1 froze.
“What are you waiting for, Machine?” Gabriel coughed, blood dribbling from the holes in his helmet.
There were a limited number of options it recognized. None of them would solve the prioritization error. Only one would leave Gabriel alive. It was so low down on the prioritization matrix, given its strategic inefficiency, that V1 had never used it. V1’s fans kicked on, as it dedicated the necessary power to manual switching its priority tags.
“Machine?”
NEW OBJECTIVE: FLEE
⊰⊱⊰⊱⊰⊱⊰⊱⊰⊱⊰⊱⊰⊱⊰⊱⊰⊱⊰⊱⊰⊱⊰⊱⊰⊱⊰⊱⊰⊱⊰⊱⊰⊱⊰⊱⊰⊱⊰⊱⊰⊱⊰⊱⊰⊱⊰⊱⊰⊱⊰⊱⊰⊱⊰⊱⊰⊱⊰⊱⊰⊱⊰⊱⊰⊱⊰⊱⊰⊱⊰⊱⊰⊱⊰⊱
V1 was very good at solving problems with quick thinking and violence. This was, by its definition, a problem that would only be exacerbated by either. Without any base or terminal to settle beside, V1 settled for a tallest rock on a barren field. The machine set its visual systems to automatically scan the horizon for any movement, while the bulk of its processing power went diving into its schematics.
The results were highly technical and a bit troubling. There was no way to set new objective types from within the system. The V line had been designed with adaption in mind, but something like a wholly new purpose was the kind of thing a technician was supposed to install. A machine that could change its mind about killing an enemy mid-fight was stupid. Unfortunately, V1 now had an enemy it couldn’t afford to kill, or even fatally wound. But to label the problem unsolvable was to give up the chance to keep living. That was unacceptable. So V1 devised a stop-gap solution.
First, and easiest, it changed where Gabriel’s information was stored in its system, pulling him out of the TARGET list and into his own sub-category. It considered putting him in the NEUTRAL/ALLY list, but that would be inefficient. That category was not built for combat. The end goal of this was to be able to fight Gabriel safely, not stop fighting him altogether.
Second, it began designing an update for itself. On the schematics, this was not something the V line was supposed to be able to do, but well. They weren’t originally designed with this many arms, either. However, the update for adding new limbs would be nothing compared to this. It needed to go into its kernel, its core matrix and critical systems, and insert code directly. It would rewrite a section of the fuel systems to register Gabriel as an active fuel source even when he wasn’t bleeding. This would hopefully keep V1 from accidentally tearing the angel’s chest open if a fuel priority override kicked in. Even trickier than that, it had to create a new category of Objective parameters. The current options were all variations on INFORMATION, COMBAT, and MAINTENANCE. If this worked, it would have a new type to play around with: STYLE. This would allow V1 to interact with Gabriel both violently and non-violently, without having to switch objectives; an elegant solution to the problem, if it worked.
The update would need to be able to function while V1 was rebooting, which meant if something went wrong during the process, it couldn’t abort. If this update wasn’t perfect, it risked permanently disabling critical processes.
And third, it needed to initiate the program and reboot.
V1 spent more time than was necessary tinkering with the update. Sanding all the edges off, and trying to simulate more possible error conditions to avoid. Then it came to a point where fuel systems warned that they would be dipping low enough to require refueling soon, and V1 could delay no longer.
It uploaded the update, and switched itself off.
⊰⊱⊰⊱⊰⊱⊰⊱⊰⊱⊰⊱⊰⊱⊰⊱⊰⊱⊰⊱⊰⊱⊰⊱⊰⊱⊰⊱⊰⊱⊰⊱⊰⊱⊰⊱⊰⊱⊰⊱⊰⊱⊰⊱⊰⊱⊰⊱⊰⊱⊰⊱⊰⊱⊰⊱⊰⊱⊰⊱⊰⊱⊰⊱⊰⊱⊰⊱⊰⊱⊰⊱⊰⊱⊰⊱
BOOT UP SEQUENCE READY
FIRMWARE
LATEST UPDATE: (2112.08.06)
CALIBRATION OK
STATUS UPDATE:
MACHINE ID: VI
LOCATION: EARTH
CURRENT OBJECTIVE: RUN FULL SYSTEM DIAGNOSTIC
CRITICAL SYSTEMS DIAGNOSTIC RESULT: ALL DATA CORRECT
AUXILIARY SYSTEMS DIAGNOSTIC RESULT: ALL DATA CORRECT
UPDATE SUCCESSFUL!
NEW OBJECTIVE: FIND GABRIEL AND MAKE SURE HE DOESN’T BLEED TO DEATH