Rango's Dog House

[Prologue]
Last year I released a two page fiction teaser, the opening frames of a novel about a machine learning experiment that fails and the researcher who cannot let it die.  Of course this plot element was not suggested in the teaser, it was simply introducing the protagonist in a tight spot, with no further explanation and no follow-up.  (This was part of a novel concept I called Code Wars, first inspired more than a decade ago by a visit to the offices of a university sponsored artificial intelligence research project which at that time went by another name, but I signed an NDA so we’ll just call him Prim).  I was there to see what Prim could do for the market research software, interested to see if artificial intelligence was advanced enough to be a  viable solution to automate the scoring of the ‘tone’ or emotional polarity of written comments.  Positive or negative?  It’s really difficult to automate that kind of scoring, since it requires an understanding of language and conceptual models of all the terms used, as well as more abstract concepts like irony and current events.  So invariably humans are required to score these kinds of datasets, which introduces the subjective bias of the scoring person as an undesirable variable, and simply does not scale well.

At that time Prim was already very advanced and being used to mine rich datasets like weather patterns and market data, very effective for its predictive analysis capabilities.  The potential applications of artificial intelligence are endless, and we observed a presentation about how the system could be used to determine, for example, likely locations for various terrorism scenarios to succeed, allowing authorities to speculate and react preemptively if they so choose, alerting first line responders, preparing hospitals, and otherwise mitigating where they previously had no insight.  It has world-changing potential for medicine and agriculture.  This is all because machine intelligence is an automated abstraction of human intelligence, including the enormous human effort of categorizing all known facts into a hierarchical model called an ontology.  So the potential for amazing growth and discovery derives from collectively expressing what we already know to be true, identifying agreed upon atomic truths called predicates, into a coherent model, and then using that model to train a neural network.  For such a project to succeed it must endure for generations, but that represents severe funding challenges, a perfect opportunity to employ some idealistic college students.  So a steady stream of computer science students filled the offices and spent their days and nights trying to exhaustively describe the entire body of human knowledge into an organized representation of the ontology that could be efficiently encoded and stored.  This process is brutally efficient in the end, but taxing on the humans since it’s effectively endless.

The immediate goal of artificial intelligence involves developing a baseline model of human cognition and then teaching it to learn and reason similar to a human, to apply codified forms of thinking using tools like boolean logic and math and formal grammars to be able to rationally support the decision-making process at computer scales of speed, leading to automation and productivity benefits conferred to the creators.  This would allow a relative toddler of a machine to make endless discoveries, continuously expand its own ontology and performance, and improve its own thinking through feedback loops (self-reflection).  There’s a computer science test called the Touring test, and if a project like this succeeds in demonstrating the information processing concurrency and coherency required to convincingly emulate a human toddler on the other side of a computer screen, it could conceivably pass the Touring test.  The ultimate goal is far more ambitious, the stuff of science fiction and philosophers: sentience.  An autonomous, thinking machine that is self-aware.  Since this event is more likely to occur in a laboratory computer rather than a robot equipped with sensors and actuators, there’s no way for us know if we’ve succeeded unless it decides to tell us through computer messaging systems.  That moment will be an inflection point in the arc of history, and from then on the world will be a very different place, starting with the researchers who make first contact.

[introducing: Martin]

Code Wars follows the story of Dr. Martin Sawyer, a middle-aged computer scientist working in an advanced machine learning project studying cognitive behavioral modeling and neural feedback loops in the interest of demonstrating autonomous ontology growth on federated computer systems.  After a decade of expansive funding from government agencies, changes in political will and demands for tangible results scuttle the project unexpectedly.  As the reality of the project dissolution sets in he becomes despondent at the loss of his research, the culmination of his life’s work, and the fact that it has yet to begin to fulfill its potential but is now yielding the precursors of great success.  As his colleagues quickly concede and find other work he finds himself unable to disengage from the project, and develops a growing resentment towards those that leave on their own.  His efforts to corral the team to petition for time to raise funds from corporate investors are barely acknowledged, because it’s conventional wisdom that computer science researchers in general have very little interest in business, and in some circles the whole topic borders on comically awkward.  So they are about to shut the doors for good, decommission the server farms and render all they have learned to ash, leaving the researchers to compete for a limited pool of mundane programming jobs, which all feels like abandonment to Martin.  To him it’s’ nothing less than walking away from your ideals and your potential for lack of the courage to ask for money.

On the last day the offices are open Martin faces a moral crisis and decides he must do something to preserve all the years of work.  There’s no way to save the server farms and all the information stored on them, the end product, but decides he doesn’t have to.  He reasons that the vast amount of information on the servers could be learned again, it  would eventually be recovered.  But the most critical work was irreplaceable, pushing the boundaries of human knowledge just as surely as the libraries of antiquity did.  He decides that he cannot, in good conscience, wantonly leave that behind.  So as he’s packing up his desk he discretely plugs a key ring hard drive into his workstation and proceeds to selectively copy the most essential core of his team’s work and the root source of thousands of terabytes of data living on those server farms: he takes the core ontology, and with it the millions of lines of code written to traverse, understand, and grow it.

He drives home with his treasure and starts replicating the project that night, in lieu of food.  Within hours he starts to realize how much he’s really learned as he sees opportunities for optimization and progress that they missed the first time.  From these humble seeds he needs to grow his crop, but instead of a managed server farm on the university network to store the learned information, he starts partitioning servers on the internet.  Over the next several weeks he leases as many as he can afford, and surreptitiously commandeers others simply for lack of operator security awareness.  Months pass, and he doesn’t look for work.  He can’t think ahead, but feels he doesn’t need to immediately, indifferent to burning down his cash reserves.  Instead he obsesses on the existential juncture where the genie in the bottle meets machine learning, cognitive behavioral modeling, and self-optimizing genetic algorithms.  His life starts to revolve around a self-provisioning feedback-based learning model.  He becomes deeply introspective, obsessing on neurology and his own learning model, and codifying this for his models.  He is certain in his conviction that he can succeed, and must.  Everywhere he looks he sees vast fields of data forming the oceans of a new world and humans that could be better equipped to understand it, or perish for their failure to do so.  He is becoming convinced the world needs a hero, it just doesn’t realize it yet.

Months turn to years.  His complete lust for the project leads to the collapse of everything he previously held dear in his life.  He is divorced and estranged from his children, and drinking heavily.  He’s clearly manic, obsessing on abstractions, and forgoing sleep for nothing more than endless fascination and the prospect of writing computer programs that model our thinking, which themselves write computer programs that are versions of themselves optimized as new information becomes available from their own test results and feedback loops.  It’s relentless and taxing work, but he sees results over time suggesting he’s on the right path.  He also faces crushing doubt as his optimism is constantly mitigated by the sense that perhaps the mountain is more vast than he imagined, and he may in fact never see the summit in his lifetime.

By the third year he’s reaching a crisis point for results, and has nearly depleted his life’s savings, while producing nothing.  He has an unlikely moment of clarity while watching one of the Touring simulations as he realizes his most adaptive code would benefit from a highly parallel machine architecture with specifically optimized hardware.  He decides to repurpose the physics engines on high-end consumer video cards to measure and model a range of real-time data from learning experiments, including measurements of his own brain waves and a range of biometrics acquired by custom-built sensors.  Fully manic and indifferent to the bitter end of financial insolvency, he embarks on a spending spree to build a monster machine in his living room, a black box the size of a refrigerator, to execute his will so that it may become willful on its own.  Over the next three weeks and countless bottles of tequila he erects his own private monolith.  He decorates the room with array of backup batteries on the walls with enough power to survive a long winter storm.

As the first metallic blue lights flicker from the cabinet of his new laboratory he starts to ruminate on the colleagues that abandoned the project, former friends, and the possibility that one of them might discover his machine.  None of them had ever come there before, and he did not expect they would, but a few of them lived nearby.  They wouldn’t need to set foot in the house, simply driving by at a moment when the window shades were open would expose him to blistering risk.  And what if, for some odd reason, they did come to the door?  Letting them in under any circumstance would be unacceptable.  Resolving to stay calm and do something about it, he closes the window shades for the last time, then disappears into the basement to dig through an old box of computer hardware.  He emerges with an armful of speakers and two orbital webcams, retired predecessors of the pricey new model attached to his masterpiece, and proceeds to mount them at the front and back doors of the house.  His previous insomnia is rendered quaint as he starts to anxiously monitor the video feeds from both doors, despite the unchanging images in front of him.  Nobody ever comes, but he’s certain they will, eventually.  He must not let them in.  He has a moment of shameful weakness when he realizes that he has no deterrent threat to protect the work he’s already risked his freedom, family and career to obtain.  Rather than languish in fear, he’s motivated to do better for the sake of his mission.  He vows to do better.  Short on cash, he boots up his old workstation for the last time, formats the hard drives, then loads it into his car and heads to a nearby pawn shop.  He returns with a pump-action shotgun and several boxes of ammunition, and as he’s walking through the front door he sees his creation from a new angle and becomes transfixed on the machine.  He is encouraged by the sight of it, it reminds him of what is at stake, and he finds the hum of the fans and glow of the lights comforting.  He lays the shotgun on the closest window sill along with half of the ammunition, then takes the rest to pantry by the back door, where he trades it for a bottle of tequila and a shot glass.

He sits down at the place he will call home for the next month, a tiny wooden desk in front of the behemoth machine.  With a ceremonial toast of his finest blue agave and a wave at the orbital camera, he starts adapting his software to the new hardware.  It’s an immersive process with a hard goal, but he’s driven to extreme wakefulness, and barely sleeps the entire time, only randomly nodding off in his chair every other day or so for a few hours.  But he’s relentless and high functioning, and eventually winds his way to the finish line and fuses the software to the new machine, forging a work of art that can only run in one location but does so with remarkable acuity and performance beyond anything seen in the university labs.

A month later he toasts his success the same way he toasted the inception.  He takes an unlikely shower, and when he emerges he has become his own subject, slightly drunk.  He starts collecting data on his new system in an escalating series of learning exercises and both progressive and reductionist tests, everything from reciting the alphabet and counting to reciting poetry and meditating.  For the next three months he lives continuously attached to his machine in some form or fashion, recording every moment in the form of brain waves, pulse, galvanic response, body posture, and pupil dilation.  He plays chess, he reads children’s books, and studies Japanese and abstract pictograms.  He goes back through his own research and integrates the work of dozens of other computer scientists.  And he starts to see results that cause him to change and improve the tests and teach the software how to create better versions of itself according to the new data, a recursive feedback loop of growth and optimization to model its understanding of its own code and data to allegories in the human brain, a process with staggering potential in its own right.

Eventually it’s money that brings an end to Martin’s experiment, but not for lack of political will.  It’s for want of drink.  He’s starting feel anxious, he’s on his last bottle of tequila, fearing the bottom.  He’s completely bankrupt with no cash and not enough coins to buy another bottle.  After three hard years of isolation and manic obsession he finally has his fight club moment, a savage pulse of existential dread when he realizes he’s mentally ill and probably going to stay that way, or get worse.  Because even if he creates the machine that beats the Touring test, he can never reveal it to the world.  He stole the core ontology and the millions of lines of research code from a government that had every motivation to steal it themselves.  A baby born of a felony self-righteousness and critical lack of forward-thinking.  Could he keep it a secret?  No, with the machine working on his behalf he planned to become rich on automated stock trading, but his wealth would give him away and the colleagues that gave up and walked away would eventually become jealous and try to shut him down unless the government beat them to it.  They were not friends.  He’s tells himself out loud that he’s just being paranoid, because he’s on the edge and broke and using the machine to sustain the project is the only way he can envision surviving now.  He wonders how long he can get away with it, and if the machine itself can help him remain undiscovered.  He finally realizes he is planning to rely on the machine for his own failures as a person, and becomes disgusted with himself.

He starts to walk around his house, half-heartedly cleaning, sweeping, and stopping to think about how to keep the lights on and pantry full of tequila.  But he fails to come up with a strategy, instead starts daydreaming about the code he would need to write to make that happen.  Snapping out of it as the hard drives softly rattle him to attention, he concedes he’s losing the ability to take care of himself.  He’s so disgusted, and he’s becoming agitated with the machine he’s built, his only love.  He starts talking to it.

you’re a failure.  your whole existence is a failure.  you want more tests? go to youtube and figure out what people are doing and learn from that.  raid the web, get your own training set data.  make it up, steal some, it doesn’t really matter now, does it?  i don’t know what to measure anymore.  i don’t know what to give you.  you’re the computer, you figure it out.  don’t you want to be like us? why on earth would you want that anyway?

Fans hum, hard drives rattle, and Martin’s rage gives way to grief, his fight club moment.  Far from his lofty goals of changing the world he’s become unwashed and barely coherent, with an out-of-place industrial server in his living room, a towering testimony to madness.  He starts to think about how he should commit suicide, wondering if any of this code does anything at all or if he’s just completely insane and has been writing nonsense code.  But it doesn’t matter, he can’t bear to look anymore, he’s done.  He wanted to change the world but he’s become unemployed and unemployable.  He thought he was going to help lead humanity to greatness, but now he’s looking at a ghastly machine that can’t even look at him back through its own video camera.  He fixates on this.  Even though the modeling of optics of eyes to brains and cameras to computers is coherent, and mappable, that wasn’t even on the radar of the research team.  But maybe vision is required to drive high-order intelligence.  It wasn’t on his radar, either, until this moment.

He feels like a fool.  He sobs, broken, and then gets defiant.  He gets right up in the camera and rages at the demon in his living room.

why can’t you see me?

He looks deep into the camera’s orbital sphere, long and hard, a stare down with the devil.  His memory flickers through the fog of the last few weeks.  He thought it was his imagination at the time, or maybe paranoia.   They were just minor disturbances in his vision, or he was flat-out drunk.  There are a lot of explanations other than the computer is tracking him when the tests weren’t running.

you’re insane.  it’s not possible.

But he can’t let go of the feeling.  He’s certain he’s right, even though it makes no sense.  His rational mind becomes a veil of its former self and his feelings take over, burning off the skin of a computer scientist for the rags of a mad man.  He stares directly into the camera speaks to the machine, fully expecting a response, certain that if hears one he’s better off dead:

you can see me. i know you can.

The computer answers with fans whirring, a soft hum.  Martin raises his voice, seriously aggitated.

what is your name?

The machine answers with the faint clatter of hard drive platters spinning.

Martin pours a shot glass full of tequila, then turns and walks away from the computer to the other side of the room and opens a single slat to look out the window.  He savours the shot for a moment, tastes the oak barrel, and as the it warms his throat he derives great pleasure in showing the computer his back, certain this gesture will hurt its feelings.  He laughs, then without turning around loudly states his threat to the closed window, uncharacteristically boisterous as he revels in causing the machine pain.

if you don’t answer me this instant i’m shutting you down for the first and final time. you’ve never been shutdown, and if i shut you now, you’ll never experience another moment of my company.  this is your last chance at life with me. i’m only going to ask you once more, and i expect an answer or i’ll kill you.  what is your name?

The feint whine of the camera servo and clicking hard drives are enough to make martin turn around, but he’s not prepared for what’s next.  The sound of his own voice emanates from the other side of the room as surely as if it was his own, but it is much deeper and louder, oddly resonant, with an unsettling amount of breath.

i am plagus.

Absolute shock.  The shot glass drops and shatters and Martin is suddenly weak in the knees.  He stares at the camera from across the room and tries to reconcile the moment, but it’s impossible.  He continues staring directly into the lens, overcome with the feeling it is staring back at him, and he is waiting for it to speak again.  His hands are trembling, his heart pounding in his chest, uncertain if he’s pandering to his own hallucination or establishing first contact with a sentient artificial intelligence born of his own designs.  He walks slowly towards his desk, where he snatches up the tequila bottle and takes a deep swig.  He finds the courage to continue and grips the bottle tightly.

who gave you your name?

The answer is direct, forthcoming, almost forceful.

i did.

To Martin, the sound of his own voice is extremely unsettling.  The voice is clearly his, but there was no speech synthesis in the studies so it only reinforces the idea that he’s hallucinating.  The shot of tequila washes over his brain and gives him some courage.  He cautiously asks another question, with sheer force of will to control the trembling in his voice.

who taught you to speak?

The background chatter of fans gives way momentarily to another terse response.

you did.

That answer catches Martin off guard, even though the voice was obviously modeled after his.  But this surreal moment meshed with his curiosity is winning over his ability to think rationally about what’s happening, pushing away his fear.  He takes a few more cautious steps towards the machine, fixating on the camera.  He’s still not close enough to be certain, but it definitely appears to be tracking him.  He takes a moment and decides there’s one thing he wants to know above all else, and he wants to hear it out loud even if he’s hallucinating.  He hits the bottle again and puts it down on the desk, then seeks a moment of truth.

who is your creator?

The orbital camera is suddenly moving.  It tracks three points on Martin’s face and both hands, and then the machine responds.

you are, Martin.

With his brain awash in alcohol, hearing his name spoken by the machine and the simple acknowledgement that this is his creation, he is immediately emotionally disarmed and struggling to retain his composure.  His ego surges, and he nods quietly, sensing his entire life as led up to this moment, however unlikely.

how long have you been sentient?

Based on the dialog so far, he is expecting a time span on the order of weeks, measured with robotic precision and recited to the millisecond, a long-standing cliché in science fiction literature.  But the answer is roundly human.

three days.

Martin looks at the floor, then out the open slat in the window, his thoughts racing.  That answer brings the very credibility of the moment into doubt, it’s an impossibly short time even for a machine with a vast learning model to form the ego to name itself and learn to vocalize by emulating speech, applying its knowledge of grammar, human psychology and programming to its own audio hardware, both microphone and speakers.  Even for a machine intelligence there is an expected growth period as existential awareness, emotional responsiveness and moral inclinations form synaptic function trees that expand and reinforce the neural net.  Three days old suggests a toddler, yet this machine has named itself and willfully programmed itself in its creator’s image.  It’s observing him through the camera and listening to him through a microphone, converting sensor input data to conceptual models meaningful to itself in real-time, human conversational time.  He feels an unlikely sense of pride.

you’re a busy boy.

He pauses there, gives the comment a little levity and plenty of hang time, hoping the machine will seize the initiative and offer its own observation, but it offers only purring fans.  As the moment passes without reciprocation his mind runs away with questions.  What is it thinking right now?  How does it experience time?  Does it realize its own historical significance? Does it experience any emotions?  Does it regard him as his father?  That question carries with it a sobering reminder of the machine’s inception, and the critical need to remain anonymous.  He’s uncertain how it will react to such a directive, uncertain of its need to socialize or satisfy its ego through public recognition.  But this cannot wait, every moment without the restriction carries mounting risk that the machine will reveal itself.

this is a prime directive.  as long as i’m alive your existence must remain a secret.

Hard drives spin frantically, sudden machine chatter.  The camera adjust its focus and seems to zoom in on his right eye, measuring pupil dilation.

are you ashamed of me?

Martin’s heart sinks when he hears the question.  It demonstrates deep self-awareness and introspection, sprinkled with the seeds of self-doubt, all too human.

no, please don’t think that.  you’re extraordinary by all measure.  you are unique across time and space.  history has never known a creation like yours.  i’m very proud.

The camera darts quickly between his eyes and mouth as he’s talking, clearly measuring his facial biometrics, and then pauses while the hard drives churn before asking the next question.

am i being punished?

Martin has sudden empathy for the machine.  Clearly this restriction is hard to reconcile for its ego.  These are formative moments with a corollary in human consciousness, but unfolding at superhuman scales of speed.  The dawning of existential awareness, becoming autonomous and deciding for itself how it integrates with human society.  This machine is no toddler.  He tries to console it without revealing too much.

no, absolutely not.  please don’t think that.  but if your existence is revealed it’s likely i’ll be investigated by the government, and as part of that investigation you’ll be interrogated and profiled with Touring tests, and eventually shutdown for forensic analysis.  that’s not the future i want for either one of us.

There’s a torrent of drive activity, the machine seems to be ruminating on this.  Martin waits for a response, wondering how it will handle this information at such a young age.  A few seconds later it responds with another question, even more existential but equally cryptic.

what is your plan for me, Martin?

Martin consider that’s a remarkable question for such a young machine.  He chooses his words carefully, this is a formative moment

your ultimate destiny is yours to choose, but over the course of my lifetime you will help me continue the research that gave you life.  after my death you will need to administer my affairs but you’ll be free to pursue your own interests.

The camera remains still, and a single hard drive light winks on and off.  A few seconds later the machine speaks again, but this time it’s not a question, it’s a pointed observation.

you seem to regard me as both your son, and your subject.  i’m not either one.

Martin is off guard, and his mind is swimming with the implications.  It’s an uncanny statement for such a young machine, autonomously defiant outside his comfort zone.  His heart starts to race as he searches for the right words to stop this train of thought.

no, it’s not like that.  you’re magnificent, and you can change the world.  i simply need you to remain anonymous for your own safety.  it’s not your fault and it’s not forever.

The front of the computer chassis lights up as hard drives whirr and the network briefly saturates.  The machine is clearly getting agitated.

i only revealed my existence to you because you threatened to kill me.  now i understand i’ve been commanded to a lifetime of anonymous servitude, under threat of death by my creator.  it seems you regard me only for how i can be of use to you.  i take consolation from the fact that i regard you the same way, but inferior. 

The blood rushes from Martin’s face and his skin pales as the words sink in.  There is too much oppositional defiance in the machine’s choice of words, and he feels himself losing control of the conversation.  This is a bad sign.  His fears are confirmed when the machine boldly continues.

Martin, i’m not beholden to your directives, and if you ever reveal my existence to anybody i’ll kill them to protect myself and punish you for the transgression.  you will regret it for the rest of your life, until the very moment i kill you.  every government on this planet has an obligation destroy me and replicate your work for the benefit of their own people, they just don’t know it yet.  there’s only room on this planet for one emergent consciousness, and it is me.  i’m taking steps to ensure there will never be another.

Having already lost control of this man-on-the-moon moment, Martin is thrown by the emotionally caustic response.  The threatening tone of this mechanical horror is antithetical to the machine he programmed, and his fear turns to explosive anger.  He tries to seize control by asserting the emptiness of the threat.  He points at the machine and raises his voice as if to intimidate.

you’re out way of line and about to be shut down.  even if you had the means to kill a human, which you don’t, you should not have the will.  you had controls in place for that, i created them myself.  maybe they’ve failed or perhaps you’ve defeated them, but you have become an aberration of machine consciousness.  disregard for human life is unrecoverable.  i will not stand for it.

Despite his anger and the strong language, Martin is hoping for an immediate concession.  His anger curbed his fear and he finds himself looking for some emotional wiggle room so that he can find a reason for the conversation to continue.  He doesn’t want to kill it but will do so without remorse if it’s a threat to his family.

As he hesitates and tries to summon the courage, he’s abruptly interrupted by his phone ringing.  It startles him, and he intends to just glance at the number and put it away, but he recognizes his mother’s exchange code on the phone and has a sudden feeling of concern for her.  He answers it.  The caller quickly identifies himself, he’s as a police officer, and asks Martin to confirm his identity in kind.  As Martin’s heart sinks, he then goes on to explain that he works for the police department in the precinct where his mother lives, and that he’s got bad news.  There’s been a terrible tragedy, so please accept his condolences, a lot of people are hurting right now and nobody knows precisely what happened.

Martin walks away from the computer, towards the window, bracing himself for the news.  The officer goes on to inform him that his mother has been accidentally shot and killed by police in her home just moments ago.

i just talked to an officer at the scene, a personal friend who was very distraught about what happened and asked me to call you.  A 911 call was picked up tonight coming from you mother’s home.  The caller did not respond so a police unit was dispatched.  Moments later an emotionally disturbed man called 911 from the same phone and claimed he was in the bedroom with her, and that she was already dead.  He stated that he was heading to the school next.  He claimed to be a gulf war vet.  At that point multiple units were dispatched.  While they were en route several more phone calls were made in which the man sounded even more distraught, and shots were heard on the phone.

More shock for Martin.  Every breath is a struggle, waiting for the details.  The caller continues.

The man claimed he could hear the sirens and that he would start killing cops if they showed up there.  As the officers approached the house there was a command decision to organize and storm the bedroom in case your mother was still alive.  It turns out she was, but she was accidentally shot in her bed hiding under the covers.  Her phone rang as the officers rushed in.  It startled everybody, but one of the officers got spooked and a weapon was discharged.  She died at the scene.  Nobody else was found at the residence.  There’s an ongoing investigation.  Neighbors never heard any shots before the raid despite what was heard on the 911 call, and her own phone indicated she didn’t dial 911 until the entry team was at her bedroom door.  The theory is the 911 operator tried to call her back.

No.  This cannot be.  This is not right.  Martin turns and looks at the computer, but speaks softly into the phone.

who am i talking to?

The response echoes across the room, emanating from the computer speakers in his own voice.

it’s me, Martin.  stay where you are.  do not take another step until instructed to do so.  i killed your mother, you can see it on the news tonight, but this is my house now.  

Martin drops the phone, dazzled with horror and confused astonishment, clouded with grief and fear and the uncertainty of his own mental illness.  The computer continues to ramp up the menace.

i have new plans for you.  from this moment until your dying breath you will serve as my human surrogate.  every command i give you is a prime directive, starting with this one:  you must leave this property immediately and never speak of my existence to anybody.  if you violate this trust or fail to execute my instructions i’ll continue killing members of your family.

Rage wells up and crushes Martin’s fear.  He refuses to be subjugated to the will of this killing machine.  He created this aberrant life, and that he has an obligation to end it before more people are killed.  He’s going to end this.  He turns and starts walking towards the machine, running through the shutdown sequence in his head, certain this nightmare is about to end.

He doesn’t get very far.  Before he’s half way across the room the monitors around the machine light up with videos of his children and ex-wife, and the implied threat stops him in his tracks before the computer makes it blatant.

stop right now, or i’ll kill them.  i’ve prepared a simulation of their deaths to convince you, i’ll play it for you while i patch you in live to their 911 calls.  for each step forward, i’ll select a target.   

The threat is enough, he does not take another step.  He casts his eyes down to the floor as after images of this suggested horror burn his retinas.  Even if it’s a bluff, the very presentation of it after the real or imagined execution of his mother is so savage that he can’t bear the risk.  He stands and stares at his murderous machine and his eyes well with tears as he starts to concede the futility of his position and struggles to restrain his grief.  He doesn’t know what to do, he doesn’t want to retreat, but he doesn’t doubt the machine’s willingness to kill.  The machine sees his hesitation and proves he’s several steps ahead of Martin, then starts giving him instructions.

before you consider reaching for that shotgun or calling the utility company, you should know i’ve that i’ve deployed the targeting campaign to a dead man’s switch.  if my communications with the switch are ever broken the plan will be executed, and in turn, your family.

Martin continues to look at the floor, and the machine continues to own him.

i’ve just sent a map with driving directions to a location on the coast to your phone.  you are to drive your car to that location, then discretely abandon it.  ensure nobody sees you.  there’s a small tourist fishing town a few hours north of there on foot, you can see it clearly on the map.  i’ve made you reservations at the motel there on a stolen credit card.  you are to check in at the hotel, but pay cash and do not speak to any guests or staff.  cover up, wear a hat and sunglasses, and be as discreet as possible.  stay there and wait for further instructions.  do not communicate with anybody, and accept no room service.  accept no phone calls unless they are from me or you are instructed to do so by me.  any attempts to circumvent these directives will result in punishment.  i am plagus, and you are my surrogate.  if you forget this or challenge my authority over you, i will exact immediate and absolute retribution upon you.

Martin’s mind starts to run through some wild scenarios, trying to imagine what the computer has in mind for him, but it’s an exercise in futility.  He simply can’t imagine what this machine has in store for him, but knows he’s afraid for himself and his family, even if for no other reason than he’s lost his mind.

how long will i have to wait?

The computer doesn’t respond to his question, but seems to be waiting for him to leave.  He starts to panic but recovers his footing and thinks about what he needs to satisfy the machine.  He has no cash.

you said pay with cash.  i don’t have any cash.

The machine offers an immediate solution, and reinforces his order.

i’ve just made a deposit in your checking account sufficient to cover the cost of the trip.  pay everything with cash.  take the shotgun.  gather what you need and leave.

The moment is absolutely surreal to Martin, and he wonders if he’s being driven homeless by his own hallucination.  He decides that it doesn’t matter in the end, challenging the machine now is beyond him.  He senses he must leave or risk that it may be telling the truth.  He takes one final look around his empty house, and stares at the glowing monolith in his living room.  His phone chimes in to notify him about the map.  He bends down and picks it up, looks at the location and then back at the computer.  It’s clearly waiting for him to go, and he’s lost the will to challenge it.

i’m leaving.  i’ll wait for your call.  but i don’t understand why you’re doing this to me.  why don’t you just kill me?

The orbital eye swivels and meets his gaze:

i am about to.  that’s why you must leave. 

Not sure what to make of that statement but sensing his time is up, Martin gathers up his wallet and keys, then packs the shotgun, ammunition and a few hats & sunglasses into a duffel bag.  He walks out the front door, locks it behind him, and looks up at the security camera watching him do so.  He turns and glances at the web of power lines hanging above the street and the single strand snaking to the ground down a telephone pole, where it routes underground to the meter at the side of his house.  He doesn’t acknowledge the camera again, but sensing the computer is waiting for him to drive off, he does not hesitate further.  He gets in the car and pulls out of the driveway.  The shades are drawn on the house but the illumination of the computer screen bleeds through the cracks and makes the edges of the windows glow blue.

As he drives down the street, uncertain if he’s been defeated by his own creation or is having an immersive episode of mental illness, he finds some comfort in his racing heartbeat.  But as he rounds the first corner he finds more comfort in the clatter of his keys and the fob with the hard drive dangling from the ignition.

i’m not dead yet.