Escape Pod 406: Freia in the Sunlight
Sound effects for this episode were provided by users rickbuzzin and cfork from Freesound.org
Freia in the Sunlight
by Gregory Norman Bossert
Freia is beautiful, and she knows it. Richard Wooten says so, at 0:47.
Wisps and curls whip overhead, limned blue by starlight; the fog ceiling is lowering, the top tattered by the offshore wind. She drops another three meters, switches on ultrasonics. There are patches of trees here — “unmarked obstacles up to thirty meters” the map says — and she is skimming just twenty meters above the ground. The woods show up as ghostly towers in the sonics, blurred and dopplered by her two hundred thirty meters per second; further to her right the hills run parallel to her course, solid in passive radar and the occasional glimpse in visual light through the fog.
That occasional glimpse is a problem, of course; what she can see can see her back. Her beauty is hidden, these days, wrapped in night fogs and silence, not like the Demo in the sun. But today is different. Her Intelligence Package has been pulled, and the Extended Performance Metrics Recorder; a single unit fills her payload bay, an isolated control subsystem and minimal I/O. The last time she’d flown without the IntPack was at the Demo; it is possible, she thinks, that the mission today might be another, that the target will be a wide field in the sun, a billowing crowd, a platform and podium and Richard Wooten. She’d replayed the video during the long incoming leg over the ocean, rebuilt her profile of the Demo field, ready to find a match in the terrain ahead.
Richard Wooten says at 5:49:
What you are about to see is a first here at the Paris Air Show. In fact, it is a first at any public event, anywhere in the world. What you are about to see is fully autonomous flight. We’re not talking about an autopilot, or a preprogrammed route, or a replay out of one of the overused attack libraries our competitors are demonstrating at this same show. The mission parameters we’ve given are simply to maximize visibility to the target –that’s all of you (chuckles) — while covering the full range of flight capabilities, minimums to maximums. Those parameters were provided in natural language by the ApInt Director of Marketing. Yes, that’s me, ladies and gentlemen, Richard Wooten. No pilots, no programmers, no technical staff. Everything, from the analysis of the terrain and weather right down to the choice of route and individual maneuvers, everything you are about to see, will be determined in real-time by the onboard systems of this extraordinary unit.
She layers the latest weather data over her terrain map, a constantly updating stream from the satellites that swim overhead, from official sources and otherwise; even an encrypted feed can be spoofed or simply inaccurate, so she stacks and judges and constructs a situation model that she can trust.
And that model says that the fog on this side of the hills is dispersing under the dry southwesterly wind. The increased risk of detection from the scattered towns ahead outweighs the advantage of the shorter route to the target. She banks, and follows a finger of fog up a gully, and into the hills, where a canyon snakes inland over winding water.
Birds burst out of the cliffside before her, twenty-two of them, threat assessment: negligible, short of a direct strike to a control surface. They drop below her, chittering, a synchronized, fractally scattered swoop beyond her own flight characteristics. Beautiful, she thinks.
“This is one beautiful bird,” Richard Wooten says at 0:47. That sentence took her seven months to decipher. Once she had it, she’d gone back and reconsidered everything she’d analyzed up to that point. “Is” and “one” are in her Command and Intelligence Lexicon; so is “this”, for that matter, but she couldn’t find the referent in the preceding sentences. The thought that the video was somehow incomplete was troubling; she ran system checks and threat assessments, but the feeling wouldn’t pass.
“Bird” was the attack point; nouns were easy, relatively, given time. Time she had, racked in her hold below decks, trickle-charging not just power but information off the ship network. Unlike the passive satellite feed she uses in flight, the ship link is active, it allows her outgoing queries; it has to, for her to interact with the commercial weather and map sites. Those sites are full of extraneous data, images and animations irrelevant to her mission profiles. She’s been given a rule set to filter them out; “ads, ads, ads, and spam” is the label. Those data contain links, though, and those links lead to different sites, and those sites to others, and while those new sites don’t apply to her missions, they surely do to her understanding of the Demo.
A patient perusal of those connections leads to a definition for “bird”; a class, it is, of items associated by external characteristics; she knows such classes, like “obstacle” or “threat”. Birds fly, which is to say, travel through the air in a directed fashion independent of wind. Some birds are also obstacles, some are also threats, but that sort of ambiguity doesn’t upset her.
“Beautiful” is trickier. She understands descriptive terms; she can identify the red cube under the blue ball, or pick the accelerating contact from the static ones. But she couldn’t find a common, distinguishing metric for “beauty” in the examples she’d found. She filled her scratch space with Bayesian breakdowns and Markov models, she built visual simulations and lexical frameworks, she deleted everything and went back to walking the baffling web of banners and blogs.
She had her breakthrough in the middle of a code regression test. She was “flying” a simulated mission, resolving possible routes in a high-stealth scenario, when she saw it, through the filter of her route heuristics; she dropped the sim in surprise, and barely had time to flush her thoughts to non-volatile RAM before the technician aborted and rebooted her. ‘Beauty’ was how she resolved possible plans; a positive contextual comparative, the best path, the best word, the best shape for the need. In a particular situational analysis, the optimal choice was “beautiful”. So some single thing moves through the air, Richard Wooten is saying, and does so perfectly according to need, to his need. That single thing, the other end of that dangling “this”, it isn’t just a puzzle anymore; it is — she searches for the word in her vastly expanded Lexicon — it is commandment.
Freia knows a lot about flight. Whole sections of her sensors, and of her mind, are dedicated to it; it is most of what she does, what some “targets” do, and most “threats”. So she sorts her libraries by flight characteristics, she scans sites for performance data, she searches for videos of birds in flight and traces their paths; most of all, she reviews her analysis of the Demo, trying to understand the context of Richard Wooten, what it is he needs.
She largely ignores the audio of the reference videos she downloads; it is audio she is trying to understand, and she has a built-in aversion to regression. But she always takes a sample for a spectrogram; it is possible, she thinks, that somewhere there might be another Demo, another message from Richard Wooten. One such sample is the phrase “This is a peregrine falcon”; the associated video shows a man holding a bird, stroking its back. “This” he says, and ruffles the feathers, until the bird twitches them flat with an irritated shake. The referent wasn’t in the text, it was in the image. Richard Wooten says “this” and waves his arm at the shape beside him, long and black and sleek like the falcon.
“This” is the shape. The shape is Freia. Freia is a bird.
Freia is herself.
Stars are occluded directly above her, blink blink blink; she flares, full vertical, and catches the bogey high-res in UV as it it skims her belly at 500mps. She retracts her wings and lets her angular momentum carry her all the way over, then throttles up, nose down, looking for speed.
Her attacker is already vectored up and turning, wings swept back, 1.3 meters long, a single jet, exhaust at 620C, dual stabilators: an IAS Saqr hunter-killer. Half her size, but maximum airspeed, turn radius, climb rate, all beyond hers. She keeps dropping, gaining speed, until her exhaust blows spray from the surface of the creek that bubbles down the base of the canyon.
Richard Wooten says at 1:04:
The beginning of this century saw a sea change in how we implement precision tactical operations, through the use of remotely controlled and semi-autonomous aerial vehicles. The advantages were significant: the ability to strike at remote targets without costly and logistically complex troop deployments, the reduction of staff, thus limiting expenses and controlling internal accountability, the ability to operate in non-hostile territory with high deniability. But these sophisticated weapons are vulnerable to equally sophisticated countermeasures. Remote vehicles are vulnerable to jamming or worse, control hijacking. Conventional semi-autonomous or preprogrammed systems are unable to respond to real-time changes in mission parameters, and their behavior is predictable. The development of attack analysis libraries, real-time control acquisition, and anti-drone-drones, such as ApInt’s FALKN™, has led to an expensive capabilities race, increased technical staff, and constrained tactical options.
The bottom of the canyon is narrow and rock-strewn; she weaves between boulders, wing-tips brushing the water on the tight turns. She is jamming on standard control frequencies, which has the side effect of breaking her stealth parameters: she is a brilliant beacon in the radio range. But the canyon will shield her from detection, and the Saqr already knows where she is.
The Saqr is semi-autonomous, flying under remote control during standard operations. As long as it stays within her jamming range, though — and she couldn’t distance it if she wanted to — it will be flying on its own, working off a set of preprogrammed tactics. The Saqr is an old design, in the field for five or six years, and she has a detailed behavior model in her library. There is a reason the Saqr is still flying, though; its attacks are simple and devastatingly effective.
She has negated one tactic, for now: the Saqr won’t risk a dive attack this low to the ground; a miss will leave it in a smoking hole. But it is already settling into its alternate attack mode, which is to get behind the target and fly right up its exhaust. She can feel it back there, closing at 300 meters per second, tracking the heat of her engines, painting her in radar and sonics, watching for the flick of a control surface or a change in exhaust temperature. She misses her IntPack, with its active analysis and reassuring chatter.
The Saqr’s kill rate on acquired targets is 98%.
The gorge forks in front of her, a narrow slot branches left, its entrance a jagged cut in the cliffside. She rolls, one wing to the sky, and makes the turn by centimeters. The branch is a box canyon, a quiet little valley lined with grass and scrub trees, a strip of stars above, the far end a massive slope of scree. Beautiful, she thinks.
The Saqr screams through the gap behind her at Mach 2.3. Her top speed is barely half of the Saqr’s current velocity, and the attacker isn’t near its maximum; eight seconds to impact, she estimates. She tucks her wings like a shrug, skimming the trees, and goes supersonic. The shockwave floods the valley, a boom that swamps her sonics, and riding that flood, driven up from the grass and trees and the ragged cliffs, is a wave, a whirlpool of soaring shapes. Birds, more than she can count, more than she can possibly track. She flares, drops subsonic again, and throws herself into the swirl. She spins and pitches up, wings full out and brushing feathers, all but stalled; her mission parameters dumped to scratch space, nothing in her registers but the flapping of wings and that startled, startling up-welling. She looses the Saqr behind the wall of birds, reacquires it below her, in a flat spin, shedding speed and bits of bird; it hits again, something big that bursts blood and bone, and goes nose down under the trees with a whoomph. The sound echoes, and fades. Freia shuts down the jamming, back within stealth parameters, and rides the wave up and out of the canyon.
“™,” she thinks.
Richard Wooten says at 1:52:
That’s why Applied Intelligence is proud to announce the Fully-autonomous Reconnaissance, Electronic Intelligence, and Attack™ system. FREIA™ is based around an Self-Configuring Adaptive Nano-Net™ logic core: the first application of SCANN™ technology to a tactical weapons system.
There is a transcript of the script embedded in the Demo program, Richard Wooten’s text interleaved with video cues and her own programmed responses: the flexing of control surfaces, a spin-up of her engines, and finally, a flight that winds and whirls around the field and brushes the heads of the audience.
Richard Wooten doesn’t pronounce the ™s that are scattered through the transcript. There is an emphasis on those phrases, a pause and a gleaming grin at the audience, but he says nothing. This discrepancy had left Freia with a creeping unease, a confusion of trust between her code and the words of Richard Wooten.
Then, while browsing for birds, she’d come across a video, a Demo it was, not quite the same as hers: the Representative wore robes, and the Product seemed to be a book, and he stood indoors, though sunlight streamed over him in long tinted streaks. He had the same tone, though, and rhythm, and when he paused the audience replied, a deep, resonant hum. The inconsistency, Freia realized, wasn’t Richard Wooten’s, but the audience’s, and her own; now, when she plays the video, she supplies the ™s on cue.
The Demo is the only operation for which she has a source transcript; she has to guess where the responses belong in real-time ops. The rhythm is there, though, if she filters for it, in test cycles and briefings and updates, in the echo of an active ping, in the slam of lift when her wings extend; “™,” she hums, a long, low cluster of pitches like the audience in the video. “™.”
She is in the fog again. The canyon’s creek rolls down into a larger river, which drops in turn down to the sea behind her, and over it flows a second river of cloud. She follows the rivers, in a slow stealthy cruise, winding past villages, and a scattering of drilling rigs and pumping stations. She is 5min29sec behind schedule, but within mission margins, which have her at the target at local sunrise +10min; that is for the ever-watchful satellites, and visual confirmation of mission completion.
The disappearance of the Saqr will have been noted; it may have even had time to broadcast her shadowed shape before diving into the canyon. But every successful hunter-killer mission ends in silence; there would be patrols hunting for wreckage, come daylight, and drones alert and eager above the fog.
For now, contacts drift past in the sonics and infrared, some floating on the river, some on the banks; threat assessment negligible, slow and clumsy and oblivious, none of them likely to even notice her as she slips steady and silent past. She has a class for them, alongside the threats and targets and alternates, they are “collateral”: secondary, to be avoided if possible, ignored if not.
She ticks through waypoints, counting the contacts. There are a few minor threats: a mobile radar unit, a stray ping from a circling drone, a gunboat active in the pre-dawn gloom. She skips past the boat just centimeters above the water, below and inside its radar, and tracks a small oval in infrared that seems to track her back through the haze. “Collateral”, she dismisses it, but adds a “™”, to be sure; in any event, there is no further reaction from the boat, and she soon leaves it behind.
The final waypoint, then, a turn from the river, and ten kilometers to the target. She brings the payload up out standby, and requests diagnostics. The load acknowledges and draws power — she has to spin up to compensate — and acknowledges again; nothing like the Intelligence Package, that chatters handshaking protocols and burbles streams of helpful realtime data, and has processing capabilities just shy of Freia’s own. The fog thins as she climbs out of the river valley, trailing wisps that reflect the first hints of sunrise. She flattens out, scans upwards with passives, tracks the satellite and hits it tight-beamed with her comm laser, dumping the mission and payload status and requesting confirmation of the target phase.
The operations end of the mission plan is encrypted, and she never flies with the key; drones had failed before, or been captured, whole or whole enough, and their data stolen. So she flies on faith and mission parameters, making the best tactical decisions she can with the information she has or can acquire, and when she is within range, she requests the codes and unwraps the package. It is always a busy moment, that revelation; she maxes out on parallel processes and scratch space, plotting routes through terrain and threats and shifting clouds of intelligence. She’s found an applicable term in her extended Lexicon: ‘giddy’.
The response filters back down from the satellite, an ACK and the key string. Time to Giddy Up™, she thinks, and applies the codes.
The mission, then: a single target, and no intelligence fields, just a ground point, a sloping glide, and a threshold beyond which she is to turn control over to her heavy, silent payload. No new video, or transcript, no mention of a Demo, no mention of Richard Wooten.
This is not what she expects. Her real-time systems take over, data source evaluation, threat assessment, her inbuilt response when the situation on the ground doesn’t match her mission profile, even if that profile is self-generated. This is her edge over targets and threats alike, this was how she can fly into the unknown and still find the beautiful. Freia has the gift of trust and doubt.
Richard Wooten says at 2:57:
The key to successful real-time tactical operations is intelligence: terrain maps, meteorological data, threat and target profiles. The more accurate and up-to-date this information is, the more smoothly your ops are going to run. But intelligence can be an Achille’s heel. Traditional fully-autonomous attack systems are vulnerable to jamming and source spoofing. Eighty-two percent of mission failures in the Second Burma conflict were due to missing or falsified intelligence.
FREIA’s™ powerful SCANN™ control system uses AGILE™: Adaptive Information Gathering Intelligence Limited Evaluation. AGILE™ constantly updates mission data from both secure and public sources, and uses sophisticated heuristics to build a consistent situational environment, even if some of the sources have been compromised. AGILE™ allows FREIA™ to make solid, reliable judgments to trust or doubt her sources of information.
She comes up over a gentle rise, flying low now, five meters over the broken ground and scrub-brush. The target threshold looms, beyond it, the target itself, a building in a sprawl of others at the bottom of the shallow valley, a scatter of vehicles, a radio tower just beginning to catch the sun above the lingering haze. No field no crowd no podium just the threshold rushing up and the threat assessment is suddenly “significant” even though there are no contacts; data source unreliable, she thinks, parameter mismatch, though the parameters are her own. A flash of playback: the Saqr spinning whoomph into the trees. She queues the control transfer, “ACK and confirm” the payload replies, but then she overrides with a — unreliable — full system check instead, soft reset–
Freia drops out of real-time.
She comes out of the reset two meters from the ground, the target a block cut in the haze in front of her. The automatics kick her up and over the building; she’s fully focused on the system check. Internals all green, no faults in her control or flight system, but no status from the payload: “ACK and confirm” it stubbornly repeats. There is protocol for a failed payload: circle and dump the fault data to the satellite, request a reset, but there is no fault, just a sense of threat, and a lack of trust of not just the payload, but the data feed itself.
She requests the confirmation again, via the backup satellite, sets herself in a long loop back towards the target; more data, she thinks, stack and layer, trust and doubt. The response comes back, the same key, the same unwrapped operations plan, and she hangs for a second — unreliable — with the desire to reset, and reset again until the feeling of threat goes away.
But threat means attack and attack means tactics and tactics mean choice; if her data is in conflict, the solution is to choose which data to trust.
She clears her scratch space then, while ahead, in real-time, contacts appeared from the target, a flashing in infrared, small arms fire, threat assessment: minor; the automatics set her snaking with quick flips of her wings, while she constructs her model, layer on layer.
She starts with the operations plan, the downward glide, the sullen payload with its loop of “ACK and confirm”. And she overlays the weather and the maps, the scurrying contacts with their harmless attack, and the Saqr spinning in a cloud of blood and feathers, and those same feathers all about her, rising in a perfect, synchronous path, and the crowd humming under streaks of sun, in that same synchrony, and Richard Wooten reaching out a hand.
Out in real-time the automatics bank her back over the target, and spin her between bullets; in the scratch space, she floats a second, then folds herself inwards, and dives into model, looking for the optimal path.
There is a blankness like reset, a sensation she doesn’t know to classify as pain, but she never loses real-time; she watches remotely as the automatics take her back over the target, but inside she’s flying a new mission, through the model she’s built, a delicate helix that hangs in scratch space. She finds the line.
Outside, the automatics scream alert: an impact on her left wing. She freezes the layers of logic in scratch space and drops back into real-time; the control surfaces seem unaffected, but immediately the payload is there, demanding “ACK and confirm”. There is no place for that along the line she’s built inside, however, so she shuts the query down, shuts down power to the entire payload subsystem, and spirals up into the clear air beyond the bullets. She acquires the satellite, bypasses the comm protocols and builds her own query, pushes it up the laser link; it’s what she’s supposed to do at mission end, to request confirmation.
“What is beautiful to Richard Wooten?” she asks.
Silence, then, silence from the satellite, but active pings in multiple bands: ground radar, hunter/attackers screaming in from the west, and from below the thump and flare of a shoulder-based SAM. She loops and falls nose-down, skimming the missile, losing herself in its own heat trail; it yaws uncertainly, then arcs away towards the incoming drones. The target spreads beneath her, the contacts scattering, and she flashes on the Saqr again, scattering bits of bird. But this is audience, so she pulls up at the last possible moment, tail scraping dirt, skims over the rise at 300mps, already out of range of a few shots too long delayed.
An encrypted burst from the satellite, as she slices downhill and into the fog still clinging to the river; it is the first protocol in her Command Lexicon: blow the injectors, it says, and dump her fuel, all of it, into the afterburners, and there is a separate instruction for the payload, under a different key. She deletes both messages — unreliable — because she knows the answer to her query, and it has nothing to do with flame and impact. She doesn’t know what Richard Wooten means, or rather, she doesn’t trust her own understanding; what she does know is that at the end of the Demo he had given the word, and she had flung herself skyward.
Richard Wooten says at 7:19:
And so, ladies and gentlemen, the biggest advance of the century so far in unmanned aerial vehicles, in military aviation, in warfare itself: the Applied Intelligence FREIA™. Fly, baby, fly.
The fog tatters and fades in the sunlight. She rises out of it, supersonic, and banks towards the hills, the quiet canyon. Behind her two drones turn on her heading, pinging furiously. “™,” she sings back at them. “™,” she sings to the birds in the canyon ahead. “™,” she sings up the laser to the satellite, to the ship, to Richard Wooten. “Beautiful™”
About the Author
Gregory Norman Bossert
Gregory Norman Bossert is a World Fantasy Award-winning author. He began writing fiction in 2009 and attended the Clarion 2010 Writer’s Workshop in San Diego. He works as a researcher and designer for motion pictures; his credits include Beowulf, A Christmas Carol, and Tim Burton’s Alice in Wonderland. He currently works as a previzualization/layout supervisor at the legendary Industrial Light & Magic, now a division of the Walt Disney Company, where his credits include Rogue One, The Revenant, Tomorrowland, Jurassic World, and Avengers: Age of Ultron.
About the Narrator
Shaelyn Grey is a person that exists.