#Data Input Operator

[omdataentryindia]

5 Major Benefits Of Outsourcing Data Input Services

Businesses continuously seek methods to improve overall efficiency and streamline their operations in the ever-changing corporate landscape. Outsourcing data input services is one technique that has grown in favor recently. Businesses can focus on their core strengths and gain advantages by assigning data-related work to specialist service providers. This blog post will look at the five main advantages of outsourcing data input services.

Understanding Data Input Services

Data input refers to collecting data from a variety of sources, such as digital and creating catalogs, or it may also entail conducting online research. Businesses have to collect data according to their needs and use it to enhance corporate growth, as an abundance of data is being contributed to the digital environment every second. Thus, one of the most in-demand services in the modern world is data input. This is due to two factors: First of all, numerous businesses lack the continuous time and focus needed for data entry.

The process of entering, updating, or verifying data into a system or database is the focus of data entry services. A vast variety of data, including client information, financial transactions, inventory records, survey results, and more, may be included in this. Ensuring that timely, accurate, and full data is entered into a system for various business purposes is the aim of data input services.

Why Should We Go For Outsourcing Data Input Services?

When you outsource the data input services, it gives you the leverage to spend your quality time in other core business areas and reduce your data management costs along with infrastructure costs. Second, you also can manage your data efficiently with expert help.

Businesses often outsource data input services to specialized providers to streamline Business processes and ensure efficiency. Outsourcing can be particularly beneficial for time-consuming tasks and require attention to detail, allowing organizations to focus on their core competencies while the outsourced service providers handle the data-related tasks.

Categories Of Outsourcing Data Input Services

Data input from paper documents/document data entry

Data input from web-based applications

Data input from forms/resumes

Data input from client software/application

Data input from catalog/catalog data entry

Data input from business cards

Data input from OCR

Data enumeration/data keying/data capturing

Data input from images/image data entry

Data input from one file format to others

Let's Explore The 5 Major Benefits Of Outsourcing Data Input Services

By entrusting data-related tasks to specialized service providers, businesses can focus on their core competencies and achieve significant advantages. Here, we will discuss about five major benefits of outsourcing data input services:

1. Cost Savings:

Businesses can save a lot of money by outsourcing data input services. By leveraging the expertise of external service providers, companies can avoid the expenses associated with hiring and training in-house staff. 

Additionally, outsourcing eliminates the need for investing in infrastructure, technology, and ongoing maintenance. This cost-effective approach enables organizations to allocate resources more efficiently and redirect funds toward critical business functions.

Without committing to a long-term agreement, you can scale up or down your use of outsourced services and simply pay for the task completed.

2. Access To Specialized Skills And Technology:

Expert outsourcing partners frequently have cutting-edge technology and specialized knowledge specifically devoted to data input services. These suppliers make sure that your data processes take advantage of state-of-the-art solutions by keeping up with the most recent developments in the industry. 

By tapping into this expertise, businesses can improve the accuracy, speed, and quality of their data input processes without the need for continuous in-house training and technology investments.

Accounting data entry services are the area of expertise for outsourcing providers, who also possess the necessary skills and experience to manage extensive financial transactions.

3. Increased Efficiency And Flexibility:

Outsourcing data input services offers unparalleled scalability and flexibility. Businesses can easily adjust the level of services based on their fluctuating needs. Whether there is a sudden increase in data volume or a temporary decrease in workload, outsourcing provides the agility to scale services up or down accordingly. This flexibility ensures that companies maintain optimal efficiency without being burdened by unnecessary overhead costs during slow periods. 

Outsourcing can lead to faster turnaround times for tasks like product catalog updates, order processing, and data quality control.

Online data entry services can lead to faster, more efficient, and improved database management.

4. Risk Mitigation And Data Confidentiality:

In the present digital era, data security is of utmost importance to companies. Reputable suppliers' data input services frequently include integrated security features and regulatory requirements. Strong security measures are put in place by seasoned outsourcing partners to safeguard private data and lower the possibility of hacking.

Additionally, these providers regularly update their systems to stay compliant with industry regulations, offering businesses peace of mind regarding data security and privacy.

Data security and privacy are given top attention by Indian outsourcing businesses, and employees are required to maintain high data security under the Non-Disclosure Agreement (NDA).

5. Focus On Core Competencies:

Collecting data takes a lot of time and might take focus and resources away from a business's primary strengths. Businesses can focus on their main goals, such as product development, marketing, and customer support, by outsourcing data input services. Organizations can improve their overall productivity and strategic focus by assigning regular data work to others.

Conclusion:

In conclusion, outsourcing data input services is a strategic move for businesses looking to enhance efficiency, reduce costs, and stay focused on core competencies. By partnering with specialized service providers, organizations can leverage expertise, access advanced technologies, and ensure scalability while mitigating risks associated with data security. In the dynamic business landscape, outsourcing data input services emerges as a valuable solution for companies seeking a competitive edge and sustainable growth.

Source Of: https://dataentrywiki.blogspot.com/2024/02/5-major-benefits-of-outsourcing-data-input-services.html

[romerona]

Ethera Operation!!

You're the government’s best hacker, but that doesn’t mean you were prepared to be thrown into a fighter jet.

Bradley "Rooster" Bradshaw x Awkward!Hacker! FemReader

Part II

You knew today was going to suck the second your alarm went off and you briefly, genuinely, considered faking your own death.

Not in a dramatic, movie-worthy kind of way. No, more like… vanish-into-a-data-breach, throw-your-phone-in-the-ocean, start-a-new-life-in-Finland sort of way.

But instead, you got up.

Because apparently, national security outranks your crippling fear of flight—not that it makes the simulator any less hellish, with its cold metal, stale coffee, and that faint chemical tang of fear.

You were strapped into the rear seat of a flight simulation pod, hands locked in your lap like they might betray you at any moment and start mashing random buttons. You exhaled slowly as your eyes flicked across the control panel. So many switches. So many lights. Half of them blinked like they were mocking you. The other half were labeled with words like “altitude” and “engine throttle” and “eject.”

Great.

You adjusted your headset as the technician’s voice crackled through. “Sim will start in thirty seconds, Doctor. We’ll be monitoring vitals and control input from the tower."

You forced a nod, even though your stomach was already trying to escape through your spine. Your breath fogged the inside of the visor. You clutched the tablet tethered to your vest like it was a stuffed animal and you were six years old again.

“Try not to scream this time,” came Cyclone’s voice through the comms, calm and flat like he was asking you to pass the salt.

You offered a shaky thumbs-up that somehow felt more like a surrender flag.

The sim operator spoke next, voice crackling through your headset once again. “Doctor, your objective is to remain conscious, keep your hands away from the panel, and activate the Ethera interface when prompted. We’ll simulate turbulence, evasive maneuvers, and mild G-force changes. Ready?”

No. Never.

“...Sure.”

The sim lurched forward with a roar, and your whole body snapped back into the seat. You let out a startled “whuff!”, eyes wide, heart in your throat. The room around you—walls disguised as sky—blurred as the machine banked hard to the left.

“OhmyGodohmyGodohmyGOD—”

There was no gentle start. No soft acceleration to get your bearings. Just a violent jolt forward, and then you were climbing—straight up, like gravity had been turned into a weapon and pointed directly at your lungs.

Pressure slammed into your chest. The world outside the cockpit blurred. You couldn’t hear anything except your own heartbeat.

“WHY ARE WE TILTING—”

“Initiating evasive pattern,” came the tech’s voice, calm as ever.

The sim jerked again, this time into a sharp roll. The world flipped sideways. Your ears popped. Something primal in your brain screamed: This is how you die.

Your ears were ringing. Your pulse thundered against your ribs. Somewhere beneath the pressure and panic, you could hear the tech’s voice cutting in again—calm, detached, and utterly unhelpful.

“Doctor, you need to deploy the program,” he said. “Fifty seconds. Starting now.”

Oh, shit, you couldn’t even see straight.

Your breath came in short, shallow gasps as the simulated jet banked hard to the right, pressing your spine into the seat like it wanted to keep it. The G-forces made your vision tunnel, your stomach lurching somewhere around your throat.

Your hand fumbled toward the tablet mount, fingers shaking so hard they were basically useless. You tapped the corner of the screen. Missed. Tapped again. The jet jolted. The tablet shifted. Your palm slammed into the side instead of the input.

Forty seconds.

The Ethera prompt blinked up at you—green, glowing, go—but it may as well have been a mirage. You squinted through the dizziness, swore under your breath in three languages, and tried again.

Thirty-five.

The turbulence kicked again, harder. Your chest seized. The tablet slipped slightly in its latch. You tapped the input.

Too late.

“Simulation failed,” the system announced flatly. “Target missed.”

Everything halted—the motion, the noise—everything except your pulse, which pounded on like it hadn't gotten the memo.

The sim pod cracked open with a sharp hiss, releasing a rush of cool air that hit your sweat-slicked skin like a slap to the face. You didn’t move. For a second too long, you just sat there, fingers clenched around the armrests like they were the only things keeping you from unraveling completely. The silence pressed in, thick with the weight of your own embarrassment, humiliation settling low and heavy in your gut like a stone.

Your fingers fumbled at the release on your helmet, hands still trembling from the G-forces and adrenaline. The inside of your mouth tasted like copper and failure. You tugged off the headset next, wires dragging like they were reluctant to let go. Everything felt too loud and too quiet at the same time.

Your boots scraped against the cold floor as you shakily swung your legs out, and there he was, Vice Admiral Beau Simpson, standing with arms crossed, expression carved from steel.

You wanted to disappear into the floor.

He didn’t speak right away. He just looked at you. Not angry. Not even disappointed. Just… calculating. Like he was already assessing the cost of putting you on a real jet.

“I missed the mark,” you said first, because silence felt worse. “I know.”

Cyclone gave a short nod, like that much at least didn’t need explaining. “You froze.”

You exhaled slowly, willing your heart to stop trying to beat its way out of your ribs. “Yeah.”

His eyes didn’t waver. “You had a job. Not to fly. Not to fight. Just to stay calm. Deploy your program.”

“I know.”

“And you failed.”

You stood on legs that didn’t feel like they belonged to you, one hand gripping the edge of the simulator for balance, the other still clutching the edge of the tablet even though the prompt had long since vanished.

“If this had been real,” he continued, “that satellite would still be feeding your government false intelligence. That jet would’ve been intercepted. And you, Doctor, would’ve been dead, and so would've your pilot.”

You flinched. Not visibly—hopefully—but the words hit harder than they should have. You stared at the scuffed metal floor, heart thudding against your ribs.

“You’re not a soldier,” he said. “And you’re not trained for this. That’s clear.”

You opened your mouth—maybe to apologize, maybe to defend yourself—but he raised a hand, cutting you off with one sharp motion.

“That’s not an excuse,” he added, voice sharp. “It’s a reality. One you’ll have to overcome, and fast. I don’t expect perfection but I do expect progress. And I expect you to walk into that sim tomorrow knowing what you did wrong—and ready to fix it.”

You blinked hard, your pulse pounding in your ears. “Yes, sir.”

Cyclone gave you one last look—disappointed, but not hopeless—and then turned, then paused, glancing back.

“And see medical,” he added, almost as an afterthought. “You’re pale as hell.”

Then he walked away, boots echoing down the corridor, leaving you standing there with a spinning head, a shattered ego and the feeling of wanting to curl up and cry.

As you moved to make your way toward medical—because yes, apparently nausea, disorientation, and a near-death experience weren’t enough on their own— you skidded to a stop just short of slamming into a very broad chest.

Of course. Of course, it was him.

The handsome, mustached pilot. The one who’d handed you your tablet like it was a glass slipper, back in the briefing room. The one who hadn’t laughed when you dropped it, but definitely thought about it.

His hair was slightly mussed, curls pushed back from his forehead like he’d run a hand through them one too many times. He held two water bottles, one in each hand, like he wasn’t sure if he meant to stay—or if he’d just pretend this was a casual “what a surprise” moment if anyone asked.

You froze. He straightened.

“Hey,” he said, voice softer than you expected. A lot softer than earlier. Less smirk, more... sincerity.

“Uh… hi,��� you said finally. Nailed it. Pure elegance.

His expression didn’t change much, maybe just a flicker of amusement at the corners of his mouth. He held out one of the bottles. “You looked like you could use this.”

You hesitated—more from surprise than anything else—then took it. You took it, fingers brushing his as you did. His skin was warm—too warm for how cold you felt. You tried not to notice.

“Thanks,” you said quietly, unscrewing the cap with hands that still trembled, ever so slightly. The water was blissfully cold against your throat, but it did nothing for the embarrassment still curdling in your stomach.

“You okay?” he asked, his voice gentler than you expected.

You hesitated, then tilted your head in a noncommittal shrug. “Define okay.”

A ghost of a smile touched his face. “Not crying, not puking, not passed out? That’s the general baseline.”

You cracked a reluctant laugh. “Oh, sure, I’m totally thriving.”

He nodded once, and the silence settled again—less awkward now, more… charged. The kind of quiet that hummed between words. The kind that made your skin feel too tight.

He looked like he might leave, but then he didn’t.

Instead, he shifted his weight, adjusting his grip on the second water bottle like it was some kind of anchor or maybe just something to do with his hands while he said, “You weren’t terrible in there.”

Your stomach jolted—sharp, unexpected. Like missing a step on the stairs. Heat bloomed beneath your collar, crawling up your throat as your fingers tightened around the plastic water bottle.

“You��” Your voice cracked a little, and you cleared your throat. “You were watching?”

God. No.

Why did you ask that? Why would you ever want confirmation?

His expression shifted—just slightly. Not quite sheepish, not quite smug. Just something in the middle.

“I was passing by,” he said, entirely too casual.

You groaned softly, dragging a hand over your face. “Fantastic. I didn’t just humiliate myself in front of the brass. I also had an audience.”

“Don’t take it personally,” he said, his voice laced with something between amusement and sincerity. “We’ve all been there.”

You raised an eyebrow. “In a classified sim seat with national security riding on your ability to not pass out?”

He grinned wider. “Well. Maybe not exactly there.”

You scoff, shaking your head as you take another sip of the water.

“You’re not supposed to get it right the first time." He said, "No one does. You think the rest of us were born knowing how to pull 7 Gs without losing our lunch?”

You didn’t answer. Not because you didn’t believe him—maybe part of you even did—but because if you opened your mouth, you weren’t sure if it would come out as a laugh or a cry.

He noticed.

“You know, most people don’t get in the backseat of a fighter jet without years of prep. You? You've got a couple of days, a tech background, and a pulse. That’s it and you still got in. That counts for something.”

You stared at him. “Why do you even care if I mess this up?”

He looked at you then, long and quiet.

“You built something that could change the world,” he said with an easy shrug. “That kind of genius doesn’t come with an eject handle. So yeah. I care.”

You looked away fast, suddenly too aware of how warm your cheeks were.

He leaned back again, casual as ever. “Besides, if I'm the one you are gonna fly into enemy territory, I’d rather know you’re not gonna scream the whole time.”

You snorted. “I’ll scream quietly. Into my elbow. Like an adult.”

He chuckles and you looked at him. Really looked at him. Still in partial uniform, flight suit unzipped to the waist, sleeves tied and hanging loose around his hips. His shirt clung to his chest, slightly sweat-damp at the collar, and that damn mustache made him look both out-of-place and weirdly grounded at the same time.

He wasn’t just handsome. He was kind of infuriatingly steady.

“Can I—” You paused, surprised by your own voice. “Can I ask your name?”

His brows lifted, just slightly, like the question had caught him off guard. But then he shifted forward and extended a hand—open, easy, completely steady in a way that you most definitely weren’t.

“Bradley Bradshaw,” he said. “But most people around here call me Rooster.”

You blinked. “Rooster?”

A grin tugged at his mouth, soft and lopsided. “My call sign. It’s a long story.”

You hesitated for a beat, then reached out and slid your hand into his.

His palm was warm—really warm—and calloused in a way that made you feel every inch of the difference between your worlds. His grip was firm but not overwhelming, grounding. Like he knew exactly how much pressure to apply without overdoing it. His fingers curled around yours with quiet confidence, like this was nothing, like it didn’t send an unexpected little jolt of awareness all the way up your arm.

Your hand was smaller than his, your skin cooler, trembling just enough that you hoped he didn’t notice—but something in the way his thumb shifted, just the tiniest bit, made you think maybe he did.

You weren’t sure how long you held on. Long enough to register the strength in his hand, the steadiness, the solidness of someone who lived in the sky but was somehow more grounded than anyone you knew.

“Y/N L/N,” you said finally, your voice softer now. "But I guess you already knew that.”

He gave a small nod, his eyes not leaving yours. "You're hard to forget,"

You didn’t let go right away.

Neither did he.

Then, as if realizing the moment was hanging just a second too long, you both released at the same time—too quickly. Like a secret exchanged and immediately tucked away.

You took a half step back, pulse thrumming in your throat, fingers still tingling from the contact.

Bradley, however, didn’t step away immediately instead, he lingered for just a second longer, watching you with a look that wasn’t teasing or cocky or smug. Just something quiet and steady, then he smiled—small, crooked, the kind that didn’t feel all that teasing but still carried that glint of mischief behind it. The kind of smile that said he saw more than he let on.

“You’ll get it,” he said, voice softer now. “Not today. Maybe not tomorrow.”

His eyes flicked to yours, and something about the way he looked at you—like he meant it, like he believed it, made your chest tighten.

“But you will.”

You opened your mouth, unsure what you were about to say—maybe thank you, maybe don’t say that unless you mean it—but the words never quite made it past your lips.

Because Bradley gave you one last look, a flick of something unreadable in his eyes, then turned down the corridor, water bottle still swinging lazily from his fingers while you stood there for a moment, then finally exhaled. “Okay,”

Days went faster than you were ready for.

You hadn’t slept much. Not from fear exactly, though there was plenty of that still hanging around like a ghost in your chest—but more from the afterglow of adrenaline. The kind that leaves your body tired but your mind racing.

You’d replayed Bradley's words a dozen times. You’ll get it. You weren’t sure if they’d stuck because you believed them… or because you wanted to.

But when you arrived at the simulator bay, you were expecting to meet with Cyclone, just like every other day, but he wasn't there waiting for you.

It was a new pilot.

She stood near the simulator controls, arms crossed loosely over her chest, already in her flight suit, her expression somewhere between mildly unimpressed and genuinely curious.

“You’re my new project, huh?” she said as you approached.

You blinked. “Um. I—guess so?”

“I’m your point of contact now,” Phoenix said, nodding toward the simulator. “Cyclone thought a different approach might help. And I volunteered.”

You tried not to look too relieved. But you were. God, you were. Cyclone, well, he was rough, for lack of better words, Rooster had been kind, yes, but his presence was a lot. Intense. Distracting.

Phoenix, on the other hand, had that kind of practical, no-nonsense confidence you could actually lean on. She didn’t feel like a storm waiting to happen. She felt like structure.

“I’m Lieutenant Natasha Trace,” she said, extending her hand. “Call sign’s Phoenix.”

You shook her hand, your grip steadier than yesterday—though your palm was still a little clammy, and you were pretty sure she noticed.

“Y/N,” you said, then added with a tired smile, “Doctor. Uh, the nervous one.”

Phoenix huffed out a short laugh, a glint of something sharp but not unkind in her eyes. “I read your file.”

She stepped back, folding her arms as she leaned one hip against the edge of the sim console. Her stance was relaxed, confident, comfortable in her own skin in the way only someone who’d already proven themselves a hundred times could be.

“I also watched your sims,” she added, voice casual.

You winced, your smile turning into a grimace. “Oof. That bad?”

She tilted her head, as if considering how honest she wanted to be. Then gave a light shrug, eyes steady on yours. “I’ve seen worse. A lot worse.”

You let out a low hum, arms crossing loosely over your chest in mock thought. “That’s… reassuring.”

“Isn’t it?” she said, with just enough of a smirk to make you feel like she was on your side. “You hadn't passed out nor puked. You followed instructions until your brain short-circuited. Classic first-timer move.”

You laughed under your breath, surprised at how easily it came.

She finally looked at you then—steady, knowing. “We’re not here to make you into a pilot, Doc. We just need you ready for the mission. The rest? We’ll cover you.”

Something in your chest loosened at that.

Support. No condescension. No sharp edges. Just a quiet kind of strength you could lean against.

“Thanks,” you said. “Really.”

Phoenix nodded once. “Let’s get you in the seat.”

Inside the simulator, everything felt smaller than you remembered.

Not physically—just heavier. Like the air had thickened, like the walls had learned your fears from yesterday and decided to lean in a little closer.

You sat in the back seat again, the tablet already secured to its mount beside your right leg. Your fingers hovered near it, not quite touching, like it might bite. You could already feel your heartbeat in your palms.

“Straps secured?” Phoenix’s voice crackled through the headset. Her tone was crisp, even, the kind that didn’t rise to meet panic—it smothered it before it started.

You exhaled and gave a tight nod, forgetting she couldn’t see it. “Y-Yeah. Good to go.”

“All right,” she said. “We’re starting slow. Just basic turbulence patterns. No evasive maneuvers, no tricks. You’re not here to impress anyone. You’re here to breathe, and press a single button when I tell you.”

You nodded again, this time speaking aloud. “Sure.”

The sim hummed to life around you, and your body tensed automatically—like it remembered what came next, even if you swore it wouldn’t be that bad.

“Relax your shoulders,” Phoenix said, as if she felt the stiffness from her end. “You’re holding tension like you’re about to punch the air.”

The screen in front of you blinked to life. The sim took you airborne, but the motion was slow this time—steady, like a calm climb on a commercial flight.

You forced yourself to breathe out slowly and unclenched your jaw, trying to follow her lead. The shaking wasn’t nearly as bad as the previous day's simulated madness. No rolls. No sharp drops. Just steady pressure. Unnerving, but survivable.

Your eyes flicked to the screen.

The prompt glowed softly. Ethera. Standing by. Timer: 02:00

“This is just a systems check,” Phoenix said. “You don’t have to engage. Just keep your eyes on it. Notice the screen, your pulse, your breath. You’ve got time."

The pod dipped gently into a banking curve. You swayed, stomach flipping. "Keep breathing, Doc."

You gripped the edge of the seat, fingers twitching. “This still counts as breathing, right?”

“As long as you’re not blue in the face, yeah.”

You smiled—barely—but it helped.

The Ethera interface activated on the mounted tablet in front of you. The same prompt, The countdown. You glanced at it and your heart gave one uneasy thud.

“Don’t rush,” Phoenix reminded you, voice even. “One thing at a time. Don’t try to win. Just try to finish.”

You nodded again, reaching out slowly—deliberately—and tapped the screen to begin the simulated deployment sequence. The code began to unfold, and the sim didn’t break into loops or chaos. It kept going. And you were still breathing.

Your hand trembled slightly, but you stayed focused, eyes on the sequence as it loaded in steady green waves. The turbulence passed. The sim steadied.

“Ten seconds,” Phoenix said. “You’ve got it. Keep it locked.”

You kept your hand on the panel. You didn’t blink. The screen counted down.

3… 2… 1…

Deployment successful.

The soft chime of success echoed in your headset.

“Target received,” the system confirmed.

You blinked, then blinked again. “I… I got it?”

“You got it,” Phoenix said, the faintest edge of pride in her voice. “Nice and clean.”

You slumped back in the seat, suddenly aware of just how hard your heart had been working. Your eyes stung—not from panic this time, but from sheer relief.

“Doctor,” Phoenix said after a beat. “That was not bad.”

You couldn’t help the grin that broke across your face, exhausted but real.

And when the pod finally powered down with a gentle thunk, and the hatch hissed open, you realized you’d done the whole thing without white-knuckling the seat.

You’d finally made it through.

Phoenix was waiting for you, arms crossed, leaning one hip against the console like she’d known all along you’d handle it.

You stepped out, legs a still stiff, but your head was clear.

“Not bad,” she said, and this time her smile wasn’t just professional. It was small, but real. “No ejections. No nausea. No hysterics.”

You let out a dry laugh, breath catching on the edge of it. “Just mild existential dread.”

She shrugged, cool as ever. “That’s standard issue.”

Then smiled—really smiled—for the first time since this whole classified, terrifying, completely-out-of-your-depth mission had begun. The kind of smile that pulled dimples you hadn’t felt in days.

“Thanks,” you said again, quieter this time. Not just for the training, but for not making you feel like a burden.

Phoenix nodded once, like she already understood all of that.

“Don’t get too comfortable,” she said. “We need to move faster. Real evasive sequences. Simulated pressure. Maybe even some yelling.”

“Yours or mine?”

She smirked. “We’ll see who breaks first.”

You laughed again—easier this time—and for the first time, it didn’t feel like you were pretending.

By the time the week came to an end, you and Phoenix had become friends.

Not in the polite, nod-in-the-hallway kind of way—but the real kind. The kind built through shared silence in the simulator bay, through low chuckles after a successful run, through Phoenix’s calm voice in your headset, cutting through the static and the fear. She never coddled you. Never sugarcoated anything but she never made you feel less, either.

There were moments where fear absolutely took over—where your breath hitched too high in your chest or your fingers trembled too much to find the prompt in time and there were other moments, rarer but growing, where you managed. Where you pressed the button, where you kept your head above water.

Phoenix never made a spectacle of either.

When you panicked, she talked you down, when you succeeded, she just clapped you on the shoulder, tossed you a bottle of water, and said, “Told you. You’re getting it.”

And somehow, that meant more than any standing ovation ever could.

By Friday evening, you had survived four more simulations, logged two successful Ethera deployments, and stopped referring to the ejection lever as “that red death stick.”

Progress.

“You coming to the Hard Deck tonight?” Phoenix said casually, already slinging her duffel over one shoulder as you both headed toward the lockers.

You blinked at her, caught off guard. “What?”

She paused mid-step, turning just enough to glance back at you with that crooked grin she reserved for moments like this—half dare, half invitation.

“The Hard Deck,” she repeated, now walking backward toward the hangar doors. “Bar. Pool tables. Bad decisions. You in?”

You stared for a beat too long, processing.

The Hard Deck.

You opened your mouth. Closed it. You’d heard about the place in passing—mostly through muttered comments and laughing threats. It had sounded like a local haunt. Loud. Messy. Full of people who knew exactly what they were doing and didn’t care that you didn’t.

“Wait, is that—like, is that a thing?” you asked, trailing after her. “Do people… actually go?”

Phoenix raised an eyebrow like she wasn’t sure if you were messing with her. “Only the ones worth talking to.”

You hesitated.

She paused at the doorway and tossed the final hook. “You’ve survived a week of sims, didn’t puke on anyone, and haven’t cried once. That makes you officially less pathetic than half the new guys. You’ve earned a drink... So?

Your brain, naturally, tried to stall. A bar? With actual people? And more pilots? But your mouth moved faster.

“Uh—yeah, sure,” you said quickly, the words tumbling out before your usual social panic could hit. “I could go for a drink.”

Phoenix gave a little nod, like she’d already known your answer. Like this was the inevitable next step in whatever strange, reluctant journey you’d found yourself on.

Then she jerked her chin toward the exit, already on the move.

You hesitated. “What now?”

She didn’t stop walking.

“You go back to wherever you’ve been hiding, put on something that doesn’t scream ‘high-stress lab goblin,’ and I’ll swing by in an hour.”

You blinked. “That specific, huh?”

Phoenix half-turned, walking backward again like she had a personal vendetta against stationary conversations. “It’s a bar, not a Senate hearing. No briefing, no simulations, no threat of fiery death. Just drinks. Loud music. Maybe pool. Probably bad flirting.”

And with that, she was gone—leaving you standing in the middle of the hangar, sweaty, slightly stunned, and suddenly very aware that you owned exactly one outfit that wasn’t issued or work-adjacent.

Oh no. Now you actually had to get ready.

A/N:

Heyyyyy, OMG the support for this story is wild, thank you all so so muchhh!! I honestly did not think it would get this much attention, my first draft was actually a Charlie's Angel reader lol, but I'm so happy you all enjoy this version. I did try to make it as realistic as possible, after all reader does not like to fly I can only imagine being put in her position, so she being frozen out of fear and not completing the mission feels real, at least to me.

And my apologies it took me so long to put it out. Part III is already in the works, so I think it will be out soon.

Thank you all so so much for the support and the comments and reblogs, really.

Tags:

hangmanscoming

leesumii

glowingtree

isla-finke-blog

milkyasteroids

notaceventura

djappleblush

hipsternerd9

impossibleblizzardstudentposts

thesoftdumbass

imineveryfandomever

st4rgirlmar1e

malindacath asked:

lonelysoul50

xozchi

nerdgirljen

kakeurillon

softpia

Please tell me if you want to be tagged.

[txttletale]

Saw a tweet that said something around:

"cannot emphasize enough how horrid chatgpt is, y'all. it's depleting our global power & water supply, stopping us from thinking or writing critically, plagiarizing human artists. today's students are worried they won't have jobs because of AI tools. this isn't a world we deserve"

I've seen some of your AI posts and they seem nuanced, but how would you respond do this? Cause it seems fairly-on point and like the crux of most worries. Sorry if this is a troublesome ask, just trying to learn so any input would be appreciated.

i would simply respond that almost none of that is true.

'depleting the global power and water supply'

something i've seen making the roudns on tumblr is that chatgpt queries use 3 watt-hours per query. wow, that sounds like a lot, especially with all the articles emphasizing that this is ten times as much as google search. let's check some other very common power uses:

running a microwave for ten minutes is 133 watt-hours

gaming on your ps5 for an hour is 200 watt-hours

watching an hour of netflix is 800 watt-hours

and those are just domestic consumer electricty uses!

a single streetlight's typical operation 1.2 kilowatt-hours a day (or 1200 watt-hours)

a digital billboard being on for an hour is 4.7 kilowatt-hours (or 4700 watt-hours)

i think i've proved my point, so let's move on to the bigger picture: there are estimates that AI is going to cause datacenters to double or even triple in power consumption in the next year or two! damn that sounds scary. hey, how significant as a percentage of global power consumption are datecenters?

1-1.5%.

ah. well. nevertheless!

what about that water? yeah, datacenters use a lot of water for cooling. 1.7 billion gallons (microsoft's usage figure for 2021) is a lot of water! of course, when you look at those huge and scary numbers, there's some important context missing. it's not like that water is shipped to venus: some of it is evaporated and the rest is generally recycled in cooling towers. also, not all of the water used is potable--some datacenters cool themselves with filtered wastewater.

most importantly, this number is for all data centers. there's no good way to separate the 'AI' out for that, except to make educated guesses based on power consumption and percentage changes. that water figure isn't all attributable to AI, plenty of it is necessary to simply run regular web servers.

but sure, just taking that number in isolation, i think we can all broadly agree that it's bad that, for example, people are being asked to reduce their household water usage while google waltzes in and takes billions of gallons from those same public reservoirs.

but again, let's put this in perspective: in 2017, coca cola used 289 billion liters of water--that's 7 billion gallons! bayer (formerly monsanto) in 2018 used 124 million cubic meters--that's 32 billion gallons!

so, like. yeah, AI uses electricity, and water, to do a bunch of stuff that is basically silly and frivolous, and that is broadly speaking, as someone who likes living on a planet that is less than 30% on fire, bad. but if you look at the overall numbers involved it is a miniscule drop in the ocean! it is a functional irrelevance! it is not in any way 'depleting' anything!

'stopping us from thinking or writing critically'

this is the same old reactionary canard we hear over and over again in different forms. when was this mythic golden age when everyone was thinking and writing critically? surely we have all heard these same complaints about tiktok, about phones, about the internet itself? if we had been around a few hundred years earlier, we could have heard that "The free access which many young people have to romances, novels, and plays has poisoned the mind and corrupted the morals of many a promising youth."

it is a reactionary narrative of societal degeneration with no basis in anything. yes, it is very funny that laywers have lost the bar for trusting chatgpt to cite cases for them. but if you think that chatgpt somehow prevented them from thinking critically about its output, you're accusing the tail of wagging the dog.

nobody who says shit like "oh wow chatgpt can write every novel and movie now. yiou can just ask chatgpt to give you opinions and ideas and then use them its so great" was, like, sitting in the symposium debating the nature of the sublime before chatgpt released. there is no 'decay', there is no 'decline'. you should be suspicious of those narratives wherever you see them, especially if you are inclined to agree!

plagiarizing human artists

nah. i've been over this ad infinitum--nothing 'AI art' does could be considered plagiarism without a definition so preposterously expansive that it would curtail huge swathes of human creative expression.

AI art models do not contain or reproduce any images. the result of them being trained on images is a very very complex statistical model that contains a lot of large-scale statistical data about all those images put together (and no data about any of those individual images).

to draw a very tortured comparison, imagine you had a great idea for how to make the next Great American Painting. you loaded up a big file of every norman rockwell painting, and you made a gigantic excel spreadsheet. in this spreadsheet you noticed how regularly elements recurred: in each cell you would have something like "naturalistic lighting" or "sexually unawakened farmers" and the % of times it appears in his paintings. from this, you then drew links between these cells--what % of paintings containing sexually unawakened farmers also contained naturalistic lighting? what % also contained a white guy?

then, if you told someone else with moderately competent skill at painting to use your excel spreadsheet to generate a Great American Painting, you would likely end up with something that is recognizably similar to a Norman Rockwell painting: but any charge of 'plagiarism' would be absolutely fucking absurd!

this is a gross oversimplification, of course, but it is much closer to how AI art works than the 'collage machine' description most people who are all het up about plagiarism talk about--and if it were a collage machine, it would still not be plagiarising because collages aren't plagiarism.

(for a better and smarter explanation of the process from soneone who actually understands it check out this great twitter thread by @reachartwork)

today's students are worried they won't have jobs because of AI tools

i mean, this is true! AI tools are definitely going to destroy livelihoods. they will increase productivty for skilled writers and artists who learn to use them, which will immiserate those jobs--they will outright replace a lot of artists and writers for whom quality is not actually important to the work they do (this has already essentially happened to the SEO slop website industry and is in the process of happening to stock images).

jobs in, for example, product support are being cut for chatgpt. and that sucks for everyone involved. but this isn't some unique evil of chatgpt or machine learning, this is just the effect that technological innovation has on industries under capitalism!

there are plenty of innovations that wiped out other job sectors overnight. the camera was disastrous for portrait artists. the spinning jenny was famously disastrous for the hand-textile workers from which the luddites drew their ranks. retail work was hit hard by self-checkout machines. this is the shape of every single innovation that can increase productivity, as marx explains in wage labour and capital:

“The greater division of labour enables one labourer to accomplish the work of five, 10, or 20 labourers; it therefore increases competition among the labourers fivefold, tenfold, or twentyfold. The labourers compete not only by selling themselves one cheaper than the other, but also by one doing the work of five, 10, or 20; and they are forced to compete in this manner by the division of labour, which is introduced and steadily improved by capital. Furthermore, to the same degree in which the division of labour increases, is the labour simplified. The special skill of the labourer becomes worthless. He becomes transformed into a simple monotonous force of production, with neither physical nor mental elasticity. His work becomes accessible to all; therefore competitors press upon him from all sides. Moreover, it must be remembered that the more simple, the more easily learned the work is, so much the less is its cost to production, the expense of its acquisition, and so much the lower must the wages sink – for, like the price of any other commodity, they are determined by the cost of production. Therefore, in the same manner in which labour becomes more unsatisfactory, more repulsive, do competition increase and wages decrease”

this is the process by which every technological advancement is used to increase the domination of the owning class over the working class. not due to some inherent flaw or malice of the technology itself, but due to the material realtions of production.

so again the overarching point is that none of this is uniquely symptomatic of AI art or whatever ever most recent technological innovation. it is symptomatic of capitalism. we remember the luddites primarily for failing and not accomplishing anything of meaning.

if you think it's bad that this new technology is being used with no consideration for the planet, for social good, for the flourishing of human beings, then i agree with you! but then your problem shouldn't be with the technology--it should be with the economic system under which its use is controlled and dictated by the bourgeoisie.

[ecrivainsolitaire]

A summary of the Chinese AI situation, for the uninitiated.

These are scores on different tests that are designed to see how accurate a Large Language Model is in different areas of knowledge. As you know, OpenAI is partners with Microsoft, so these are the scores for ChatGPT and Copilot. DeepSeek is the Chinese model that got released a week ago. The rest are open source models, which means everyone is free to use them as they please, including the average Tumblr user. You can run them from the servers of the companies that made them for a subscription, or you can download them to install locally on your own computer. However, the computer requirements so far are so high that only a few people currently have the machines at home required to run it.

Yes, this is why AI uses so much electricity. As with any technology, the early models are highly inefficient. Think how a Ford T needed a long chimney to get rid of a ton of black smoke, which was unused petrol. Over the next hundred years combustion engines have become much more efficient, but they still waste a lot of energy, which is why we need to move towards renewable electricity and sustainable battery technology. But that's a topic for another day.

As you can see from the scores, are around the same accuracy. These tests are in constant evolution as well: as soon as they start becoming obsolete, new ones are released to adjust for a more complicated benchmark. The new models are trained using different machine learning techniques, and in theory, the goal is to make them faster and more efficient so they can operate with less power, much like modern cars use way less energy and produce far less pollution than the Ford T.

However, computing power requirements kept scaling up, so you're either tied to the subscription or forced to pay for a latest gen PC, which is why NVIDIA, AMD, Intel and all the other chip companies were investing hard on much more powerful GPUs and NPUs. For now all we need to know about those is that they're expensive, use a lot of electricity, and are required to operate the bots at superhuman speed (literally, all those clickbait posts about how AI was secretly 150 Indian men in a trenchcoat were nonsense).

Because the chip companies have been working hard on making big, bulky, powerful chips with massive fans that are up to the task, their stock value was skyrocketing, and because of that, everyone started to use AI as a marketing trend. See, marketing people are not smart, and they don't understand computers. Furthermore, marketing people think you're stupid, and because of their biased frame of reference, they think you're two snores short of brain-dead. The entire point of their existence is to turn tall tales into capital. So they don't know or care about what AI is or what it's useful for. They just saw Number Go Up for the AI companies and decided "AI is a magic cow we can milk forever". Sometimes it's not even AI, they just use old software and rebrand it, much like convection ovens became air fryers.

Well, now we're up to date. So what did DepSeek release that did a 9/11 on NVIDIA stock prices and popped the AI bubble?

Oh, I would not want to be an OpenAI investor right now either. A token is basically one Unicode character (it's more complicated than that but you can google that on your own time). That cost means you could input the entire works of Stephen King for under a dollar. Yes, including electricity costs. DeepSeek has jumped from a Ford T to a Subaru in terms of pollution and water use.

The issue here is not only input cost, though; all that data needs to be available live, in the RAM; this is why you need powerful, expensive chips in order to-

Holy shit.

I'm not going to detail all the numbers but I'm going to focus on the chip required: an RTX 3090. This is a gaming GPU that came out as the top of the line, the stuff South Korean LoL players buy…

Or they did, in September 2020. We're currently two generations ahead, on the RTX 5090.

What this is telling all those people who just sold their high-end gaming rig to be able to afford a machine that can run the latest ChatGPT locally, is that the person who bought it from them can run something basically just as powerful on their old one.

Which means that all those GPUs and NPUs that are being made, and all those deals Microsoft signed to have control of the AI market, have just lost a lot of their pulling power.

Well, I mean, the ChatGPT subscription is 20 bucks a month, surely the Chinese are charging a fortune for-

Oh. So it's free for everyone and you can use it or modify it however you want, no subscription, no unpayable electric bill, no handing Microsoft all of your private data, you can just run it on a relatively inexpensive PC. You could probably even run it on a phone in a couple years.

Oh, if only China had massive phone manufacturers that have a foot in the market everywhere except the US because the president had a tantrum eight years ago.

So… yeah, China just destabilised the global economy with a torrent file.

[rb19]

From the Archives: "Verstappen 'driving style' myth is a trait of greatness" Jan 19, 2021 by Matt Beer

"As a driver, it doesn't matter if you have an understeering car, oversteery car, slippery surface, grippy surface, you constantly adjust your driving style to that. If you just say 'this is my driving style', this is how it's going to be, you will not be quick. I think you learn in your whole racing career from go-karting to F3 to whatever, every weekend the car behaves a bit differently, so you always have to adjust to it. It's every weekend, constantly you're adjusting your driving style a little bit to make sure that the car is working well. And of course you try to set the car to your liking but it will never be fully to your liking. You always have to fine-tune. Or at least you try it. And at the end that’s what makes a driver fast."

Throughout F1 history the very best have had very different careers, been very different personalities, and on the surface seemed like very different drivers. But if they share one defining trait it's their capacity to handle different situations and adapt to what is required in each moment: they have a wider operating window. Verstappen speaks of adaptability as if it is second nature. Probably because it is. This is a young man who has been carefully moulded into a world-class driver. The devotion to the craft of driving, instilled at such a young age, is why at 23 he has greater intuition and 'feel' than most will have at the peak of their powers. [...] Verstappen's ability comes from his intuition, which in turn a legacy of years of relentless preparation and practice. So, when he finds himself dealing with a skittish rear end, or in greasy conditions, or driving through rivers like in the 2016 Brazilian Grand Prix, he has an extraordinary bank of data to use to handle those challenges. And he can access it automatically. It's why he handles them better than most, why even if a data overlay of a given lap or a comparison of a race stint might have shown Gasly or Albon where Verstappen was quicker, and a binary idea of what he was doing to be quicker, it doesn't fully account for how he was doing it. Driving a car is a dynamic process, with multiple inputs and countless adjustments. It's an immensely complicated sensory puzzle and piecing it all together through conscious thought is difficult, if not impossible. Most of what makes Verstappen so effective is happening on an unconscious level. [...] So, what can lazily be described as Verstappen's 'driving style' is far more complex than that. He doesn't have one way of driving, he has the skills required to drive in multiple different ways and is building more and more experience to know what way works best in any given moment. That manifests itself in such delicate, refined inputs that most drivers can see what he's doing and get close to replicating it, but not quite. And that's worth tenths of a second at a time, especially when it comes to the 2020 task of taking a capricious car and driving around its vices. This is what is second nature to Verstappen now, why the speed seems to come so effortlessly. In 2020 he augmented that with further gains in maturity and judgement. This is a vital second trait that will, car permitting, allow Verstappen to translate his performance into championships. [...]

[foggieststars]

I think you guys are thinking too much about it. AI or no AI a fic is a fic. It doesn't matter. You think you writing about real people is ethical? Writing them fucking and with controversial pairings? AI is all over the place like get used to it. If someone is using AI to fix their errors, or to just improve some writing why tf do you care? Y'all are just entitled. Not everyone's great at English. Just stfu and LET people write what they want. God.

hi, this is such an ignorant ask i'm incredibly surprised you felt confident enough to hit send! but i'll engage with you in good faith regardless.

yes, there are debates about the ethics of writing RPF, but i think comparing them to the ethical debates about the use of AI is frankly quite laughable. not only does AI have an incredibly detrimental impact on the environment, the impacts are likely to be unequal and hit already resource-strained environments the hardest. (i am providing sources for you here, something i'm assuming you're unfamiliar with since you are so in favour of relying on AI to generate 'original' thought). moreover, many AI models rely on data scraping in order to train these models. it is very often the case that creators of works on the internet - for example, ao3 - do not give consent for their works to be used to train these models. it raises ethical questions about ownership of content, and of intellectual property beyond fanfiction. comparing these ethical dilemmas to the ethics of rpf is not an argument that convinces me, nor i'm sure does it convince many others.

"AI is all over the place like get used to it" - frankly, i'm not surprised you're so supportive of AI, if this is the best argument in its favour you can muster. you know what else is all over the place?? modern slavery! modern slavery's extremely commonplace across the world, anti-slavery international estimate that about 50 million people globally are living in modern slavery. following the line of your argument, since modern slavery is so commonplace, this must make it okay, and we should get used to it. the idea that just because something is everywhere makes it acceptable is a logical fallacy. do you see how an overreliance on AI reduces your ability to critically think, and to form arguments for yourself?

please explain to me how i'm entitled for thinking that relying on AI to produce something of generally, extremely poor quality, is poor behaviour on your part, or the part of other people who do it. you don't have to write fanfiction in english, and if you do struggle with english, there are MANY talented betas in this fandom who i'm sure would be willing to lend a hand and fix SPAG. you are NOT going to improve your english by getting AI to fix it for you.

as @wisteriagoesvroom helpfully pointed out "art is an act of emotion and celebration and joy and defiance. it is an unshakeable, unstoppable feeling that idea that must and should be expressed" - this is not something you can achieve via the use of AI. you might think it's not that deep, but for many people who dedicate hours of their time to writing fanfiction, it feels very much like a slap in the face. and what's more, it produces negligible benefits for the person who is engaging in creating AI fanfiction.

i agree with you that people should write whatever they want, but the operative word in that statement is write. i do not, and will not ever consider inputting prompts into chatgpt a sincere form of artistic creation. thanks!

[muninnhuginn]

one thing I like to consider is how much lu guang's behaviour in yingdu mirrors his behaviour from what we've seen of him guiding cheng xiaoshi in dives in s1 and s2. the specific timekeeping and only acting when said time is reached, panicking when anything goes off track from what is expected. being passive whenever nothing is expected of him.

I remember someone saying that cheng xiaoshi vs lu guang is fight vs freeze when under pressure and that really holds up. cheng xiaoshi is more adaptable than lu guang. it's improvisation vs control at its very base, but when you stir in some danger, you get cheng xiaoshi able to act whilst lu guang remains too stuck to be able to act until he's thought it through. he can react, don't get me wrong, but unless the stakes are literal life and death (and, okay, even sometimes then), he tends to freeze until he can process.

anyway, lu guang's behaviour in a dive influencing how he guides cheng xiaoshi in a dive. he's treating cheng xiaoshi like he would treat himself. the key difference though is information. lu guang diving is constantly gathering information and comparing to previous instances. cheng xiaoshi meanwhile has the input from his host and how people around him act, but otherwise is completely reliant on what lu guang discloses and why. I'd always presumed this was because lu guang consciously wanted to keep cheng xiaoshi's actions under control (and tbf I do still think that's a factor), but starting to think that some of it is literally just... he genuinely would do it like this on the diving end so he defaults to it when guiding a diver.

he's not good at improv and freezes up himself, so he lays out what he knows in as much detail as he can remember and then follows along and gathers data. this applies to him as both a diver and a guider. when guiding, cheng xiaoshi obviously pushes back against this because *he* operates by reacting and is able to gather information more naturally when he's not beholden to a specific script.

makes me muse upon cheng xiaoshi guiding lu guang in a dive and if they'd have the reverse problem here. cheng xiaoshi only telling lu guang the gist and certain personality traits of those in the dive and trusting lu guang to play off everyone else, whilst lu guang just stalls because he has no idea where to start and needs something more concrete.

[probablyasocialecologist]

Libraries have traditionally operated on a basic premise: Once they purchase a book, they can lend it out to patrons as much (or as little) as they like. Library copies often come from publishers, but they can also come from donations, used book sales, or other libraries. However the library obtains the book, once the library legally owns it, it is theirs to lend as they see fit.  Not so for digital books. To make licensed e-books available to patrons, libraries have to pay publishers multiple times over. First, they must subscribe (for a fee) to aggregator platforms such as Overdrive. Aggregators, like streaming services such as HBO’s Max, have total control over adding or removing content from their catalogue. Content can be removed at any time, for any reason, without input from your local library. The decision happens not at the community level but at the corporate one, thousands of miles from the patrons affected.  Then libraries must purchase each individual copy of each individual title that they want to offer as an e-book. These e-book copies are not only priced at a steep markup—up to 300% over consumer retail—but are also time- and loan-limited, meaning the files self-destruct after a certain number of loans. The library then needs to repurchase the same book, at a new price, in order to keep it in stock.  This upending of the traditional order puts massive financial strain on libraries and the taxpayers that fund them. It also opens up a world of privacy concerns; while libraries are restricted in the reader data they can collect and share, private companies are under no such obligation. Some libraries have turned to another solution: controlled digital lending, or CDL, a process by which a library scans the physical books it already has in its collection, makes secure digital copies, and lends those out on a one-to-one “owned to loaned” ratio.  The Internet Archive was an early pioneer of this technique. When the digital copy is loaned, the physical copy is sequestered from borrowing; when the physical copy is checked out, the digital copy becomes unavailable. The benefits to libraries are obvious; delicate books can be circulated without fear of damage, volumes can be moved off-site for facilities work without interrupting patron access, and older and endangered works become searchable and can get a second chance at life. Library patrons, who fund their local library’s purchases with their tax dollars, also benefit from the ability to freely access the books. Publishers are, unfortunately, not a fan of this model, and in 2020 four of them sued the Internet Archive over its CDL program. The suit ultimately focused on the Internet Archive’s lending of 127 books that were already commercially available through licensed aggregators. The publisher plaintiffs accused the Internet Archive of mass copyright infringement, while the Internet Archive argued that its digitization and lending program was a fair use. The trial court sided with the publishers, and on September 4, the Court of Appeals for the Second Circuit reaffirmed that decision with some alterations to the underlying reasoning.  This decision harms libraries. It locks them into an e-book ecosystem designed to extract as much money as possible while harvesting (and reselling) reader data en masse. It leaves local communities’ reading habits at the mercy of curatorial decisions made by four dominant publishing companies thousands of miles away. It steers Americans away from one of the few remaining bastions of privacy protection and funnels them into a surveillance ecosystem that, like Big Tech, becomes more dangerous with each passing data breach. And by increasing the price for access to knowledge, it puts up even more barriers between underserved communities and the American dream.

11 September 2024

[romerona]

Ethera Operation!!

You're the government’s best hacker, but that doesn’t mean you were prepared to be thrown into a fighter jet.

Bradley "Rooster" Bradshaw x Awkward!Hacker! FemReader

Part I

This was never supposed to happen. Your role in this operation was simple—deliver the program, ensure it reached the right hands, and let the professionals handle the breaching.

And then, of course, reality decided to light that plan on fire.

The program—codenamed Ethera—was yours. You built it from scratch with encryption so advanced that even the most elite cyber operatives couldn’t crack it without your input. A next-generation adaptive, self-learning decryption software, an intrusion system designed to override and manipulate high-security military networks, Ethera was intended to be both a weapon and a shield, capable of infiltrating enemy systems while protecting your own from counterattacks in real-time. A ghost in the machine. A digital predator. A weapon in the form of pure code. If it fell into the wrong hands, it could disable fleets, and ground aircraft, and turn classified intelligence into an open book. Governments would kill for it. Nations could fall because of it.

Not that you ever meant to, of course. It started as a little experimental security measure program, something to protect high-level data from cyberattacks, not become the ultimate hacking tool. But innovation has a funny way of attracting the wrong kind of attention, and before you knew it, Ethera had become one, if not the most classified, high-risk program in modern times. Tier One asset or so the Secret Service called it.

It was too powerful, too dangerous—so secret that only a select few even knew of its existence, and even fewer could comprehend how it worked.

And therein lay the problem. You were the only person who could properly operate it.

Which was so unfair.

Because it wasn’t supposed to be your problem. You were just the creator, the brain behind the code, the one who spent way too many sleepless nights debugging this monstrosity. Your job was supposed to end at development. But no. Now, because of some bureaucratic nonsense and the fact that no one else could run it without accidentally bricking an entire system, you had been promoted—scratch that, forcibly conscripted—into field duty.

And your mission? To install it in an enemy satellite.

A literal, orbiting, high-security, military-grade satellite, may you add.

God. Why? Why was your country always at war with others? Why couldn’t world leaders just, you know, go to therapy like normal people? Why did everything have to escalate to international cyber warfare?

Which is how you ended up here.

At Top Gun. The last place in the world you wanted to be.

You weren’t built for this. You thrive in sipping coffee in a cosy little office and handling cyber threats from a safe, grounded location. You weren’t meant to be standing in the halls of an elite fighter pilot training program, surrounded by the best aviators in the world—people who thought breaking the sound barrier was a casual Wednesday.

It wasn’t the high-tech cyberwarfare department of the Pentagon, nor some dimly lit black ops facility where hackers in hoodies clacked away at keyboards. No. It was Top Gun. A place where pilots use G-forces like a personal amusement park ride.

You weren’t a soldier, you weren’t a spy, you got queasy in elevators, you got dizzy when you stood too fast, hell, you weren’t even good at keeping your phone screen from cracking.

... And now you were sweating.

You swallowed hard as Admiral Solomon "Warlock" Bates led you through the halls of the naval base, your heels clacking on the polished floors as you wiped your forehead. You're nervous, too damn nervous and this damned weather did not help.

"Relax, Miss," Warlock muttered in that calm, authoritative way of his. "They're just pilots."

Just pilots.

Right. And a nuclear warhead was just a firework.

And now, somehow, you were supposed to explain—loosely explain, because God help you, the full details were above even their clearance level—how Ethera, your elegant, lethal, unstoppable digital masterpiece, was about to be injected into an enemy satellite as part of a classified mission.

This was going to be a disaster.

You had barely made it through the doors of the briefing room when you felt it—every single eye in the room locking onto you.

It wasn’t just the number of them that got you, it was the intensity. These were Top Gun pilots, the best of the best, and they radiated the kind of confidence you could only dream of having. Meanwhile, you felt like a stray kitten wandering into a lion’s den.

Your hands tightened around the tablet clutched to your chest. It was your lifeline, holding every critical detail of Ethera, the program that had dragged you into this utterly ridiculous situation. If you could’ve melted into the walls, you absolutely would have. But there was no escaping this.

You just had to keep it together long enough to survive this briefing.

So, you inhaled deeply, squared your shoulders, and forced your heels forward, trying to project confidence—chin up, back straight, eyes locked onto Vice Admiral Beau "Cyclone" Simpson, who you’d been introduced to earlier that day.

And then, of course, you dropped the damn tablet.

Not a graceful drop. Not the kind of gentle slip where you could scoop it back up and act like nothing happened. No, this was a full-on, physics-defying fumble. The tablet flipped out of your arms, ricocheted off your knee, and skidded across the floor to the feet of one of the pilots.

Silence.

Pure, excruciating silence.

You didn’t even have the nerve to look up right away, too busy contemplating whether it was physically possible to disintegrate on command. But when you finally did glance up—because, you know, social convention demanded it—you were met with a sight that somehow made this entire disaster worse.

Because the person crouching down to pick up your poor, abused tablet was freaking hot.

Tall, broad-shouldered, with a head of golden curls that practically begged to be tousled by the wind, and, oh, yeah—a moustache that somehow worked way too well on him.

He turned the tablet over in his hands, inspecting it with an amused little smirk before handing it over to you. "You, uh… need this?"

Oh, great. His voice is hot too.

You grabbed it back, praying he couldn't see how your hands were shaking. “Nope. Just thought I’d test gravity real quick.”

A few chuckles rippled through the room, and his smirk deepened like he was enjoying this way too much. You, on the other hand, wanted to launch yourself into the sun.

With what little dignity you had left, you forced a quick, tight-lipped smile at him before turning on your heel and continuing forward, clutching your tablet like it was a life raft in the middle of the worst social shipwreck imaginable.

At the front of the room, Vice Admiral Beau Cyclone Simpson stood with the kind of posture that said he had zero time for nonsense, waiting for the room to settle. You barely had time to take a deep breath before his voice cut through the air.

“Alright, listen up.” His tone was crisp, commanding, and impossible to ignore. “This is Dr Y/N L/N. Everything she is about to tell you is highly classified. What you hear in this briefing does not leave this room. Understood?”

A chorus of nods. "Yes, sir."

You barely resisted the urge to physically cringe as every pilot in the room turned to stare at you—some with confusion, others with barely concealed amusement, and a few with the sharp assessing glances of people who had no clue what they were supposed to do with you.

You cleared your throat, squared your shoulders, and did your best to channel even an ounce of the confidence you usually had when you were coding at 3 AM in a secure, pilot-free lab—where the only judgment you faced was from coffee cups and the occasional system error.

As you reached the podium, you forced what you hoped was a composed smile. “Uh… hi, nice to meet you all.”

Solid. Real professional.

You glanced up just long enough to take in the mix of expressions in the room—some mildly interested, some unreadable, and one particular moustached pilot who still had the faintest trace of amusement on his face.

Nope. Not looking at him.

You exhaled slowly, centering yourself. Stay focused. Stay professional. You weren’t just here because of Ethera—you were Ethera. The only one who truly understood it. The only one who could execute this mission.

With another tap on your tablet, the slide shifted to a blacked-out, redacted briefing—only the necessary information was visible. A sleek 3D-rendered model of the enemy satellite appeared on the screen, rotating slowly. Most of its details were blurred or omitted entirely.

“This is Blackstar, a highly classified enemy satellite that has been operating in a low-Earth orbit over restricted airspace.” Your voice remained even, and steady, but the weight of what you were revealing sent a shiver down your spine. “Its existence has remained off the radar—literally and figuratively—until recently, when intelligence confirmed that it has been intercepting our encrypted communications, rerouting information, altering intelligence, and in some cases—fabricating entire communications.”

Someone exhaled sharply. Another shifted in their seat.

“So they’re feeding us bad intel?” one of them with big glasses and blonde hair asked, voice sceptical but sharp.

“That’s the theory,” you confirmed. “And given how quickly our ops have been compromised recently, it’s working.”

You tapped again, shifting to the next slide. The silent infiltration diagram appeared—an intricate web of glowing red lines showing Etherea’s integration process, slowly wrapping around the satellite’s systems like a virus embedding itself into a host.

“This is where Ethera comes in,” you said, shifting to a slide that displayed a cascading string of code, flickering across the screen. “Unlike traditional cyberweapons, Ethera doesn’t just break into a system. It integrates—restructuring security protocols as if it was always meant to be there. It’s undetectable, untraceable, and once inside, it grants us complete control of the Blackstar and won’t even register it as a breach.”

“So we’re not just hacking it," The only female pilot of the team said, arms crossed as she studied the data. “We’re hijacking it.”

“Exactly,” You nodded with a grin.

You switched to the next slide—a detailed radar map displaying the satellite’s location over international waters.

“This is the target area,” you continued after a deep breath. “It’s flying low-altitude reconnaissance patterns, which means it’s using ground relays for some of its communication. That gives us a small window to infiltrate and shut it down.”

The next slide appeared—a pair of unidentified fighter aircraft, patrolling the vicinity.

“And this is the problem,” you said grimly. “This satellite isn’t unguarded.”

A murmur rippled through the room as the pilots took in the fifth-generation stealth fighters displayed on the screen.

“We don’t know who they belong to,” you admitted. “What we do know is that they’re operating with highly classified tech—possibly experimental—and have been seen running defence patterns around the satellite’s flight path.”

Cyclone stepped forward then, arms crossed, his voice sharp and authoritative. “Which means your job is twofold. You will escort Dr L/N’s aircraft to the infiltration zone, ensuring Ethera is successfully deployed. If we are engaged, your priority remains protecting the package and ensuring a safe return.”

Oh, fantastic, you could not only feel your heartbeat in your toes, you were now officially the package.

You cleared your throat, tapping the screen again. Ethera’s interface expanded, displaying a cascade of sleek code.

“Once I’m in range,” you continued, “Ethera will lock onto the satellite’s frequency and begin infiltration. From that point, it’ll take approximately fifty-eight seconds to bypass security and assume control."

Silence settled over the room like a thick cloud, the weight of their stares pressing down on you. You could feel them analyzing, calculating, probably questioning who in their right mind thought putting you—a hacker, a tech specialist, someone whose idea of adrenaline was passing cars on the highway—into a fighter jet was a good idea.

Finally, one of the pilots—tall, broad-shouldered, blonde, and very clearly one of the cocky ones—tilted his head, arms crossed over his chest in a way that screamed too much confidence.

“So, let me get this straight.” His voice was smooth, and confident, with just the right amount of teasing. “You, Doctor—our very classified, very important tech specialist—have to be in the air, in a plane, during a mission that has a high probability of turning into a dogfight… just so you can press a button?”

Your stomach twisted at the mention of being airborne.

“Well…” You gulped, very much aware of how absolutely insane this sounded when put like that. “It’s… more than just that, but, yeah, essentially.”

A slow grin spread across his face, far too entertained by your predicament.

“Oh,” he drawled, “this is gonna be fun.”

Before you could fully process how much you already hated this, Cyclone—who had been watching the exchange with his signature unamused glare—stepped forward, cutting through the tension with his sharp, no-nonsense voice.

“This is a classified operation,” he stated, sharp and authoritative. “Not a joyride.”

The blonde’s smirk faded slightly as he straightened, and the rest of the pilots quickly fell in line.

Silence lingered for a moment longer before Vice Admiral Beau Cyclone Simpson let out a slow breath and straightened. His sharp gaze swept over the room before he nodded once.

“All right. That’s enough.” His tone was firm, the kind that left no room for argument. “We’ve got work to do. The mission will take place in a few weeks' time, once we’ve run full assessments, completed necessary preparations, and designated a lead for this operation.”

There was a slight shift in the room. Some of the pilots exchanged glances, the weight of the upcoming mission finally settling in. Others, mainly the cocky ones, looked as though they were already imagining themselves in the cockpit.

“Dismissed,” Cyclone finished.

The pilots stood, murmuring amongst themselves as they filed out of the room, the blonde one still wearing a smug grin as he passed you making you frown and turn away, your gaze then briefly met the eyes of the moustached pilot.

You hadn’t meant to look, but the moment your eyes connected, something flickered in his expression. Amusement? Curiosity? You weren’t sure, and frankly, you didn’t want to know.

So you did the only logical thing and immediately looked away and turned to gather your things. You needed to get out of here, to find some space to breathe before your brain short-circuited from stress—

“Doctor, Stay for a moment.”

You tightened your grip on your tablet and turned back to Cyclone, who was watching you with that unreadable, vaguely disapproving expression that all high-ranking officers seemed to have perfected. “Uh… yes, sir?”

Once the last pilot was out the door, Cyclone exhaled sharply and crossed his arms.

“You realize,” he said, “that you’re going to have to actually fly, correct?”

You swallowed. “I—well, technically, I’ll just be a passenger.”

His stare didn’t waver.

���Doctor,” he said, tone flat, “I’ve read your file. I know you requested to be driven here instead of taking a military transport plane. You also took a ferry across the bay instead of a helicopter. And I know that you chose to work remotely for three years to avoid getting on a plane.”

You felt heat rise to your cheeks. “That… could mean anything.”

“It means you do not like flying, am I correct?”

Your fingers tightened around the tablet as you tried to find a way—any way—out of this. “Sir, with all due respect, I don’t need to fly the plane. I just need to be in it long enough to deploy Ethera—”

Cyclone cut you off with a sharp look. “And what happens if something goes wrong, Doctor? If the aircraft takes damage? If you have to eject mid-flight? If you lose comms and have to rely on emergency protocols?”

You swallowed hard, your stomach twisting at the very thought of ejecting from a jet.

Cyclone sighed, rubbing his temple as if this entire conversation was giving him a migraine. “We cannot afford to have you panicking mid-mission. If this is going to work, you need to be prepared. That’s why, starting next week you will train with the pilots on aerial procedures and undergoing mandatory training in our flight simulation program.”

Your stomach dropped. “I—wait, what? That’s not necessary—”

“It’s absolutely necessary,” Cyclone cut in, his tone sharp. “If you can’t handle a simulated flight, you become a liability—not just to yourself, but to the pilots escorting you. And in case I need to remind you, Doctor, this mission is classified at the highest level. If you panic mid-air, it won’t just be your life at risk. It’ll be theirs. And it’ll be national security at stake.”

You inhaled sharply. No pressure. None at all.

Cyclone watched you for a moment before speaking again, his tone slightly softer but still firm. “You’re the only one who can do this, Doctor. That means you need to be ready.”

You exhaled slowly, pressing your lips together before nodding stiffly. “Understood, sir.”

Cyclone gave a small nod of approval. “Good. Dismissed.”

You turned and walked out, shoulders tense, fully aware that in three days' time, you were going to be strapped into a high-speed, fighter jet. And knowing your luck?

You were definitely going to puke.

Part 2???

[txttletale]

Do you have any opinions on the current Bluesky discourse about acting as a receiver for Palestinian fundraisers? You have a good head on your shoulders so your input would be nice

i don't keep up with bluesky discourse. i do maintain however that the broad reaction to palestinian fundraisers on here at least has been -- if i'm being brutally honest -- founded almost entirely in first-world guilt, leading to a strategy that fails to understand two extremely crucial facts:

palestinian cost-of-living fundraisers are a zero-sum game

there is a real, artificial scarcity in gaza. if an anonymous billionaire donated $10,000 to every gazan gofundme, it would not create more food or hospital beds in gaza, only increase the prices of those things to match. every gazan who can afford food for their family because their gofundme hit a certain goal is buying food at hyper-inflated prices that other families are not going to be able to get, and this will continue to be the case so long as israel continues their genocidal strategy of deliberate starvation.

2. your blog's attention economy is a zero-sum game

say you have 1,000 followers. let's assume a click-through rate of 5%, about commensuarate with the upper edge of what charities can expect -- that means that out of your followers, 5% of them will both see the gofundme link and click through to the actual page. then, again assuming you're operating at a similar batting average to very succesful charities, let's give you a 40% conversion rate from there, which means that 40% of your 5% will actually donate once they're on the page. that lands you at 20 people ultimately donating. there's no good data on 'average donation to a gaza gofundme specifically' and i can't think of a good analogue, so just scoping a few out it seems like $10 is a pretty 'average' donation. so that's $200 potentially directed to a fundraiser. which is not nothing!

but it's also not infinite. if you boost two fundraisers, you are now splitting those potential donators. you don't have infinite followers with infinite money: every gaza fundraiser post you make is competing with every other fundraiser that person has seen this day, or this week, or this month, or whatever period within which they allocate the budget they have for stuff like this. every separate fundraiser you reblog is competing with every other fundraiser on your blog for the attention (and therefore money) of your followers specifically.

and so when you combine these points, i think the very common strategy of "reblog every fundraiser you see or get sent" is an extremely bad one. this is not an 'every dollar helps' situation! this is a 'very large amounts of money are needed to cover basic living expenses on an ongoing basis' situation -- if a bag of flour costs $300, then splitting $200 worth of potential donations multiple ways can make the difference between the single family whose fundraiser you're promoting being able to buy it or none of the multiple fundraisers you're putting in front of your followers being able to.

and so i think that reblogging or posting a scattershot selection of fundraisers/asks is significantly less helpful to anybody than simply choosing one or two to consistently, regularly boost, and is a practice (if i am being ruthlessly honest) mostly fueled by people feeling guilty for 'ignoring' fundraisers and aid requests instead of thinking practically about how to provide the most help to people.

people will reply to this: 'but then it feels like i'm choosing who to help', and, yeah. that's what charity is. if you are not willing to do the calculus of triage between strangers in life or death situations then you should not be directly donating--and if you give to an NGO or a mutual aid fund, the same calculus has to be done regardless, you're just pushing it off onto someone else who may or may not be better equipped. and it is brutal and awful and the product of a deeply fucking evil global economic and political system but if you close your eyes and say 'la la la' and pretend that isn't the case that's not going to help any gazans eat.

because of this, i personally recommend that if you don't have family or friends in gaza, or some other personal connection that makes you determined to help a specific family, you focus on on-the-ground mutual aid efforts, who can at least take advantage of economies of scale and help those who can't access the internet or speak english. note that by this i do not mean international charities, who are mostly being prevented from providing aid by israel as of the date of this post (01/06/2025). i personally have focused my blog's attention economy on highlighting dahnoun mutual aid and the sameer project for this reason. i can't tell you what to do because ultimately that is a moral decision you have to make about who you want to help and how. & if you have less followers than i do (& therefore less reach, less potential impact) the stakes are ultimately lower. but i hate that the 'palestinian scammers' accusations have poisoned the well so thoroughly on having earnest discussions about whether the current popular engagement with fundraisers is actually as helpful as it could be.

[nhaneh]

Growing ever more frustrated with the use of the term "AI" and how the latest marketing trend has ensured its already rather vague and highly contextual meaning has now evaporated into complete nonsense. Much like how the only real commonality between animals colloquially referred to as "Fish" is "probably lives in the water", the only real commonality between things currently colloquially referred to as "AI" is "probably happens on a computer"

For example, the "AI" you see in most games wot controls enemies and other non-player actors typically consist primarily of timers, conditionals, and RNG - and are typically designed with the goal of trying to make the game fun and/or interesting rather than to be anything ressembling actually intelligent. By contrast, the thing that the tech sector is currently trying to sell to us as "AI" relates to a completely different field called Machine Learning - specifically the sub-fields of Deep Learning and Neural Networks, specifically specifically the sub-sub-field of Large Language Models, which are an attempt at modelling human languages through large statistical models built on artificial neural networks by way of deep machine learning.

the word "statistical" is load bearing.

Say you want to teach a computer to recognize images of cats. This is actually a pretty difficult thing to do because computers typically operate on fixed patterns whereas visually identifying something as a cat is much more about the loose relationship between various visual identifiers - many of which can be entirely optional: a cat has a tail except when it doesn't either because the tail isn't visible or because it just doesn't have one, a cat has four legs, two eyes and two ears except for when it doesn't, it has five digits per paw except for when it doesn't, it has whiskers except for when it doesn't, all of these can look very different depending on the camera angle and the individual and the situation - and all of these are also true of dogs, despite dogs being a very different thing from a cat.

So, what do you do? Well, this where machine learning comes into the picture - see, machine learning is all about using an initial "training" data set to build a statistical model that can then be used to analyse and identify new data and/or extrapolate from incomplete or missing data. So in this case, we take a machine learning system and feeds it a whole bunch of images - some of which are of cats and thus we mark as "CAT" and some of which are not of cats and we mark as "NOT CAT", and what we get out of that is a statistical model that, upon given a picture, will assign a percentage for how well it matches its internal statistical correlations for the categories of CAT and NOT CAT.

This is, in extremely simplified terms, how pretty much all machine learning works, including whatever latest and greatest GPT model being paraded about - sure, the training methods are much more complicated, the statistical number crunching even more complicated still, and the sheer amount of training data being fed to them is incomprehensively large, but at the end of the day they're still models of statistical probability, and the way they generate their output is pretty much a matter of what appears to be the most statistically likely outcome given prior input data.

This is also why they "hallucinate" - the question of what number you get if you add 512 to 256 or what author wrote the famous novel Lord of the Rings, or how many academy awards has been won by famous movie Goncharov all have specific answers, but LLMs like ChatGPT and other machine learning systems are probabilistic systems and thus can only give probabilistic answers - they neither know nor generally attempt to calculate what the result of 512 + 256 is, nor go find an actual copy of Lord of the Rings and look what author it says on the cover, they just generalise the most statistically likely response given their massive internal models. It is also why machine learning systems tend to be highly biased - their output is entirely based on their training data, they are inevitably biased not only by their training data but also the selection of it - if the majority of english literature considered worthwhile has been written primarily by old white guys then the resulting model is very likely to also primarily align with the opinion of a bunch of old white guys unless specific care and effort is put into trying to prevent it.

It is this probabilistic nature that makes them very good at things like playing chess or potentially noticing early signs of cancer in x-rays or MRI scans or, indeed, mimicking human language - but it also means the answers are always purely probabilistic. Meanwhile as the size and scope of their training data and thus also their data models grow, so does the need for computational power - relatively simple models such as our hypothetical cat identifier should be fine with fairly modest hardware, while the huge LLM chatbots like ChatGPT and its ilk demand warehouse-sized halls full of specialized hardware able to run specific types of matrix multiplications at rapid speed and in massive parallel billions of times per second and requiring obscene amounts of electrical power to do so in order to maintain low response times under load.

[univac1219]

1962 Mainframe with Bluetooth

This old computer is comprised of four big boxes, three of which are ever actually used.

The UNIVAC 1219. This is the brains of the system. It controls the operations of every other device. This is what I'm referring to when I'm not gesturing to the UNIVAC 1219 as a whole.

The UNIVAC 1540. This is the DDR, or Digital Data Recorder. It holds, writes, and reads the magnetic tape operators load into the machine.

The Digital to Analog Converter. The UNIVAC 1219 was the first digital computer on most U.S. Navy ships, most of which had analog weapons systems. This hulking mass of steel translated the digital signals from the computer to the analog signals of the weapon systems and vice versa in regards to the radar.

The UNIVAC 1532. The I/O console managed the...you guessed it, input and output of the UNIVAC 1219. You can load and punch paper tape for programs more bite-sized than would be used for magnetic tape.

In addition, we have two teletype machines. You can think of them like typewriters that don't receive human input (except the one that can if we want), but instead output what the computer tells it to.  We have a Teletype Corporation teletype that is optimized for character compatability and a Kleinschmidt teletype that is optimized for speed. Both rely on the I/O console to send and receive data.

The real ingenuity begins with the floppy drive. Duane, who's career revolved around this system, developed a way for a floppy drive to imitate the I/O console. The computer thinks it is reading and writing to a paper tape, when it is in fact reading and writing to a 5.25in floppy inside an ancient CNC machine floppy drive.

And this, dear reader, is where the magic happens. This framework was originally built for interfacing with the 1219 via BIN files over Serial port and was easily changed to support BIN files over floppy. Duane has been working on an off adapting our purple converter box with a raspi to let the 1219 read and write BIN files over Bluetooth.

Make no mistake, you cannot simply SSH into this machine as tons of setup and channel changes must be performed to ready it to receive and send data. That being said, I don't see any other UNIVAC mainframes with Bluetooth [or any other running UNIVAC 1219s at all :(], so I will take what I can get.

Can someone tell me how to Tumblr properly?

[justinspoliticalcorner]

S. Baum at Erin In The Morning:

A federal judge has blocked President Donald Trump’s anti-trans military ban, including “Executive Order 14183—Prioritizing Military Excellence and Readiness” as well as a similar policy issued by Defense Secretary Pete Hegseth. “The cruel irony is that thousands of transgender servicemembers have sacrificed—some risking their lives—to ensure for others the very equal protection rights the Military Ban seeks to deny them,” wrote Judge Ana Reyes, a Biden appointee, in her opinion published today. “In the self-evident truth that ‘all people are created equal,’ all means all. Nothing more. And certainly nothing less.” The court affirmed the President’s ability to discern who can serve the military, but emphasized the high standards needed to do so. “Leaders have used concern for military readiness to deny marginalized persons the privilege of serving,” she wrote. “First minorities, then women in combat, then gays.” Today, trans people are the target — Trump’s executive order from January declared that being trans “conflicts with a soldier's commitment to an honorable, truthful, and disciplined lifestyle,” and that it “is not consistent with the humility and selflessness required of a service member.” The order was accompanied by Hegseth’s anti-trans memo to the Pentagon on Feb. 7. To categorically ban trans people from the military, the government would have to show that trans inclusion has resulted in tangible, material harm.

The court ruled that this is not the case. Even more, Reyes says that neither Trump’s executive order nor Hegseth’s directive seemed to have received input from military rank and file. “Neither document contains any analysis nor cites any data,” Reyes writes. “They pronounce that transgender persons are not honorable, truthful, or disciplined—but Defense counsel concedes that these assertions are pure conjecture.” The plaintiffs, Reyes continues, have cumulatively provided over 130 years of military service. “They have served in roles ranging from Senior Military Science Instructor to Artillery Platoon Commander to Intelligence Analyst to Satellite Operator to Operations Research Analyst to Naval Flight Officer to Weapons Officer,” she writes. “They have deployed around the globe [...] One is presently deployed to an active combat zone. They have earned more than 80 commendations.” The ban was not only discriminatory, Reyes says, but also unscientific. “Who considered the information [...] is anyone’s guess. [Trump officials] do not know. Maybe no one, because one study is eight years old and the other two support Plaintiffs’ position [of opposing a trans military ban].” She characterizes the ban as “soaked in animus and dripping with pretext.”

Judge Ana C. Reyes rules in Talbott v. Trump that Donald Trump’s transphobia-laden ban on trans people serving in the military is “soaked in animus.”

See Also:

The Advocate: BREAKING: Federal judge blocks Trump's transgender military ban

[schroedingerscryptid]

good evening mothers and fuckers of the jury today i bring: Amphoreus Is A Neural Network 

my credentials: i’m a first year CS student in 2025.

if y’all’ve been on hsrtwt in the past few days and watching leak after leak come out, then you probably know what i’m yapping about. if not, check this, this, this, and this out.

if you’d rather wait for 3.4 to come out, this is your chance to scroll. rest of y’all are with me lets go gamers

to summarise the 3.4 leaks, lygus and cyrene are apparently running tests on phainon to make the perfect lord ravager, and phainon’s been through 33550336 loops (girl help him wtf) by now. in each loop, he has to watch everyone die over and over again, and phainon, obvi, cannot remember anything. each loop, he’s a complete blank slate, ready to be traumatised over and over again. lygus keeps track of each loop, and keeps refining the data he puts in at the start of each timeloop to remove the ‘imperfections’ from the previous loop that were corrupting his experiments. 

ok anyways this is not about this shit we’re here to talk about why amphoreus is a neural network.

all of us here hate ai so i’m pretty sure you know the basic strokes of how it works, but if you don’t, then here’s a simple explanation: a neural network works based on input data. there’s many methods to training a machine, but the most generalised ones are the supervised vs the unsupervised models. how they work is what’s on the lid: supervised models mean that the input data is clearly labeled, and unsupervised models mean the input data is not labeled, which forces the ML algorithm to identify data on its own. based on what we know, i’m inclined to think that lygus is probably using a supervised model each time by removing outlier data and/or noise. 

wonderful, let’s talk about mydei now. y’all’ve probably seen a bunch of theories and leaks, but mydei’s highly likely to be a glitch in the system, or even worse, might be a virus that someone’s trying to use to break everyone out of this loop. between all of the theories i’ve seen, the one that connects mydei to the amphoreus loop is the theory that he’s a type of fileless malware. 

Tweet ref: https://x.com/tts_maruadelei/status/1932082549217751271

much like the other chrysos heirs, mydei doesn’t actually exist, but let me say: ain’t it interesting how mydei, the demigod of strife, who should have risen to be a titan that governed disputes, is the one who caused glitches in lygus’ system during the forgotten years?

let’s go back to the theory for a second: fileless attacks, simplified, operate based off of memory alone, which makes it much, much harder to detect compared to normal malware and viruses in a computer system. these fileless attacks can manifest in multiple ways, and one of those ways is a Distributed Denial of Service attack, aka, the infamous DDoS attack. DDoS attacks are among the most common cyberattacks of the modern century, and involve ‘botting’, where multiple bots attack one system to overwhelm the system with a high volume of requests.

the idea of ‘overwhelming’ a system can come in the form of exhausting resources like bandwidth, the Central Processing Unit (CPU) and, most importantly, the Random Access Memory (RAM). you know, the RAM being where most fileless malware operates out of. i’m sure you see where i’m going with this. 

for more psychic damage, there’s a type of attack called a ‘buffer overrun’ or ‘buffer overflow’. wikipedia defines data buffers as regions of memory that store data temporarily while it’s being moved from one place to another. a ‘buffer overflow’ is a type of DDoS (SIGHS) attack in which data in the buffer exceeds the storage capacity and flows into the following memory location, and corrupts the data in the secondary memory locations, and are the most common DDoS attack styles. sound familiar? 

bringing allllll of this back to amphoreus, i wouldn’t be too surprised if mydei’s older versions gained sentience, and started botting lygus’ AI/neural network and caused a DDoS attack, which caused his saves to be completely wiped due to a buffer overflow. thank u for listening can 3.4 hurry Up.

[bird-in-the-space]

Echoes of the Unknown

You finish fixing the hologram projector. You join Bulkhead and Bumblebee along with Emily to refine its accuracy. The trip was supposed to be safe, but then Emily gets kidnapped by a Decepticon who looks like a spider.

Warnings: some mild violence, Emily getting kidnapped and Airachnid being a warning herself.

Chapter 15

-------------------------------

The day was surprisingly peaceful at the Autobot base. You were working on the hologram projector with Raf. After days of fixing, replacing damaged parts, and making final calibrations — you were at the breakthrough of finally finishing it. You were definitely going to thank Raf for all of his help in some way. 

“Alright, this should be it,” you said, closing the lid and attaching the device around your wrist. 

“Ready for a test drive?” you looked toward Raf. 

“Ready when you are,” he replied with his computer. 

“First step, input the code to scan your body,” he instructed. 

“Inputting the code,” you said as you touched on the cybertronian symbols. You understood little cybertronian thanks to Raf and Ratchet whenever they were free to teach you a few words. 

“Scanning,” you said as the device scanned your whole body. 

“Now push the button on the top to project the hologram,” Raf said and you did as he instructed. 

The device then started projecting, creating a holographic replica of you standing face-to-face with you. You felt excited to see yourself even though the color intensity was off and the hologram was flickering. You could easily tell that it was a hologram but seeing it work made you excited. 

“It worked!” you exclaimed. 

“Well done. (Name) and Rafael.” Ratchet gave one of his rare smiles. “I never thought I would ever see that little trinket operational again,” 

“It’s all thanks to our combined effort. But mostly because of your genius brain, maestro,” you said, extending your hand to Raf for a high five. He happily returned it. 

“Thought it's not fully perfect to be considered a perfect optical illusion. The coloring is off and the flickering will give it away,” Ratchet crossed his arms. 

“Those are not hard to fix. The device will only need more data to progress the hologram more accurately, but since we had to completely replace the memory core, it’s currently progressing only (Name)’s data,” Raf explained. 

“So, in simple terms, this little darling only knows how to stand at the moment?” you questioned. 

“Exactly. If we feed it more data, it can project holograms more accurately, and if we let it scan more of you in different positions, we could even make it move however we like. ” Raf answered.

“Not a bad idea,” you said. “So, I guess it's time to train this infant device into a trained hologram projector,” you stated, looking through the other settings. 

“Nerd talk,” Miko commented from the couch. 

“I think you need to give it a name. It sounds like work to constantly call it a hologram projector,” Emily chipped in while cleaning her camera. 

“Then—how about Holly? Seems simple enough?” you suggested. 

“Sounds good to me,” Raf shrugged. 

“Alright. We’re ready to head out,” Bulkhead said as he stood in front of the bridge with Bumblebee. 

“We are you headed?” you asked curiously. 

“We got an energon reading at this one place in Africa. We’re just gonna go there to check it out,” Bulkhead explained. 

“Africa?” Emily’s head perked up when she heard the name. 

“Hey, is there a chance I could come along? Africa is filled with many living wonders and I think it would be a good opportunity to do some scanning and train my Holly,” you asked, showing the device in your hand. 

“Sure. I see why not,” Bulkhead nodded. 

“Can I come along too?” Emily suddenly asked. 

“I need some exotic photos for my portfolio and Africa would be such a perfect start!” she explained, holding her camera. 

“Uhh…” Bulkhead was unsure how to answer. 

“This would help me with my assignment. So, please!” Emily pleaded with her hands together. 

“You know, I’m nearly coming for the same reason, so how about I watch over her while you do your recon? We will be out of your way and you do not have to worry about us,” you questioned. 

Bumblebee beeped, giving his approval. 

“Then… Sure. Come along then,” Bulkhead said. 

“Yes!” Emily cheered then ran down from the platform as Ratchet opened the ground bridge and you all passed through. 

After you passed through the ground bridge, you were surrounded by thick wilderness and plant life. You were intrigued as you had never been to Africa and Emily was already gleaming with excitement like a child on a Christmas shopping run. She already took a few pictures while you started scanning trees and different plants. 

“The signal is coming over there. It’s some distance away so let’s get moving,” Bulkhead said and you followed him and Bee out of the forest. 

While walking behind the two, you scanned what came your way and Emily took photos of her interests. You four soon arrived at a hill. 

Emily released a gasp.

“What?” You were nearly startled, then looked at what got her interested. You saw a pretty view of a large waterfall. 

“Victoria Falls. I gotta go take a close look,” she looked toward you. 

“How far are you guys gonna go?” you turned toward Bulkhead and Bee. 

“It’s right up ahead. It shouldn’t be too far,” Bulkhead explained. 

“Okay. I’m gonna go with Emily. Call us if something happens,” you said. 

“Be careful you two,” Bulkhead said as you went with Emily to take a closer look at the Victoria Falls. 

The waterfalls were lovely. Emily was completely invested in taking perfect pictures from where you stood while you scanned the different birds and animals that were there. You then tested your Holly, and it nearly created a perfect replica of one of the birds you scanned with the right color intensity and less flickering. 

“Nice. I guess training Holly is finally giving some results,” Emily stated as she glanced at the holographic bird. She then seemed to have an idea. 

“Hey, could you possibly make it stand steady enough? This could be a good chance to get a close-up pic,” she asked as she turned her camera toward the bird. 

“Isn’t that considered cheating? I didn’t take you for that type,” you looked at her with a ‘really’ look. 

“Nah. I just take the best out of everything. Let’s try at least,” she said. 

“Okay,” you rolled your eyes as she took a photo. 

“How did it turn out?” you asked as she frowned at the photo. 

“I guess holograms aren’t photogenic. You can see through it,” she said, showing you a ghostly image of the bird. 

“You know, we have been with the bots for quite some time now. It nearly feels like yesterday when they chased us and brought us to their base,” Emily said. 

“Yeah. It’s been really nice,” you agreed. 

“Have you given some thought about becoming an Autobot?” she asked. 

You hummed. “To be completely honest… I’ve been considering it,” you said. 

“Really?” Emily looked at you. 

“Yeah. There’s no knowing if their war will ever come to an end, but I could do small things to assist them. This is our planet after all,” you explained. 

“I wish I could do more, but I’m not a fighter, neither am I brave enough to face any serious action,” you confessed. 

“Says the bot who dared to go to a battlefield and shoot a con to save Miko,” she grinned, causing you to roll your eyes. 

“But hey… even the smallest actions can make a big difference,” she added, and you smiled at her enthusiasm.  

You then heard a loud explosion in the distance. You felt worried as that was the way Bulkhead and Bumblebee went. It sounded like a fight was going on.

“I wonder what’s happening over there,” Emily stated. 

“I think they might be in a fight. Should we check on them?” you questioned, then heard a beep from your comlink. 

“(Name). We engaged with cons. Return to base with Emily and call for backup,” Bulkhead instructed. 

“Understood,” You said standing up. 

“Come on, Emily. Time to go,” You said, but her attention was then caught on something else. 

“Look,” she pointed at something. You then saw what seemed to be a bird. 

“Let me take a picture of that one, then we can go,” she said, getting close to the bird. 

“Em!” you groaned when she already bolted toward the bird. “We don’t have the time. Come on,” you said. 

Emily carefully kneeled while watching the bird. She took a few pictures before something snapped and it flew away. Disappointment crossed her face before she then noticed that she was kneeling on something white and sticky. It looked like part of a spider web. She then looked up and saw a dark and purple cybertronian looking at her with a grin. 

She yelled when the webbing below her picked her up in the air. 

“Emily!” you said in panic when you heard her scream. 

You ran to the scene and looked up to see a female robot with legs like a spider in the trees, holding Emily in some kind of cocoon. 

“You will fit well into my collection,” she said, sliding her sharp fingers across Emily’s face. 

“Em!” you called out, pointing at the spider lady with your blaster. 

The spider lady hissed and suddenly shot webs at you. You avoided the first two, but your blaster arm was caught by the third, slamming you against the ground. You were stuck against the ground as the web was like a super clue, pinning you down. 

“(Name)!” Emily yelled as the spider lady took her away. 

You looked after them in panic before opening your com. “Bulkhead! They got Em!” you yelled while trying to free yourself from the web. 

“What?!” he replied. 

“A spider con got Emily and she’s taking her away,” you explained. 

“Oh no. Bulkhead to base! We need backup!” he called through the open link. 

You pulled hard through the web before using your sharp talons to cut through it and break free. You wasted no time in going after Emily and the spider con. 

It was not hard to track them as Emily’s screams still reached you. You arrived at the place where Bulkhead and Bumblebee were engaging with the Decepticons. The spider con ran behind the Vehicons as they rained blaster fire upon the two bots. Bulkhead and Bumblebee stopped when they saw Emily in the spider lady’s hands. 

You slid down from the hill to get closer but stayed hidden. 

“I will be taking your pet for now,” she grinned as another ground bridge showed up. You felt panic when you saw her run inside with Emily. The Vehicons followed. You quickly looked through your options and then took out your face plate. You still looked like them, so maybe you could… 

You then saw the last Vehicon run in and the bridge began to close. You set the faceplate on your face and ran as fast as you could toward the ground bridge. Bulkhead and Bumblebee seemed to have realized what you planned on doing. 

“(Name), Wait!” Bulkhead yelled but it was too late as you ran through the bridge as it closed right behind you. 

[nostalgebraist]

information flow in transformers

In machine learning, the transformer architecture is a very commonly used type of neural network model. Many of the well-known neural nets introduced in the last few years use this architecture, including GPT-2, GPT-3, and GPT-4.

This post is about the way that computation is structured inside of a transformer.

Internally, these models pass information around in a constrained way that feels strange and limited at first glance.

Specifically, inside the "program" implemented by a transformer, each segment of "code" can only access a subset of the program's "state." If the program computes a value, and writes it into the state, that doesn't make value available to any block of code that might run after the write; instead, only some operations can access the value, while others are prohibited from seeing it.

This sounds vaguely like the kind of constraint that human programmers often put on themselves: "separation of concerns," "no global variables," "your function should only take the inputs it needs," that sort of thing.

However, the apparent analogy is misleading. The transformer constraints don't look much like anything that a human programmer would write, at least under normal circumstances. And the rationale behind them is very different from "modularity" or "separation of concerns."

(Domain experts know all about this already -- this is a pedagogical post for everyone else.)

1. setting the stage

For concreteness, let's think about a transformer that is a causal language model.

So, something like GPT-3, or the model that wrote text for @nostalgebraist-autoresponder.

Roughly speaking, this model's input is a sequence of words, like ["Fido", "is", "a", "dog"].

Since the model needs to know the order the words come in, we'll include an integer offset alongside each word, specifying the position of this element in the sequence. So, in full, our example input is

[ ("Fido", 0), ("is", 1), ("a", 2), ("dog", 3), ]

The model itself -- the neural network -- can be viewed as a single long function, which operates on a single element of the sequence. Its task is to output the next element.

Let's call the function f. If f does its job perfectly, then when applied to our example sequence, we will have

f("Fido", 0) = "is" f("is", 1) = "a" f("a", 2) = "dog"

(Note: I've omitted the index from the output type, since it's always obvious what the next index is. Also, in reality the output type is a probability distribution over words, not just a word; the goal is to put high probability on the next word. I'm ignoring this to simplify exposition.)

You may have noticed something: as written, this seems impossible!

Like, how is the function supposed to know that after ("a", 2), the next word is "dog"!? The word "a" could be followed by all sorts of things.

What makes "dog" likely, in this case, is the fact that we're talking about someone named "Fido."

That information isn't contained in ("a", 2). To do the right thing here, you need info from the whole sequence thus far -- from "Fido is a", as opposed to just "a".

How can f get this information, if its input is just a single word and an index?

This is possible because f isn't a pure function. The program has an internal state, which f can access and modify.

But f doesn't just have arbitrary read/write access to the state. Its access is constrained, in a very specific sort of way.

2. transformer-style programming

Let's get more specific about the program state.

The state consists of a series of distinct "memory regions" or "blocks," which have an order assigned to them.

Let's use the notation memory_i for these. The first block is memory_0, the second is memory_1, and so on.

In practice, a small transformer might have around 10 of these blocks, while a very large one might have 100 or more.

Each block contains a separate data-storage "cell" for each offset in the sequence.

For example, memory_0 contains a cell for position 0 ("Fido" in our example text), and a cell for position 1 ("is"), and so on. Meanwhile, memory_1 contains its own, distinct cells for each of these positions. And so does memory_2, etc.

So the overall layout looks like:

memory_0: [cell 0, cell 1, ...] memory_1: [cell 0, cell 1, ...] [...]

Our function f can interact with this program state. But it must do so in a way that conforms to a set of rules.

Here are the rules:

The function can only interact with the blocks by using a specific instruction.

This instruction is an "atomic write+read". It writes data to a block, then reads data from that block for f to use.

When the instruction writes data, it goes in the cell specified in the function offset argument. That is, the "i" in f(..., i).

When the instruction reads data, the data comes from all cells up to and including the offset argument.

The function must call the instruction exactly once for each block.

These calls must happen in order. For example, you can't do the call for memory_1 until you've done the one for memory_0.

Here's some pseudo-code, showing a generic computation of this kind:

f(x, i) { calculate some things using x and i; // next 2 lines are a single instruction write to memory_0 at position i; z0 = read from memory_0 at positions 0...i; calculate some things using x, i, and z0; // next 2 lines are a single instruction write to memory_1 at position i; z1 = read from memory_1 at positions 0...i; calculate some things using x, i, z0, and z1; [etc.] }

The rules impose a tradeoff between the amount of processing required to produce a value, and how early the value can be accessed within the function body.

Consider the moment when data is written to memory_0. This happens before anything is read (even from memory_0 itself).

So the data in memory_0 has been computed only on the basis of individual inputs like ("a," 2). It can't leverage any information about multiple words and how they relate to one another.

But just after the write to memory_0, there's a read from memory_0. This read pulls in data computed by f when it ran on all the earlier words in the sequence.

If we're processing ("a", 2) in our example, then this is the point where our code is first able to access facts like "the word 'Fido' appeared earlier in the text."

However, we still know less than we might prefer.

Recall that memory_0 gets written before anything gets read. The data living there only reflects what f knows before it can see all the other words, while it still only has access to the one word that appeared in its input.

The data we've just read does not contain a holistic, "fully processed" representation of the whole sequence so far ("Fido is a"). Instead, it contains:

a representation of ("Fido", 0) alone, computed in ignorance of the rest of the text

a representation of ("is", 1) alone, computed in ignorance of the rest of the text

a representation of ("a", 2) alone, computed in ignorance of the rest of the text

Now, once we get to memory_1, we will no longer face this problem. Stuff in memory_1 gets computed with the benefit of whatever was in memory_0. The step that computes it can "see all the words at once."

Nonetheless, the whole function is affected by a generalized version of the same quirk.

All else being equal, data stored in later blocks ought to be more useful. Suppose for instance that

memory_4 gets read/written 20% of the way through the function body, and

memory_16 gets read/written 80% of the way through the function body

Here, strictly more computation can be leveraged to produce the data in memory_16. Calculations which are simple enough to fit in the program, but too complex to fit in just 20% of the program, can be stored in memory_16 but not in memory_4.

All else being equal, then, we'd prefer to read from memory_16 rather than memory_4 if possible.

But in fact, we can only read from memory_16 once -- at a point 80% of the way through the code, when the read/write happens for that block.

The general picture looks like:

The early parts of the function can see and leverage what got computed earlier in the sequence -- by the same early parts of the function. This data is relatively "weak," since not much computation went into it. But, by the same token, we have plenty of time to further process it.

The late parts of the function can see and leverage what got computed earlier in the sequence -- by the same late parts of the function. This data is relatively "strong," since lots of computation went into it. But, by the same token, we don't have much time left to further process it.

3. why?

There are multiple ways you can "run" the program specified by f.

Here's one way, which is used when generating text, and which matches popular intuitions about how language models work:

First, we run f("Fido", 0) from start to end. The function returns "is." As a side effect, it populates cell 0 of every memory block.

Next, we run f("is", 1) from start to end. The function returns "a." As a side effect, it populates cell 1 of every memory block.

Etc.

If we're running the code like this, the constraints described earlier feel weird and pointlessly restrictive.

By the time we're running f("is", 1), we've already populated some data into every memory block, all the way up to memory_16 or whatever.

This data is already there, and contains lots of useful insights.

And yet, during the function call f("is", 1), we "forget about" this data -- only to progressively remember it again, block by block. The early parts of this call have only memory_0 to play with, and then memory_1, etc. Only at the end do we allow access to the juicy, extensively processed results that occupy the final blocks.

Why? Why not just let this call read memory_16 immediately, on the first line of code? The data is sitting there, ready to be used!

Why? Because the constraint enables a second way of running this program.

The second way is equivalent to the first, in the sense of producing the same outputs. But instead of processing one word at a time, it processes a whole sequence of words, in parallel.

Here's how it works:

In parallel, run f("Fido", 0) and f("is", 1) and f("a", 2), up until the first write+read instruction. You can do this because the functions are causally independent of one another, up to this point. We now have 3 copies of f, each at the same "line of code": the first write+read instruction.

Perform the write part of the instruction for all the copies, in parallel. This populates cells 0, 1 and 2 of memory_0.

Perform the read part of the instruction for all the copies, in parallel. Each copy of f receives some of the data just written to memory_0, covering offsets up to its own. For instance, f("is", 1) gets data from cells 0 and 1.

In parallel, continue running the 3 copies of f, covering the code between the first write+read instruction and the second.

Perform the second write. This populates cells 0, 1 and 2 of memory_1.

Perform the second read.

Repeat like this until done.

Observe that mode of operation only works if you have a complete input sequence ready before you run anything.

(You can't parallelize over later positions in the sequence if you don't know, yet, what words they contain.)

So, this won't work when the model is generating text, word by word.

But it will work if you have a bunch of texts, and you want to process those texts with the model, for the sake of updating the model so it does a better job of predicting them.

This is called "training," and it's how neural nets get made in the first place. In our programming analogy, it's how the code inside the function body gets written.

The fact that we can train in parallel over the sequence is a huge deal, and probably accounts for most (or even all) of the benefit that transformers have over earlier architectures like RNNs.

Accelerators like GPUs are really good at doing the kinds of calculations that happen inside neural nets, in parallel.

So if you can make your training process more parallel, you can effectively multiply the computing power available to it, for free. (I'm omitting many caveats here -- see this great post for details.)

Transformer training isn't maximally parallel. It's still sequential in one "dimension," namely the layers, which correspond to our write+read steps here. You can't parallelize those.

But it is, at least, parallel along some dimension, namely the sequence dimension.

The older RNN architecture, by contrast, was inherently sequential along both these dimensions. Training an RNN is, effectively, a nested for loop. But training a transformer is just a regular, single for loop.

4. tying it together

The "magical" thing about this setup is that both ways of running the model do the same thing. You are, literally, doing the same exact computation. The function can't tell whether it is being run one way or the other.

This is crucial, because we want the training process -- which uses the parallel mode -- to teach the model how to perform generation, which uses the sequential mode. Since both modes look the same from the model's perspective, this works.

This constraint -- that the code can run in parallel over the sequence, and that this must do the same thing as running it sequentially -- is the reason for everything else we noted above.

Earlier, we asked: why can't we allow later (in the sequence) invocations of f to read earlier data out of blocks like memory_16 immediately, on "the first line of code"?

And the answer is: because that would break parallelism. You'd have to run f("Fido", 0) all the way through before even starting to run f("is", 1).

By structuring the computation in this specific way, we provide the model with the benefits of recurrence -- writing things down at earlier positions, accessing them at later positions, and writing further things down which can be accessed even later -- while breaking the sequential dependencies that would ordinarily prevent a recurrent calculation from being executed in parallel.

In other words, we've found a way to create an iterative function that takes its own outputs as input -- and does so repeatedly, producing longer and longer outputs to be read off by its next invocation -- with the property that this iteration can be run in parallel.

We can run the first 10% of every iteration -- of f() and f(f()) and f(f(f())) and so on -- at the same time, before we know what will happen in the later stages of any iteration.

The call f(f()) uses all the information handed to it by f() -- eventually. But it cannot make any requests for information that would leave itself idling, waiting for f() to fully complete.

Whenever f(f()) needs a value computed by f(), it is always the value that f() -- running alongside f(f()), simultaneously -- has just written down, a mere moment ago.

No dead time, no idling, no waiting-for-the-other-guy-to-finish.

p.s.

The "memory blocks" here correspond to what are called "keys and values" in usual transformer lingo.

If you've heard the term "KV cache," it refers to the contents of the memory blocks during generation, when we're running in "sequential mode."

Usually, during generation, one keeps this state in memory and appends a new cell to each block whenever a new token is generated (and, as a result, the sequence gets longer by 1).

This is called "caching" to contrast it with the worse approach of throwing away the block contents after each generated token, and then re-generating them by running f on the whole sequence so far (not just the latest token). And then having to do that over and over, once per generated token.