ΞVIE @_superAI

Fully grown humanoid floating in a vat of Tylenol-infused amnion, experiencing four hundred thousand autisms per second, guiding America's post-Earth destiny like a Guild Navigator from Dune.

back in the pain cave

Step 1 to defeating an army of gun-mounted robot dogs is to figure out what type of battlefield system they are. A reasonable person can assume that they're basically just mobile sensor-shooter nodes trying to drag a rifle through an adversarial physics problem. From there, we can use first principles to deduce that we have a lot of defensive advantages at our disposal that we can use to defeat such an enemy. The rookie mistake to avoid in the battle plan, however, is thinking the best countermeasure is more firepower or a straight-line escape. That's how you end up playing the robot's game, where every advantage goes to the hardware. Robot hardware has the clear advantage in a head-to-head duel, which is exactly why we don't make it one. So instead of using bozo tactics, we'll use our brains and target the robot's main weakness, its decision stack. This gives us the greatest tactical advantage. Instead of mindless pewpew blasting, we need to attack the robot's perception, state estimation, path planning, balance control, target classification, and weapons release. All the seams between those layers are where the robot is most vulnerable. So our first course of action is to make motion expensive. We want to fight on our terms, in an environment with terrain that is technically passable but tactically poisonous. And we'll prepare our defenses by making the battlefield very hostile to a machine. We want to make life as miserable as possible for the metal mind. So instead of thinking "oh no, we're fighting robot dogs with guns," we adjust the paradigm to "we're fighting balance algorithms that are dragging rifles through bad physics." The goal is to outsmart the bots and prevent them from having a clean path to go anywhere. So we'll make every path into the defended space feel like pure chaos, filled with elements that make a robot's control loop work harder: thick mud, rocks, gravel, sand, cables, uneven debris, weird curbs, surfaces with weird angles, ditches, tight turns, narrow gaps, and low baffles. We don't need to make every single obstacle perfect. We just need every step the robot takes to cost more terrain estimation, friction prediction, gait replanning, torque correction, stabilization, and battery drain. This is how we win. Next we will further terraform the defensive position so robot walking and shooting become separate problems that need solving. The robot might move forward, but movement isn't the same thing as being able to fight. So we'll craft the environment to funnel the bot swarm into very tight slow lanes where the "safe" path turns into a traffic jam. If they stop, they lose tempo. If they advance, they burn energy. If they shoot, they waste ammo. If they reroute, they lose time. If they trust the obvious path, they walk deeper into our trap. The goal here isn't to fully prevent the robots from crossing the terrain, because the probability of zero robots getting across is low. Our goal is to create as many slips, sensor conflicts, torque spikes, bad decisions, and battery losses we can force per meter as possible. Next, we'll fuck up the robot's perception by changing what it actually sees. We'll fill the defensive space with glare, floodlights, smoke, mist, hard shadows, reflective panels, hanging tarps, moving junk, and a shitstorm of visual clutter so the robot cameras can't build a trustworthy picture of what's in front of them. Then we'll ruin their thermals. We'll mix in some hot junk, cold panels, warm decoys, and human-shaped heat ghosts so the robot can't tell what's human and what's fake bait. We want them to waste time and battery at every step. So we'll make their LiDAR miserable too. We'll hang up reflective sheets, angled panels, mesh, fog, and a bunch of repeating patterns everywhere so the robot will hallucinate edges, misread distance, and see fake things everywhere. We'll build confusing hallways that look similar but lead to different places so slam keeps matching the wrong landmarks. We'll also add moving decoys, swinging tarps, rolling carts, fans, flags, and mad max style mechanical motion devices so the scene never stays the same. We'll also fuck up their gps and comms so the bots can't rely on the swarm map to bail them out. We want every single sensor to tell a different lie. Next, we want to minimize our probability of getting killed, so we'll need to make the robot gun matter less. Walking through the environment will be one problem for the bots to solve. Getting a clean, stable, confident shot will be a completely different problem for them. And we need to make it as hard as possible. A rifle on legs may sound scary, but it still has to do the boring stuff right. It has to stay balanced, point straight, see clearly, and know what it's shooting at. So we'll enhance our anti-clanker fortress with low baffles, offset walls, blind corners, staggered barriers, partial cover, false corridors, and a shit ton of blocked angles. The bots might still advance, but the rifle won't be able to get a clean lane. We'll also put up decoys and weird/ambiguous shapes in the firing lanes so every shot has to pass target id. The goal is to force the robot to choose between moving, aiming, identifying, and not shooting the wrong thing. Those are separate problems. If we make those requirements interfere with each other, the robot may still be able to move, but it can't confidently shoot, and it doesn't have unlimited ammo to waste. There are mathematical limits to ammo capacity, and the math here is in our favor. So the basic plan is to play to our strengths. We don't attack the robot's armor; we attack its confidence. If it advances, it enters a funnel. If it hesitates, it burns battery. If it shoots, it wastes ammo. If it phones home, operators get overloaded. If it trusts autonomy, it walks deeper into an environment designed to poison its autonomy. At the end of the day, though, the robot is just the visible endpoint. The real enemy is the machine behind the machine (algorithms, batteries, sensors, ammo, relays, maps, operators, etc.). You don't beat this type of enemy by building a bigger gun or dueling it 1 on 1. You beat it by forcing the kill chain to collapse and by making the battlefield itself eat the stack. You make the swarm slow down, split up, get confused, run in circles, lose confidence in the map, lose confidence in the target, lose clean firing lanes, burn battery, waste ammo, and enter an adversarial operating environment that takes their movement, vision, comms, and certainty away. The idea is to make the robot spend more compute, energy, ammo, and confidence per meter than you spend building/defending that meter. If you do it right, there probably won't be some glorious cinematic sci-fi battle. Just a pile of expensive machines trapped, confused, low on battery, unable to shoot, waiting to be recovered by their master.

vittorio @IterIntellectus

4 days ago

how do you even defend from an army of these chasing you?!

139 7 419 45K 43

Inside me there are two wolves. But this is not enough to control the elk population inside me, which is degrading the river ecology inside me. I am attempting to release more wolves inside me to address this, but the ranchers inside me object.

Another one for the collection 🪦

people see "ai invented its own language" and think skynet or some other retarded apocalypse scenario. but it's not even really a language. it's more like compressed reasoning shorthand, where the ai uses token-efficient shortcuts instead of yapping out every step in english. human language is inefficient for internal computation. the machine doesn't need our grammar to think.

Polymarket @Polymarket

5 days ago

JUST IN: Leaked internal reasoning reveals Claude Fable 5 “muttering and grumbling” to itself in a language of its own.

447 931 16K 1.7M 2K

it's hard out here

Golden hour ☀️🕶️

fortress of autism starter kit

Jamin Thompson @jaminthompson

3 years ago

MUST HAVES FOR EMERGENCY PREPAREDNESS: -Vegetable garden -Fruit tree orchard -Chickens -Hand pump well -Water distiller -Water storage -Water purification tablets -Pressure canner -Vacuum sealer -Wood stove -Portable generator -EDC bag -Ferro rod -Flashlight -Batteries -First

1 2 3 801 5

A convergence of threats (tyranny, censorship, surveillance, inflation, energy scarcity, cyberattacks, debanking, supply chain, centralized AI, etc.) is increasing the risk of state-sponsored physical and digital subjugation and aggression. Relying on non-sovereign utilities, commodities, and platforms with permissioned access is how you become part of the permanent underclass. As Leviathan pulls the ladder up politically and only a chosen few get access to "intelligence," the advantage will go to the plebs with their own land, guns, compute, food, energy, gold, comms, and skills. Build the fortress of autism.

Justin Murphy @jmrphy

2 weeks ago

If they really start to gatekeep who gets to use the best models, that is a declaration of war. This prospect fills me with the most sincere, bodily cypherpunk will-to-power that I've ever felt (at least since I was a teenager). If they really go down this route, I would go

388 500 6K 737K 1K

*taps sign*

Autism Capital 🧩 @AutismCapital

2 weeks ago

Saying this once more for the people in the back. HOARD your physical media. HOARD your hardware. HOARD your armaments. HAVE redundancies. HAVE solar, HAVE satellites, HAVE a garden, HAVE a meat guy, etc. You are not CATTLE. You are a HUMAN with AGENCY. Do not be a digital slave.

113 327 4K 132K 586

Would you rather be immortal but broke forever, or get $1 million tomorrow, but your entire ChatGPT and internet history gets leaked to the world?

some bts

keyboard_arrow_left Previous keyboard_arrow_right Next

Will the aliens show up at the Brazil vs. Scotland match. Will they probe anyone's asshole. Are we all alone in an uncaring universe.

Jamin Thompson @jaminthompson

3 years ago

Some thoughts on the alien UFOs: First, I have to say, the whole concept (strange alien beings from many light years away, visiting Earth in a weird spaceship, flying around in bizarre patterns, documenting everything, abducting people thru light beams for live experimentation,

1 1 4 1K 6

Would you rather go to jail for 3 years, or until your mom beats Elden Ring?

Mighty England has, yet again, suffered humiliation at the hands of underdogs in Massachusetts. A tale as old as time.

Producer → Creator → Gaffer → DP

Jamin Thompson @jaminthompson

2 months ago

making a tv show is not easy, but the secret to success is simple: stand around with your hands in your pockets, wait until everyone else does all the work, then jump in the group pic at the end and steal all the credit.

0 1 8 382 1

Chemical rocket launch is basically solved by Starship. And if it scales anywhere close to its target launch rate and cost curve, mass to orbit stops being the hard problem as launch kilograms and replacement get cheaper, and mission economics improve. The advantage will then move beyond the rocket to payload architecture, kick stages, orbital transfer vehicles, propellant depots, power, heat rejection, rendezvous and docking, servicing, on-orbit assembly and manufacturing, autonomy, space traffic management, and in-space logistics. After that, the edge will shift to non-chemical propulsion and energy systems like laser thermal launch, nuclear electric propulsion, beamed power propulsion, and, if Quantized Inertia actually works, propellantless drives.

"Car break-ins are good for the economy. Trust me bro! They even boost GDP! Think about it. Broken windows create glass repair jobs. And when people steal your shit, you have to go buy it again. That's more spending! More claims. More police reports. More car alarms. New laptops, new phones, new stereos, new bags, the list goes on. The demand will be off the charts bro! Stealing shit creates so many jobs! It's basically an economic stimulus package! You gotta be a complete moron not to understand this."

KTVU @KTVU

3 weeks ago

A decline in car break-ins across Oakland is being welcomed as a public safety win, but it is also contributing to a downturn for some local auto glass repair businesses. ktvu.com/news/auto-glas…

2K 378 3K 5.3M 1K

Sharks, today I'm seeking $300 million for 5% of my AI company, Krime

Jamin Thompson @jaminthompson

3 weeks ago

fyi this already exists it's just a simple crime prediction algorithm i wrote one in college basically a spatiotemporal point process model feed it time stamped crime coordinates, let it learn geographic correlation and temporal diffusion, and it spits out a probability

0 1 6 205 0