The map of the United States looks calm for about one second. Then red cells begin crawling inland from the East Coast. Blue territory shrinks. A bar at the top starts leaning the wrong way. The British are coming, and this time they brought a cellular automaton.

Myles built Game of Liberty as a 90-second territory battle. The player begins with a Patriot stronghold in the heartland. Redcoat forces spread across neighboring cells and arrive in stronger waves. Leave the game alone and the country can fall in roughly twenty seconds. It is an excellent cure for indecision.

The player fights back by ringing Liberty Bells. Each click sends a blue surge through nearby land. Bells need time to recharge, so random clicking quickly becomes a poor military plan. The real choice is where to push, when to wait, and which front looks one bad second away from collapse.

Ring first, plan second

That makes this entry feel different from the runners and shooters around it. Those games put one patriot on the road or battlefield. Game of Liberty hands the player the whole map. One click is small, yet its effect can roll through dozens of cells. Suddenly the revolution looks less like a duel and more like weather.

The chunky graphics help. Red advances. Blue resists. White flashes mark conversions. The score bar tells the story without a speech. The shape of the country remains familiar, but the game asks the player to think about it as connected ground. Losing one area changes the pressure on the next.

Myles also tips his tricorn hat to Eric's Megaman 1776 and Adam's Paul Revere game. Those entries made the player one person inside the fight. His answer moves the camera into the clouds and says, “Fine. You run the whole revolution.” Friendly competition rarely announces itself so neatly.

The map starts talking

Game of Liberty leaves one strong image behind: a national symbol that refuses to sit still. The flag colors have jobs. The map can panic. Liberty arrives through timing, attention, and a bell that always seems to recharge one heartbeat too slowly.

Adam built three machines for Week 2. One runs a town. One feeds a town. One turns an innocent sentence into a parade announcement. They look unrelated at first. Spend a few minutes with them and a family resemblance appears. Each begins with something ordinary, then lets consequences through the door.

Engines of Independence starts in an 1880 charter town. The player adds roads, rail, farms, factories, banks, newspapers, power, water, and public works. Growth feels good until the bills arrive. Coal brings power and smoke. Credit brings construction and panic. A bridge helps everyone, then asks to be paid for.

Food Engines moves from city hall to the dirt. The player gets 21 days and six actions per day. Crops need water and time. Food can spoil. The school wants lunch. The firehouse wants chili. The pantry wants enough stored food to face the final week without developing a cracker-based emergency plan.

Every solution sends a bill

Rain barrels, compost bins, greenhouses, kitchens, canneries, and root cellars turn crops into a working food network. The cheerful farm grid makes the rules easy to learn. Then weather and town requests arrive. A tomato may be cute, but it has appointments.

Manifest Notebooky takes aim at software that offers freedom with several exciting conditions. Type “I have a meeting,” and Mandatory Murican Autocorrect may turn it into a town hall battle report. Switch on MAXIMUM MURICAN and the setting stays locked for five minutes. Changing it back requires Form 1776-B.

The complaint form delivers the sharpest joke. The notebook takes the user's dissatisfaction and processes it into praise. A tool meant to hold private thoughts becomes a tiny office with flags, stamps, a cooldown timer, and strong views about your attitude. The app runs locally, so the invasion remains theatrical.

The notebook seizes the keyboard

Across all three projects, the machinery is funny because the trouble makes sense. Towns create costs. Food takes planning. Software settings can trap people inside choices they barely made. Adam gives each idea a bright coat of paint, then waits for the gears to start making suspicious noises.

The home screen does not offer a sunset photo or a list of charming coffee shops. It says YOUR TOWN IS NOT FINE in giant black type. The page looks like somebody stole a city report, ran it through a bad copier, circled the alarming parts, and stapled it outside a basement show.

Enter a town, county, or ZIP code. Heather's app turns it into a county-level file when possible, then checks three things people live inside every day: air, access to food, and damage from weather. These are not vacation ratings. You do not get to unsubscribe from breathing.

Three things nobody can opt out of

The air section uses EPA annual data from 2020 through 2024. It says clearly that this is recent history, not live air quality. The food section starts with the USDA Food Access Research Atlas, then rolls tract information up to the county. FEMA supplies weather and natural-disaster declarations.

The app creates a NOT FINE SCORE from zero to 100. Air and food each make up 40 percent. Disaster history adds 10 percent. Missing or odd source data adds the final 10 percent. Heather labels it as an app-made signal, not an official grade from any agency.

The first version found a very real problem: FEMA's biological disaster entries could make COVID the leading event in places where the page was supposed to discuss weather. Heather filtered the file to the right incident types. She also shows the geography used by each section, because a ZIP, county, and census tract are three different creatures wearing the same town name.

A score with a warning label

The loud design is more than clothing. A soft civic dashboard can make bad conditions look comfortably managed. This page refuses the beige wallpaper. Even so, its hardest line is aimed at the conditions, never the people living with them.

The final section gets practical. Find the county alert service. Check live air readings before hard outdoor work. Learn the flood map. Know where the next reliable grocery option is. The page arrives yelling, then hands you a list. That is surprisingly decent behavior for something dressed like a punk flyer.

Final weeks are supposed to behave. Everyone finishes a loose end, poses for a group photo, and agrees the whole month followed a sensible plan. Summer Into AI closed with a black spy plane, half a million glowing points, and British soldiers learning where Paul Revere likes to ride.

Black Budget Bird began with Eric testing how far an AI helper could push Blender. The first SR-71 shapes were rough. More attempts produced a plane good enough to move into a Three.js flight game, complete with cockpit view, high-altitude flight, cloud condensation on the glass, and a few steering bugs still clinging to the wings. Then the game became a storyboard. The storyboard became a music video. Naturally.

Three unreasonable sequels

More Perfect returned to Myles's particle flag from Week 2. Instead of holding one fixed symbol, hundreds of thousands of points now form any words the visitor types. Text is drawn on an invisible canvas. Each bright pixel becomes a target. Pull the letters apart with the cursor and the crowd finds its way home.

Midnight Machine borrowed Adam's four-lane Paul Revere game and gave the redcoats a tiny learning brain. Four times each second, it watches lane choices and movement habits. Against a rider with patterns, Myles's tests raised its catch rate from about 25 percent to 41 percent. A wildly unpredictable rider made it nearly useless. The machine learned a lesson about humans, then humans learned to wiggle.

Each project starts with something that already worked. Then each one changes the kind of thing it is. A Blender experiment becomes a music video. A flag becomes a public writing surface. A simple arcade game becomes a duel with your own habits.

Reuse changes the genre

That is more interesting than polishing the old screen and calling it version two. Creative reuse can be wasteful, funny, and gloriously sideways. It may add a cockpit when a menu would do. It may also reveal the idea that was hiding inside the first project.

The final week never settled into one house style. Good. A competition should end with a larger set of possible moves than it began with. This one ended with afterburners, particles, and redcoats who had become alarmingly attentive.

USA-OS 2.5.0 arrives without a new world or a dramatic name. It has a version number, which is software's way of putting on work gloves. Kyle went back to his Week 1 game, played it again, and started adjusting the parts that only become clear after the first rush of building.

He had larger plans for the final week, but tool work pulled him sideways. So the update focused on the game already in front of him. More achievements went in. The play was rebalanced. The ending received more attention. This is less cinematic than adding a moon base, though moon bases are famous for missing their deadlines.

The glamorous life of version 2.5.0

Kyle will not explain the new endgame because he wants players to find it. He hints at a cheat code and says unlocking every achievement may reveal something special. The patch post offers just enough information to make the old game feel unfinished in a new and encouraging way.

The tools changed too. Hermes and Codex helped with development. Kyle continued refining a local image-generation process, while GPT 5.x still handled sprite work for this release. Stable Audio 3 supplied sound effects. He lists the parts plainly, including the image process that still needs more work.

Balance sounds like housekeeping until a number moves. A longer delay can turn a safe choice into a risk. A cheaper action can become the only action anyone uses. An achievement can teach a player to try a corner of the game they ignored. Small adjustments rewrite behavior without changing the title screen.

Tuning changes the story

Endgame work matters for the same reason. A prototype can win attention in the first minute and lose it in the tenth. A revision asks whether the idea remains fun after the surprise wears off. That question is harder than inventing another feature, because the player already knows the trick.

USA-OS 2.5.0 closes the season with maintenance, secrets, and a public repository. It treats Week 1 as something worth returning to, not a disposable draft. The quietest Week 4 move may also be the most grown-up: play the thing again, notice what drags, and fix it.