Making HollowARK: the devlog

Oct 13, 11:49 pmdesign

I want to make a video game

“I want to make a video game”Kraig, 11:49 pm

“2D alien world colony game”Kraig, 11:50 pm

“Top down”Kraig, 11:50 pm

Claude: Claude's first suggestion was a calm hiking game built on Kraig's real routes. He wanted an alien world instead. Claude then offered a RimWorld-style colony with three colonists, and he turned that down too.

• The game is 2D, seen from above, on an alien world.

Oct 13, 11:52 pmdesign

Start with a base, build drones

“Not what I want”Kraig, 11:51 pm

“Start with a base. Build resource collector drones”Kraig, 11:51 pm

“Still designing”Kraig, 11:52 pm

“Like Factorio”Kraig, 11:52 pm

• You start with one base and build drones that collect resources.
• Like Factorio, the game is about building automation that grows. The difference: drones carry everything, so there are no conveyor belts.

Oct 13, 11:57 pmdesignmechanic

Chains of drones, three temperatures

“So drones have a limited distance, to be chained together for longer distances, and also upgraded for speed and cargo and strength of carrying weight.”Kraig, 11:54 pm

“Yeah, let's keep it simple. Let's just have room temperature, hot and cold. And then iron ore, copper ore, molten ores.”Kraig, 11:57 pm

• Every drone has a limited range, so far deposits need a chain of drones.
• Materials have a weight and one of three temperatures, and those decide which drones can carry them.

Oct 14, 12:00 amdesignmechanic

Power, smelters and ice

“All the buildings are connected with lines for power, and that transports power between them. There's power generation, there's solar, there's batteries for collecting, there's a day and night cycle. There are smelters for processing raw ores into metals.”Kraig, 11:59 pm

“I think ice should be a material, a resource, and it's the only way that we make water. And the ice can be used to cool the melting process.”Kraig, 12:00 am

• Solar power by day, batteries for the night.
• A smelter turns ore into metal; a melter turns ore into molten metal for bigger parts.
• Ice is the only source of water, and it keeps the melters cool.

Oct 14, 12:03 amdesignart

The look: Foundation and Dune

“Factorio is like steampunk. I want this game to be very simplistic, minimalist design, but kind of Apple-esque. Kind of like Foundation or the Dune.”Kraig, 12:02 am

“I like the light gray stone, white light, colder, more sci-fi feel.”Kraig, 12:03 am

Claude: Claude turned that into rules: big calm ground with no clutter, plain geometric shapes, almost no colour, and the day and night cycle carrying the mood.

• Light grey stone, white light, cold and sci-fi.
• Colour only where it matters: live power glows, molten metal is orange, ice is pale blue.

Oct 14, 12:06 amdesign

Enemies with a purpose, and lasers

“Like Factorio, there's pollution and bugs. But I never really liked how they do it in Factorio. It just seems like an annoyance. And I usually turn off enemies in Factorio. I want this game to be more objective and mission-based.”Kraig, 12:04 am

“A drone can almost become like a tower. So like a defensive network of drones. But they'd have to be powered and fed. Maybe laser based if it's power. Makes it easier than making bullets. I feel like bullets are stupid anyway.”Kraig, 12:06 am

• Bugs attack the base now and then, but the real game is going out to hunt them and their bosses for rare materials.
• Combat drones are the same drones with lasers, running on power. No ammunition.

Oct 14, 12:07 amdesign

Never settle: the convoy

“In Factorio, I don't like that you kind of start building the base and never really leave that area. I like the idea of a convoy, or, you have to, every couple of days, decide when you need to pack up and leave.”Kraig, 12:07 am

• The base is always on the move. Staying to mine longer or packing up before the attacks get worse is the main decision.

Oct 14, 12:10 amdesignstory

The story: crash-landed and hunted

“There should be one foreboding enemy pressure or enemy army that builds and builds and builds, and if you're really good at the game you can go far, but eventually everyone dies.”Kraig, 12:09 am

“There's just this one vehicle, command vehicle, that has people in it that live in it, and they control all of the drones around them to survive, and they just need to keep moving, collecting resources, automating, upgrading as they're being hunted, because they crash landed on an enemy planet.”Kraig, 12:09 am

• A crew of five lives in one command vehicle and steers a swarm of drones.
• One enemy force, later named the Tide, grows forever. There's no rescue and no win screen.
• Each run gets a new map. Your score is days survived plus distance travelled.

Oct 14, 12:13 amart

The first mockup

“It's a lot simpler than I expected, but maybe that's the best way to start.”Kraig, 12:15 am

“Let's keep giving details until we are ready to do the first playable version.”Kraig, 12:16 am

• Two still pictures, not the game yet: the camp on grey stone by day, and the same camp at night under attack.

(open full size)

(open full size)

Oct 14, 12:27 amdesignmechanic

Everything is a drone

“Storage drones don't make sense. You can connect drones. They don't have to be nodes. You can just have a drone be stationary and it can be a drop-off spot for another drone moving cargo.”Kraig, 12:17 am

“The enemy grows in difficulty and intensity, but you for sure can defend against the main attack if you're on track and strong and growing, so you can stay and fight. But if you want to pack up and leave and not have to fight, that makes sense as well. There's strategy to that.”Kraig, 12:18 am

“I want them all to be movable ships, but they pack into one main ship... you can move a base of like six buildings, but it unpacks to have like 40 or 50 different things.”Kraig, 12:21 am

“It should always be a spectrum of choice. Each with pros and cons.”Kraig, 12:25 am

“Packing up and leaving, you can only go as far as you have power.”Kraig, 12:25 am

• Any parked drone is a drop-off point, so chains link up without special drone types.
• Buildings are ships that unfold as they grow and fold back up to travel.
• A design rule for everything: every choice is a spectrum with pros and cons, never one right answer.
• Batteries are your travel range.

Oct 14, 12:31 amdesign

Keep it simple

“I think it seems good at first, but now, the more I think about it, the more it might be annoying from a play perspective.”Kraig, 12:31 am

Claude: Claude's answer: only materials are physical cargo (ore, metal, ice). Parts are recipes, not items: an upgrade is just a cost taken from one shared stockpile, so nobody manages warehouses of gears.

• Kraig agreed. Hauling materials is the puzzle; spending them is one click.

Oct 14, 12:32 ammilestone

Build it while I sleep

“I think we're at the point when we should just build a working version. I'm going to bed... You can take as much time as you need to build something that's really cool that I can play in the morning when I wake up.”Kraig, 12:32 am

“I guess trackpad and keyboard on my Mac is how I would play. I don't know what that would be like. A browser-based game or whatever, I don't know. Whatever's best. You know better.”Kraig, 12:32 am

• Claude picked a browser game in plain JavaScript and Canvas, played with a Mac trackpad and keyboard.

Oct 14, 10:54 amdesignstory

The big idea: start inside a mountain

“So a big idea that I have is for us to move the environment of the game into a mine. And instead of on the surface of a planet, I think we should start inside a cavern that's enclosed. And the beginning tutorial happens while we're still encased in the rock. And as we dig and excavate, it reveals resources, different storylines, and then eventually opens up into the next bigger environment. And that environment could have the next mechanic. Maybe enemies.”Kraig, 10:54 am

“So the tutorial phase is before you open up and reveal yourself to the enemy... Maybe we start with geothermal and eventually we get to solar.”Kraig, 10:54 am

“Yes, get it to a point when I can play test it quick”Kraig, 11:04 am

Claude: Claude's take: starting buried gives the slow, safe tutorial a reason. Nothing knows you're there until you break out. The plan came in three stages: the buried start, then a great cavern with the first enemies, then back to the surface and the sun.

Oct 14, 11:31 ammilestone

Let's share it

“So I'm really liking this. It's really fun. And I think it would be really cool to document and share and create content about the behind the scenes and an entire creation, because people might be interested in how I'm making this with Claude.”Kraig, 11:31 am

“And once I'm ready, I want people to be able to play this as I build it and give feedback. Honestly, if we get to a point today where I can share it with people and start getting feedback, that would be really cool.”Kraig, 11:31 am

• This page. Claude built the devlog from the conversation and the code history while the game kept changing, and rebuilt old versions from the code to screenshot the stages that had no pictures.

Oct 14, 11:48 amstorydesign

Open question: why is the Ark in a cave?

“So now that we're starting in a cave and it's becoming like a cave exploration game, it doesn't really make sense that this ship crash lands... How do we explain the beginning of it?”Kraig, 11:43 am

“Let's wait to do this”Kraig, 11:47 am

Claude: Claude suggested the crew "went to ground": the Ark came down hurt with the swarm on its trail, drove into a mountainside cave and brought the entrance down behind it to hide. Kraig parked the idea for now.

Oct 14, 1:53 pmdesignstory

Where it's going (planned, not built yet)

“Because I want the game to have surface and cave, but you start in the cave. And once you leave the cave, that's when like the actual enemy mechanics start to come in and other mechanics. But then there's also reasons why you want to go back into the cave system to like mine special resources or extract certain things... So I feel like there should be a starter base, a middle base, and then like an end base, and the end base should be like more permanent.”Kraig, 1:38 pm

“I think the tutorial chapter should end by breaking out into the surface. And I think that's actually like a really cool story point because you see like where you are... the systems turn on, the crew is still sleeping in cryo... we're in a tiny little cavern, the Ark is like stuck there for some reason, we don't understand what happened... It's literally like an Ark, like a civilization sent out a seed to restart civilization”Kraig, 1:43 pm

“The opening should start black and it should have like a chapter one. The seed or the awakening or system boot and the lights and the like indicators on the Ark should wake up and like a booting sequence and then the first drone should come out and we should start digging stone to find the ore”Kraig, 1:48 pm

“I think the outside should look very different and alien but beautiful and colorful compared to like the dark spooky cave system that gets slightly more beautiful with the bioluminescent but ultimately is like cramped and dark. But also like feels safe in terms of like being able to bottleneck and control and build walls against enemies that are trying to come in versus the like beautiful expansive outside.”Kraig, 1:51 pm

“I think the ultimate ending should be that we need to terraform the planet so that humans can live outside without habs or suits. And terraforming for us also means unterraforming for the alien species, the bugs that live there. So like ultimately end game, we're like building huge systems to terraform the atmosphere for us and against the enemy.”Kraig, 1:52 pm

Claude: Claude wrote each idea into the design notes as it came and suggested holding the build until the shape settles, so none of this is in the game yet. Its draft of Chapter One: a black screen and "CHAPTER ONE · THE SEED", a boot log in the dark (power 3%, location unknown, date unknown, colonists in cryo: 400, stable), the Ark's lights coming on section by section, the first drone flying out to dig toward the ore, the tutorial's repairs in order, a seal too hard for the drills, and daylight.

• Planned: the Ark is a seed ship, sent out to restart a civilisation. Its systems boot in a tiny cavern with the colonists asleep in cryo, and nobody knows where or when it is. The guidance becomes the Ark's own terse system log.
• Planned: Chapter One, "The Seed", ends by breaking out to the surface. The Ark reads the stars to learn how long it slept, the first colonists wake, and air, water and food begin.
• Planned: three bases. A starter base that packs up fast, a middle base worth staying at, and an anchored end base that stops moving for good and holds the biggest machines.
• Planned: the caves stay dark, cramped and safe, with tunnels you can wall off; the surface is colourful, alien, wide open and exposed, and it's where the enemies come in. Some resources are found only underground, so you keep going back down.
• Planned: the ending is terraforming. Huge machines push the planet's air toward breathable, which makes it worse for the bugs, so the swarm fights back harder the closer you get.
• Also in the notes: one rock with a hardness percentage instead of rock types, stone from digging as a resource, and a storage limit with a Storage module.
• Reshaped later the same afternoon (next chapter): the cave flora, not a geothermal vent, wakes the Ark; Chapter One leaves the AI core, drone upgrades and Surveyor for Chapter Two; and the Ark itself, not a drill upgrade, breaks through the seal.

Oct 14, 1:57 pmdesignstoryplanned

A space programme and a city to defend (planned, not built yet)

“We can do the terraforming on our own from the surface, but it takes a lot of time and very costly and very difficult. So part of the story should be like building a space program and bringing down those materials so that we can like finalize our designs for surface terraforming to like beat the game.”Kraig, 1:55 pm

“The bugs will become more incentivized towards the end to like send out waves and bosses and try to destroy those things. So the end game becomes more like defending the big logistics city that you've built.”Kraig, 1:56 pm

• Planned, the arc of a run: Buried → Nomad → Settled → Anchored → Terraform. Dig out of the caves, scout the surface with a base that packs up fast, settle somewhere worth staying, anchor for good, then remake the planet.
• Planned: mid-game, a communications array makes contact with orbit and finds the seed kit still waiting up there. Terraforming from the ground alone is possible but slow and costly; a space programme, single launches at first and a space elevator at the end, brings the kit down to finish it.
• Planned: the anchored base is a city you lay out yourself around the uplink, the launch pads, the elevator and the terraforming machines, with permanent towers and walls. The swarm's waves and bosses go for those, harder the closer you get to the end.
• Rare resources found only underground feed the launches and the machines, so the caves matter to the very end.

Oct 14, 2:05 pmdesignstoryplanned

The story: you are ARK, the Ark's own AI (planned, not built yet)

“I want, you know, AI to be a big part of this. And the player isn't a person. You're the AI system that's helping the humans. So I want that to be like a narrative.”Kraig, 1:58 pm

“I think the Ark is buried there for a while because it doesn't have power. And eventually a little bit of bioluminescent grows in that section. And the Ark is able to get the tiniest trickle of power from the small amount of biomass. And then that's what wakes it up.”Kraig, 2:05 pm

Claude: Claude suggested SEED as the AI's name. Kraig: "I like Ark better than seed", so the ship and its mind share one name, ARK, and Chapter One keeps "The Seed". Claude's first answer to why the Ark wakes was a geothermal vent slowly charging its batteries; Kraig's glowing flora replaced it.

• Planned: you play ARK, not a person. Drones are its hands, its system log is the voice, and the colonists sleep in cryo. Today's AI labs and reusable rockets are the inspiration, with no real company names in the game.
• Planned: the Ark was one ship in a convoy, with cargo landers and a seed kit for terraforming. Something hit the convoy as it arrived: the Ark buried itself in a mountainside, the landers crashed and scattered, and the seed kit is still in orbit, waiting for a signal that never came.
• Planned: the opening is a black screen, then a faint glow as cave flora grows round the dead Ark, its lights coming on section by section, and the boot log: "Power: 3%".
• Planned: the data cores found in the world are the crashed landers' computers, weathered on the surface and better kept in buried wrecks, which is one more reason to dig.

Oct 14, 2:07 pmdesigntutorialplanned

Chapter One: a power famine inside a seal (planned, not built yet)

“I also think chapter one shouldn't dive into the AI systems yet... chapter one is just like mining, smelting, getting the Ark up and running so it can burrow out of the cave system to the open world. And also we should do hardened rock... So even if a person wanted to mine endlessly... the tutorial seal, like borderline would kind of force them out into the open eventually.”Kraig, 2:07 pm

“Like bioluminescent gives some power, but we're almost like always starving. And it's only until we like finally get out into the actual surface with the sun that like the power issues can actually be solved.”Kraig, 2:05 pm

Claude: The morning's draft ended Chapter One at a seal too hard for the drills until a drill upgrade. With the upgrades moved out of Chapter One, Claude suggested the Ark itself breaks through instead.

• Planned: Chapter One is only mining, smelting and getting the Ark running, with about three drones. The AI core, drone upgrades, the Surveyor and datacenters move to Chapter Two.
• Planned: the tutorial caves sit inside a ring of hardened rock no drone can cut, and what's inside runs out, so the only way on is out. With its engines and cutter restored, the Ark bores up through the seal.
• Planned: underground, power is always short. The flora is the only source (the geothermal vent goes), so even green is fragile, and the sun at the breakout is the relief.

Oct 14, 2:13 pmdesignplanned

Anything is possible, nothing is easy (planned, not built yet)

“I just really like the idea of a player being able to do almost anything, especially once they escape the tutorial phase, but it just sucks to do it. It takes forever, it's complicated, it's difficult, and it's just time consuming. Like I love that aspect of Factorio, like you're kind of just in a sandbox from the start able to do whatever. But you need to build up systems to build more complicated things.”Kraig, 2:12 pm

• Planned, a rule for everything after Chapter One: no hard locks. Research and modules change how fast and how well you can do something, not whether you can.
• Planned: everything has a slow way by hand. The Fabricator makes almost anything, slowly, and drones take one order at a time. The reward is systems: modules that do one job fast, mining lines and haul routes that run themselves, automation rules, and compute that speeds up research.
• Even terraforming could start crudely on day one. The satisfaction is building the machine that does it for you.

Oct 14, 2:19 pmdesignmechanicplanned

Waking the colonists, one at a time (planned, not built yet)

“I like the idea of like slowly, one at a time, waking these people up. And it's very costly to do it in compute... So if you bring, yeah, a botanist specialist, they just give like a bonus to like whatever you have going on with that. And you can just like keep stacking different types of bonuses.”Kraig, 2:16 pm

“And you can have all of your crew die in the Ark, and that's fine because you have more to wake up... a good mistake that's possible and probably should happen. And the game only ends when your Ark is destroyed.”Kraig, 2:17 pm

Claude: Choosing who wakes was the second of ten ideas Claude pitched. Kraig added the compute cost and the stacking bonuses, then the rule that colonists can die.

• Planned: Chapter Two brings the enemy, deeper smelting and building, and the people, and with them oxygen, water and food.
• Planned: each colonist is a specialist whose bonus stacks: a botanist for plants and food, an engineer for building and repairs, a geologist for surveying, a pilot for launches, a medic for the crew. You choose who wakes by what you're working toward.
• Planned: life support is one gauge, coloured like power and checked about once a minute. Green: nobody dies. One of the three low, yellow: about a 5% chance a colonist dies at each check. Two low, orange: about 20%. All three low, red: about 50%. Any one run out: someone dies every check. The numbers get tuned in play.
• Planned: colonists can all die, taking their bonuses and the compute spent on them, and more can be woken. Only losing the Ark ends the run.

Oct 14, 2:23 pmdesignmechanicplanned

Compute, the game's currency (planned, not built yet)

“I feel like compute is like almost an even better, like all-encompassing resource to like spend stuff on. So like every large project needs a certain amount of compute and better models like lower the cost of compute.”Kraig, 2:09 pm

“It's the spectrum idea. Like, there's kind of two polar ways to play, and there are pros and cons to equally... your preferences, and also like the situation of your map and starting position, like would make you favor one or the other.”Kraig, 2:22 pm

Claude: Claude's first take was compute as a rate only, nothing stored: a project finishes faster the more compute runs. Kraig wanted it stored too, so it now works like the power and battery.

• Planned: compute arrives in Chapter Two and pays for every big project: research, ARK's own upgrades, waking colonists, decrypting data cores, launches, the elevator and terraforming. Each better model makes every recipe cheaper.
• Planned: datacenters make it each second, a store holds a surplus up to a cap, and research draws it at a rate. When it runs short, research slows, the way low power slows the base. Power runs the base; compute moves ARK forward.
• Planned: datacenters need power, water, space and protection, and can sprawl. Cooling is one of two ways to play: in the caves cold rock cools them for free, safe but cramped and short on power; on the surface they need water but get room and sun, and run cooler at night. Heat sinks help either way.

Oct 14, 2:26 pmdesignenemiesplanned

How the enemies notice you (planned, still open)

“I don't want like sound dampening... Maybe it's another dichotomy thing where like building and doing production and stuff in the caves does produce vibration and that alerts like pockets of enemies within the cave... And then also like when you're outside, it's just visual. So it's like line of sight... With Factorio, you kind of like knew exactly what the bugs would do. But I kind of like the idea of like kind of knowing, but also you could be attacked at any point.”Kraig, 2:26 pm

Claude: Claude pitched noise as the swarm's one sense: digging, smelters and above all launches draw bugs, with damping and quiet modes as the counter. Kraig turned it down, and didn't want a copy of Factorio's pollution either, then reshaped it into the split below.

• Planned: in the caves, work sends vibration through the rock and can stir nearby pockets of cave creatures. On the surface it's sight only.
• Planned: the bugs have their own fog of war and send out scouts now and then. Colonies mostly keep to habits you can learn, with enough randomness that an attack can come at any time.
• Still open: how the enemies spread, decide to attack, build, upgrade and change. Each gets its own design talk.

Oct 14, 2:29 pmdesignplanned

Two kinds of cave plants (planned, not built yet)

“So the bioluminescent flora in the caves is like very specifically just for lighting, like the day and night cycles... flora that we harvest and use as a resource. I think it should be different. I think it should be like patches that... can be used up similar to... iron or copper ore... but the difference is that it'll grow back slowly during day cycles.”Kraig, 2:29 pm

Claude: Claude pitched the glowing flora as living tech: plant cuttings to grow light farms, bugs graze on it, terraforming slowly kills it. Kraig kept light and harvest apart.

• Planned: the glowing flora is light only. It lights the caves and runs the day cycle, and it isn't harvested.
• Planned: plants you harvest are separate patches, mined out like ore, that grow back slowly while biolight shines on them: mine one, let it rest, come back.
• Planned: the surface gets its own plants and creatures, different in each biome. Biomes are a design talk still to come.

Oct 14, 2:29 pmdesignplanned

Ten ideas from Claude, and what Kraig did with them

“Give me a list of 10 ideas for anything with the game that you can come up with and be creative and improvements and suggestions and then I'll go through them and give my feedback.”Kraig, 1:59 pm

“Number one, I like that, but it has to tie into the mechanics and the resources or like bosses. There has to be more of a reason than just getting to a point, unlocking a thing and just getting like a block of text to read.”Kraig, 2:00 pm

“I think this complicates it too much. Maybe down the road. I don't think it would be exactly how you spelled out. I have my own ideas.”Kraig, 2:21 pm

Claude: Claude listed ten ideas, then went through them with Kraig one at a time, each with its own take.

• Reworked: 1, memory cores. Decrypting a data core costs compute, and each one gives something you use: a blueprint, a map to a rare resource or the seal's weak spot, a lasting bonus, or a boss's weakness. Some wake an old wreck's defences.
• Kept with Kraig's changes: 2, choosing which colonists wake, and 4, cooling compute in the caves, which became a choice between cave and surface.
• Turned down and reshaped: 5, bugs hunting by noise, became vibration in the caves and sight on the surface. Reshaped: 6, flora as living tech, became separate light and harvest plants.
• Parked: 3, trust and autonomy, the crew granting ARK more freedom as it proves itself. Ideas 7 to 10 are still to come.

Oct 14, 2:36 pmartstory

HollowARK, with ARK in capitals

“Let's just go with Hollowark. But can you make like the A capitalized? Or like A-R-K capitalized?”Kraig, 2:35 pm

Claude: Kraig asked for a rebrand to fit the new story. Of Claude's names he liked Hollowark best, asked for a few more built on AI and today's tech, and turned those down.

• The name is now written HollowARK, so the Ark, and ARK, the AI you play, sit inside it.
• On the title screen and the top bar, HOLLOW stays in ink and ARK is in biolight teal.
• The web address stays the same.

Oct 14, 2:38 pmdesignmechanicplanned

The Beacon: one ping, one find (planned, not built yet)

“I like it coming from the sonar tower or whatever we call that beacon. Probably just beacon... it should cost compute. So we can't build it in the tutorial... every ping should just find one resource... show an indicator that something was there, but you have to go explore to find it... mid game, when you're going through the cave systems to find rare stuff, you kind of need to set up, burrow through with defense, set up pings, and then ping to find where you're going to dig next.”Kraig, 2:38 pm

Claude: The seventh of Claude's ten ideas was a sonar ping. Kraig made it a building with a cost and a risk.

• Planned: the Beacon is a building you put up after the tutorial, a mid-to-late-game tool for hunting rare resources.
• Planned: each ping costs power and compute and finds one thing in its radius, shown as a rough marker. You still have to go and explore to find out what and where. Upgrades push the range and the strength.
• Planned: pinging gives you away. In the caves, creatures slowly burrow towards where they heard it; on the surface the sound carries and they come to look.
• The mid-game rhythm: push into deep caves, set up defences, build a Beacon, ping, choose where to dig, and defend while you dig.

Oct 14, 2:48 pmdesignmechanicplanned

The Ark always moves; what you leave behind is the base (planned, not built yet)

“We should always allow the Ark to get up and move. And any attached modules will just come with it, but there's a limit to how many modules you can have attached to it, just one of each kind... you can unattach modules from the Ark, but once you do that, it's permanent. They can't be reattached.”Kraig, 2:42 pm

“Every building should have a movable, packable version... everything coming out as a packed oval or whatever design we have. And then when we deploy it, it kind of sprawls out and builds with a slight animation.”Kraig, 2:47 pm

“I don't think there should be any limit to how many buildings you have. It just comes at a cost of complexity and power and just logistics and then your ability to defend it.”Kraig, 2:48 pm

Claude: Claude had asked how new buildings get built (placed instantly, assembled by drones over time, or a mix) and whether there should be a limit on how many. Kraig answered with pods, and no limit beyond what a big base costs.

• Planned: the Ark can always pack up and move. Its attached modules, one of each kind, fold up and travel with it and unfold wherever it makes camp.
• Planned: detaching a module is permanent. It becomes a standalone building, and a new attached one can be built at a cost. Detached buildings still move: they fold, travel and unfold where you send them.
• This replaces the idea of the Ark stopping for good at the end. The Ark stays free; the permanent base is what you detach and leave behind.
• Planned: every building comes out of the Fabricator packed, as a pod, and unfolds in place with a short animation. Late-game megastructures get a bigger builder of their own. Sizes and designs come last.
• Planned: no limit on buildings. The limits are complexity, power, logistics and having to defend it all.

Oct 14, 2:53 pmdesignenemiesplanned

Laser towers, hangars and veteran drones (planned, not built yet)

“Laser towers are probably the main idea... you just place them where you want to defend things, choke points around buildings, supply lines. I do think each drone should have a very minimal way to defend itself... maybe they stay in a hangar and they come out of the hangar if you're being attacked... they have to come back to the hangar because they run out of power because they're flying away from the supply grid and the Ark.”Kraig, 2:52 pm

“I also kind of like the idea of a promotion system... a very small and minimal automated promotion. So if a drone does work and doesn't die, it keeps getting small, tiny bonuses and you can incentivize smart strategic sorties that don't wipe out your entire fleet.”Kraig, 2:53 pm

Claude: Claude's suggestion was stone walls, power-hungry laser towers, guard drones and the Ark's own laser. Kraig kept the towers and added hangars, sorties and promotions.

• Planned: laser towers are the main defence, placed at chokepoints, round key buildings and along supply lines.
• Planned: every drone gets a small laser to defend itself, though a swarm will overwhelm it.
• Planned: hangars hold attack drones. They launch on their own when the base is attacked, or go out on sorties under your control. The battery is the leash: away from the grid they have to come home to recharge. Engine speed, battery size, a small solar trickle and the laser's draw are their stats, which fit the drone upgrades already built.
• Planned: drones gain experience on their own from working and surviving, over three to five ranks of a few percent each in their own job, shown as a small chevron or pips. Losing a veteran stings, so careful sorties pay.
• Attacking nests with the Ark itself is noted but not planned: losing the Ark ends the run.

Oct 14, 2:57 pmdesignenemiesplanned

Enemies guard the way forward (planned, not built yet)

“We should use enemies to guard areas that we want the player to go to eventually. So they should almost always guard resources or the next step or storylines or data cores... there's always the possibility that you might get attacked at any point.”Kraig, 2:55 pm

“I also want repairs to be simple... a very, very small and gradual health repair and recovery for all items. But maybe a quick fix, from the Ark, you can click one button to repair all buildings and units, but it scales to how much damage there is and it's costly.”Kraig, 2:56 pm

“Eventually, enemies or certain different types of enemies will have drops that are specific for doing something. I don't know.”Kraig, 2:57 pm

Claude: Claude asked why anyone would attack a nest, offering drops, better land or growing pressure. Kraig's answer covers it: they guard what you need. For damage, Claude suggested drones repairing nearby buildings for a little iron; Kraig wanted something simpler.

• Planned: nests and creatures guard resources, data cores, story finds and the way on, with bigger guards further along, so the enemies set the pace of the game.
• Planned: they grow and change slowly over time, tuned in playtests. Attacks come when you're close or they sense you, with a little chance on top.
• Planned: a cut relay drops power to everything past it, defences included.
• Planned: everything repairs itself very slowly. "Repair all" at the Ark fixes every building and drone at once, for a cost that grows with the damage: wait it out, or pay to be ready now.
• Still open: enemy types will drop materials for specific uses. One fit: samples from the swarm as research for the terraforming that poisons it.

Oct 14, 2:59 pmdesignmechanicplanned

One network, and every line leads to the Ark (planned, not built yet)

“There's always a power line from the Ark connecting everything. And if a module gets too far from the Ark, it'll need a relay to transfer the power and resources between. So basically a city will look like a big sprawling web of different buildings.”Kraig, 2:44 pm

“You can collect resources if it's an outpost, but it doesn't get added to your totals of inventory until the outpost is connected to the Ark... all lines kind of lead to the Ark.”Kraig, 2:46 pm

“For now we should keep it simple... right now, just longer lines, longer time. So it incentivizes building tight and complex close to the Ark to minimize the travel distances.”Kraig, 2:58 pm

Claude: Claude's first suggestion was a Hub building at the heart of each base, with bases linked by long lines. Kraig wanted one network rooted at the Ark instead. For moving goods, Claude offered travel time only, capacity limits, or one then the other; Kraig kept travel time only.

• Planned: power lines from the Ark carry both power and resources to every building, and relay stations stretch the reach. A city is a web you can arrange in clusters, branches or loops, and an outpost is a far branch at the end of a chain of relays. Smaller Arks as hubs of their own are an idea for later.
• Planned: an outpost can mine and stockpile on its own, but what it collects joins your totals only once it's connected to the Ark. The Ark is the heart and the swarm's target, so relay chains are every base's backbone, and cutting them is the enemy's best move.
• Planned: goods on the network only take travel time, with no limit on how much a line carries, which rewards building tight around the Ark. Line upgrades come later.

Oct 14, 3:01 pmdesignplanned

Four tiers of production, simple first (planned, not built yet)

“I like the four classifications. I think it's good to start simple and get it as a fun loop that makes sense and then expand and add bigger recipes, bigger production chains and complicate it more later down the road. But those would also be chapter three, chapter four.”Kraig, 3:00 pm

Claude: Claude proposed raw, refined, components and megastructure parts, and thought three tiers might be enough. Kraig kept all four.

• Planned: four tiers. Raw (ore, rock, ice, plants), refined (metal and the like), components, and megastructure parts such as elevator segments and terraforming cores.
• Start with one fun loop; the long production chains come in Chapters Three and Four. Chapter One first, then Chapter Two.

Oct 14, 3:08 pmdesignartstory

The tree: from seed to stars

“I like the metaphors with a tree. Like it starts as a seed, it pushes out of the dirt to see the sunlight, takes in resources and grows bigger until it's like a full huge tree stretching, reaching all the way into the sky and the stars above.”Kraig, 3:07 pm

“Sounds good.”Kraig, 3:08 pm

Claude: Kraig asked whether the tree could change the name. Claude suggested keeping HollowARK, since a hollow is already a tree word, the space inside an old trunk where things shelter, and making the tree the look and the chapter names instead.

• The chapters grow like a tree: The Seed (buried in the cave), Sprout (breaking ground into the sun, the first colonists wake), Roots (the network and outposts), Trunk (the permanent city), Canopy (terraforming) and Reaching (the space programme and the elevator to the stars).
• The look: the Ark drawn as a seed, a thin teal shoot rising out of ARK in the logo, lines that look like roots underground and branches on the surface, and the space elevator as the trunk into the sky. A tagline idea: "From seed to stars."
• HollowARK stays. The shoot in the logo went into Stage 1.

Oct 14, 3:10 pmdesignmilestone

Chapter One, built in four stages

“You can start building this first phase with all the steps. Or maybe like stop at each step and just like give me a report so I can okay the next step.”Kraig, 3:08 pm

“Yeah, you're free to start stage two as well.”Kraig, 3:14 pm

Claude: Claude wrote the plan as four stages, each one playable. Kraig asked whether building them all at once would be faster. Claude's answer: it costs about the same, and the stages build on each other, but the rock is separate enough to build next to the opening. So Stage 2 started alongside Stage 1.

• Stage 1, the wake: the new opening and ARK's voice.
• Stage 2, the rock: one rock with a hardness you can see, rock from digging, a storage limit, and the seal.
• Stage 3, the famine: flora the only power underground, three drones at most, the AI core, drone upgrades and the Surveyor saved for Chapter Two, and the tutorial in a new order.
• Stage 4, daylight: the Ark bores up through the seal, the sun floods the grid, ARK reads the stars to learn how long it slept, and "Chapter One complete".
• After each stage a test player runs the whole chapter, then a report and Kraig's OK before the next. Old saves start fresh.

Oct 14, 3:14 pmdesignmechanic

Rock becomes concrete

“One quick change, rock gets refined into concrete.”Kraig, 3:14 pm

• Digging yields rock, more from harder rock. A concrete mixer unfolds from the Ark like the smelters and turns rock into concrete.
• Concrete patches the hull and builds the Storage module, and later walls, defences and foundations.
• It went straight into Stage 2, which had just started.

Oct 14, 3:20 pmdesignmechanicplanned

Expeditions, outposts and one inventory (planned, not built yet)

“Cave expeditions, expeditions on the land. You're free to set up or build whatever you want, but to connect it to like your main power source, you need long lines or you need a separate power source... So you might as well build a power and supply line, but you don't have to, especially at the start.”Kraig, 3:17 pm

“A small amount of any resource can be stored in a module. If we like pack up and then leave the outpost area and come back to the Ark and connect, like we get those resources. So it's possible to like send a little outpost party out to get a special thing and then come back instead of setting up like a permanent line.”Kraig, 3:19 pm

“I kind of want to avoid sub-networks. So just to keep things simple, to use a resource to build anything, you have to connect to the Ark.”Kraig, 3:20 pm

Claude: This was the ninth of Claude's ten ideas, cave expeditions. Claude assumed a party could spend what it carried on the spot, to put up a wall or a tower; Kraig said no, since that would make little networks of their own.

• Planned: build anywhere, in the caves or on the surface. A building with no line runs on its own power: light collectors underground, the sun on the surface, batteries.
• Planned: resources reach the Ark's totals only over a line, and one line carries both power and supply, so connecting is the natural choice but never a rule.
• Planned: every module holds a little of any resource and keeps it when packed. An expedition party, a few drones with a power module and a storage module, goes out, mines, packs up full and comes home, and its load counts once it's connected. A trip or a permanent line: two ways to get far-off resources.
• Planned: one inventory, the Ark's. Anything you build or spend draws on it. A tower for a far chokepoint is fabricated packed at home, carried out and unfolded there.

Oct 14, 3:27 pmdesignmechanicplanned

Components, the Fabricator and factories (planned, not built yet)

“I think this is a good start. This is going to take a lot of iteration... a fabricator should just be able to 3D print any kind of module or drone or attacker.”Kraig, 3:25 pm

“One module that will make the modules and the drones... and like 3D prints it. But then there's like the other buildings that are more like the factories... you select your recipe and then you turn it on and it just continues to make those items if it has the available recipe items... you can upgrade the modules, or you could also build more modules with the same recipe, but that's like less efficient than upgrading.”Kraig, 3:27 pm

Claude: Claude proposed four starting components and one Assembler that could make any of them. Kraig first described makers in tiers by recipe size, then split them in two: one Fabricator for whole things, and factories for the parts.

• Planned, from Chapter Two: Frame (2 iron) for structure, Wiring (2 copper) for power, Motor (a frame and wiring) for anything that moves, and Circuit (wiring and crystal, found only in caves) for anything that thinks. Concrete stays a refined material. Chapter One is raw and refined only.
• Planned: the Fabricator prints whole things from a build queue: modules (packed), drones and attackers.
• Planned: factories make items. Pick a recipe, switch one on, and it keeps going while the ingredients are in the Ark's inventory, so factories feed each other through it, with travel time and no belts. Refiners take one input (the smelters, the concrete mixer), the Assembler two, the Heavy assembler three or four; the names are placeholders.
• Planned: to make more, upgrading is the efficient way (level 2 doubles the speed for one and a half times the power, in the same space). A second copy costs full price, full power and more room.

Oct 14, 3:29 pmdesignplanned

Build the whole game thin first, about two hours (planned, not built yet)

“Start simple and almost kind of like build the entire game with all the chapters so that you can win, but it's cheap and easy and quick to do so. And then we'll just keep layering in more complexity and detail.”Kraig, 3:25 pm

“We'll start somewhere and then we'll tweak it based on how hard it is, how easy it is, how long it takes, and then just add complexity... you can just increase the difficulty by adding a multiplication to or from the recipes for everything. So it just takes longer and more resources to do the exact same thing.”Kraig, 3:29 pm

Claude: Claude sketched a thin run with one simple goal per chapter, from digging out of the cave to bringing the seed kit down from orbit, and suggested about two hours to win.

• Planned: after Chapter One, every chapter gets a short, simple version first, so a whole run can be won end to end. Then each one is deepened from playtests: longer recipes, harder enemies, more steps.
• Planned: the thin run takes about two hours to win, about 35 minutes for Chapter One and about 15 for each of the rest, and grows longer as the chapters deepen.
• Planned: difficulty is tuned with a few global dials: one multiplier on every recipe's cost, plus how hard the enemies and the power push back.

Oct 14, 3:37 pmdesignmechanicplanned

Work the tree backwards from data centres and rockets (planned, not built yet)

“I think those make sense. And also we need to like go backwards from where we want to go. So like I said, space program, data centers. So that's like GPUs, heat sinks, optical fiber, and then eventually like rocket fuel, heat shields...”Kraig, 3:35 pm

“Yeah, I think that makes sense.”Kraig, 3:37 pm

Claude: Claude proposed four second-tier components, which Kraig accepted. Then, working back from his two end goals, Claude drafted the recipes with real-world parts under plain names, including methane harvested from the cave plants for rocket fuel.

• Planned: four second-tier components, made in the Heavy assembler from three or four inputs. Processor (two circuits and wiring) for compute; Emitter (circuit, wiring and crystal) for lasers, light collectors and the Beacon; Drive (motor, frame and circuit) for fast and attack drones, the Ark's engines and launch stages; Armour plate (frame, alloy and concrete) for walls, towers and hulls.
• Planned: the recipes come from the end goals, backwards. A data centre is GPUs, heat sinks, optical fibre and frames, plus water to cool it. A GPU (the Processor, renamed) is two circuits, wiring and crystal; a heat sink is copper and alloy; optical fibre is drawn from glass, and glass is fired from rock, so rock feeds both concrete and glass.
• Planned: a rocket stage is a rocket engine, a fuel tank, a heat shield and a GPU for the avionics. The engine is a drive, alloy and a heat sink; the heat shield is ceramic tiles fired from rock on armour plate; rocket fuel is methane from harvested alien plants plus liquid oxygen from ice.

Oct 14, 3:40 pmdesignmechanicplanned

Food, and life in the caves and on the surface (planned, not built yet)

“But also remember, we want flora and fauna on the inside and outside of the cave system. And we also have to have a recipe tree for food and yeah, anything else you can recommend?”Kraig, 3:37 pm

“Yeah, I think all these look good. What's the next question?”Kraig, 3:39 pm

Claude: Claude worked food backwards the same way, from what the colonists eat down to what the caves give, and recommended three more trees to plan like it.

• Planned: plants and creatures both in the caves and on the surface, with different kinds in each.
• Planned: food in three steps up. Algae, grown in vats from water and light, is the easy cave food. Crops from the Ark's seed bank grow in a greenhouse on soil, water and light, and soil is rock dust and composted cave plants. Native food, alien plants and creatures made safe in a lab, is the richest, and it teaches the biology terraforming needs.
• Planned: a kitchen turns them all into rations, the one food the colonists eat.
• Planned: three more trees worked backwards the same way: medicine from plant extracts (lower death odds, safer waking), what terraforming needs (air, soil, the seed kit from orbit) and the space elevator's cable (carbon from methane, then carbon fibre).

Oct 14, 3:42 pmdesignplanned

Terraforming: one number for the air (planned, not built yet)

“I'd say it should boil down to like one percent and then all of the other metrics should like affect it. I think all of these are pretty good in the right direction... swarm samples make a lot of sense. Yeah, for speeding up like how, like, you know, figuring out how to terraform the air better and faster against the enemy.”Kraig, 3:42 pm

Claude: Claude proposed terraforming as a tree of its own, worked backwards: three tracks, machines, life and orbit, all pushing one planet meter, with samples from the swarm as research.

• Planned: one headline percentage, how breathable the air is, fed by smaller numbers such as oxygen, toxins and temperature. Breathable air wins the game.
• Planned: three tracks push it. Machines: air processors, megastructures of catalyst stacks, carbon filters, turbines and GPU control, scrub out the toxic gas the swarm breathes and add oxygen. Life: seeders plant Earth crops and hardy native cover across the surface, with soil and microbes cultured in a lab. Orbit: each package of the seed kit brought down speeds a track.
• Planned: samples the swarm drops speed up the research that turns the air against them.
• Planned: the sky and the colours shift as the number climbs.

Oct 14, 3:47 pmdesignmechanicplanned

Compute: ice caves, memory banks and numbered training runs (planned, not built yet)

“I think instead of just cooled for free in a cave, I think there should be ice pockets of caves. Like certain areas of caves are freezing. And in those sections, you can build data centers and not have to cool them. I think storage for compute should hit a max. Like the Ark can hold a certain amount of compute. But you have to build modules that are like big hard drives. So data centers create compute and then the hard drive modules hold the compute... And you can increase the size of them by upgrading them or building more.”Kraig, 3:47 pm

“Yes, the training runs can be numbered like that, that's fine.”Kraig, 3:49 pm

Claude: Claude had suggested that data centres in caves stay cool for free, and that ARK's upgrades be numbered like real model versions. For the units, Claude picked real ones that grow with the game.

• Planned: some caves are freezing ice caves. A data centre built there needs no cooling; anywhere else, other caves included, it does.
• Planned: the Ark holds only a little compute. Memory bank modules, big drives, add room, grown by levels or by building more.
• Planned: real units that climb with the game: data centres make petaflops, memory banks hold exaflops of work, and the last projects cost zettaflops and yottaflops.
• Planned: ARK's upgrades are numbered training runs, ARK 1, ARK 2, ARK 3 and on. Each is a big compute project that makes every recipe's compute cheaper, lets ARK run more drones at once and adds smarter automation. They are the game's spine.

Oct 14, 3:49 pmdesignartplanned

Water only from the ice caves, and a barren surface that greens (planned, partly built)

“But also with the ice caves, I feel like that should be the only place where we find ice to make water. I don't think there should be water on the outside. Rethinking the surface, I think it should be abundant with light, but I think it should be kind of like, not quite moon, not quite desert, but like kind of barren a little bit. I don't want like flora and trees and grass everywhere because that'll help the terraform change... as our terraform percent goes up, like the surface would change and become greener with grass and like, you know, indicators that it's being terraformed.”Kraig, 3:49 pm

Claude: This replaces the rich, colourful surface of the earlier notes. Stage 4 already built the bare version: a bright, pale surface with mineral colours and no ice.

• Planned: ice exists only in the freezing ice caves, so every drop of water, for the colonists, for cooling, for the oxygen in rocket fuel and for the crops, comes from underground. Lines from the ice caves are lifelines, and the caves matter to the end.
• The surface is bright and barren, somewhere between the moon and a desert, its colour from rock, minerals and sky rather than plants.
• Planned: terraforming shows on the land: as the number climbs, grass, then plants, then more spread over it.

Oct 14, 3:58 pmdesignplanned

Six places, each with a reason to go (planned, not built yet)

“Both major biomes is inside and outside, and then each will have their sub biomes, maybe two or three, and each of those will have its own reason for us to go there, fight enemies, extract resources, or get something or do something. So ice in caves, I think, is great and forces us to go back into caves. Yeah, there should be a reason to go to every part of the map and interact with every aspect of it.”Kraig, 3:54 pm

“Yes”Kraig, 3:58 pm

Claude: Claude proposed three sub-biomes on each side, each one holding something the run needs.

• Planned: two biomes, the caves and the surface, each with three sub-biomes, and each sub-biome holds something you need: a resource, a find or an enemy to beat.
• Planned, in the caves: the Glow caves (the glowing plants: light, methane, medicine and native food, and cave creatures), the Ice caves (all the water, and free cooling for data centres) and the Crystal deeps (crystal for circuits, emitters and GPUs, buried wrecks with data cores, and the worst cave creatures).
• Planned, on the surface: the Flats (sun, common ore, room to build and launch sites, where the swarm hunts), the Badlands (colourful mineral canyons with a rare metal for heat shields and rocket engines, and natural chokepoints) and the Hive lands (the swarm's own ground: samples for terraforming research, and guarded finds).
• Planned: the surface's own life is mostly the swarm and its hives, plus hardy crusts and lichens.

Oct 14, 5:26 pmdesignenemiestutorialplanned

A first brush with the enemy, in Chapter One (planned, not built yet)

“I am considering like the smallest version of like introduction to the enemy within the cave system for the tutorial. Like a fight that you know we can always win. Cause like the Ark can defend itself with a laser. And if just like one or two pop out after you do the copper, like I feel like you're mining for one of the resources and then as soon as you unlock it, like there's some hiding with it and they come out straight to the Ark. And something like that might be interesting. If not, like another little swarm of enemies right after we break the seal and go on the outside.”Kraig, 5:24 pm

“Yes”Kraig, 5:26 pm

Claude: Claude suggested doing both, in different ways: a real fight in the cave, and only a glimpse outside.

• Planned: when the drones break into the copper cave, two cave creatures nesting there burst out and run straight at the Ark. Its laser cuts them down, so you always win, but each shot costs a little power, which stings in the famine. ARK logs it: "Movement in the copper cave. Two contacts. Defence online."
• Planned: after the breakout, a few swarm scouts on a ridge watch the Ark, then slip away. No fight, just the hook into Chapter Two.
• Planned: both go into a polish round on Chapter One, with Kraig's notes from playing it.

Oct 14, 5:29 pmdesignstoryplanned

Chapter Two, Sprout: the plan (planned, not built yet)

“And yes, you can write chapter two plan...”Kraig, 5:18 pm

“Yes”Kraig, 5:29 pm

Claude: Claude drafted Chapter Two in Chapter One's shape: a thin version, about 20 minutes, in four playable stages with Kraig's OK between them. The goal: wake the first three colonists and keep them alive while the swarm finds you.

• Planned, stage 1, sun and mind: the chapter opens on the surface after the Chapter One card, with CHAPTER TWO · SPROUT. Real days and nights with sun power: the famine's relief, but the nights bite. The AI core repair comes back and makes a small trickle of compute, and training run ARK 1 raises the drone cap from three to six and brings drone upgrades back.
• Planned, stage 2, water and the first sleeper: water comes only from the ice cave back inside the mountain, an ice miner there and a line to the Ark, the first lifeline. The still turns ice into water, the oxygen plant water into air, and algae vats water and light into food. Life support is one gauge with death odds.
• Planned: colonists wake one at a time, for compute, three in this chapter, each a specialist with a bonus that stacks: the engineer (building and repairs), the botanist (food) and the medic (lower death odds). They can die; only losing the Ark ends the run.
• Planned, stage 3, parts and factories: the Assembler makes frames, wiring and motors on its own while the inputs last, and upgrading it beats building a copy. New drones, life modules and towers cost those parts.
• Planned, stage 4, the swarm and the lander: scouts find you, and the first nest guards the convoy's crashed cargo lander. Laser towers, laser-fitted drones, slow self-repair and a costly Repair all. Clear the nest and salvage the lander: circuits, and a data core whose fragment points at the seed kit in orbit. CHAPTER TWO COMPLETE · SPROUT.
• Planned: built after the polish round on Chapter One, one stage at a time.

Oct 14, 5:52 pmdesignplanned

Chapter Three, Roots: the plan (planned, not built yet)

“What's next?”Kraig, 5:42 pm

“Yes, draft it.”Kraig, 5:43 pm

“Yes”Kraig, 5:52 pm

Claude: Claude laid out the order (Stage 4, then a polish round on Chapter One, then Chapter Two) and offered to draft Chapter Three in the same shape. The goal: grow the network outward, find crystal and make real compute.

• Planned, stage 1, lines and outposts: one network from the Ark, lines carrying power and goods, and relays to reach further. An outpost with no line runs on its own power, and its goods count only once it's linked; expedition parties pack up full and carry their load home. The swarm starts going for lines, and a cut line drops the power past the cut.
• Planned, stage 2, the Beacon and the deeps: each ping costs power and compute, finds one thing, and can wake what's nearby. It finds the Crystal deeps: an expedition in, cave creatures guarding the crystal, and the Assembler making circuits from wiring and crystal. Wake the geologist for rarer finds and better pings.
• Planned, stage 3, data centres and memory: a melter turns ore and ice into alloy and water, heat sinks are copper and alloy, and the first data centre, built from frames, circuits and heat sinks, needs cooling water, or none in an ice cave. Memory banks store the compute.
• Planned, stage 4, ARK 2 and the lander's core: training run ARK 2 raises the drone cap from six to ten and adds auto-build and auto-repair. Decrypting the lander's data core with compute gives the convoy's map: the rare metal in the Badlands and the site planned for a permanent city. CHAPTER THREE COMPLETE · ROOTS.

Oct 14, 5:57 pmdesignenemiesplanned

Chapter Four, Trunk: the plan (planned, not built yet)

“Yes”Kraig, 5:54 pm

“Yes”Kraig, 5:57 pm

Claude: With Chapter Three approved, Claude offered to draft Chapter Four next, and Kraig said yes to the draft and then to the plan. The goal: found the permanent city and hold it.

• Planned, stage 1, the city site: follow the convoy's map to where the Flats meet the Badlands, with room, cliffs and natural chokepoints. Detach modules there to found the permanent city while the Ark stays free to move, raise concrete walls, and bring in titanium, the Badlands' rare metal, by line or expedition.
• Planned, stage 2, the Heavy assembler: drives, armour plate, emitters and GPUs; a glass furnace and optical fibre; a second-generation data centre for a big jump in compute. Wake the pilot.
• Planned, stage 3, defence grows up: a hangar sends attack drones out on sorties, drones that survive earn veteran ranks, and laser towers and armoured walls get stronger.
• Planned, stage 4, ARK 3 and the siege: training run ARK 3, then the swarm's first siege on the city, with bigger waves and one large brute. Hold it, and ARK 3 finishes decrypting the lander's core: the blueprint for the uplink, the hook into Chapter Five. CHAPTER FOUR COMPLETE · TRUNK.

Oct 14, 5:57 pmdesignplanned

Chapter Five, Canopy: the plan (planned, not built yet)

“Yes build all the way to 6”Kraig, 6:27 pm

Claude: Right after Kraig approved Chapter Four, Claude drafted Chapter Five in the same shape. The goal: make contact with orbit and start turning the planet green. Kraig approved it with the Chapter Six plan.

• Planned, stage 1, the uplink: build the communications array from the blueprint, a big compute project. Contact: the seed kit is in orbit, intact and waiting, and the uplink shows what became of the convoy. Two roads open: terraform from the ground alone, slow and costly, or bring the kit down in Chapter Six.
• Planned, stage 2, life: harvest the Glow caves' plants for methane and extracts, grow crops from the Ark's seed bank in a greenhouse, make rations in a kitchen and medicine in a lab. Wake the biologist.
• Planned, stage 3, machines: the first air processor, and the planet meter appears, "Air: N% breathable", fed by oxygen, toxins and temperature. Samples from swarm kills and hive raids speed the research. Training run ARK 4 makes everything cheaper and faster.
• Planned, stage 4, seeders and the swarm's answer: seeders plant cover outward from the city and the land visibly greens, while the swarm fights harder as the number climbs and goes for the air processors. Reach 25% breathable. ARK's conclusion: from the ground alone this takes generations, and the kit in orbit would finish it. CHAPTER FIVE COMPLETE · CANOPY.

Oct 14, 6:07 pmdesignart

A design upgrade: the Lantern look, in four phases

“I also want to do a general design upgrade. Let's plan out how we can make it look better as well as better animations and design style.”Kraig, 5:44 pm

“The design changes sound fantastic. Let's do them.”Kraig, 6:07 pm

Claude: Claude audited every screen (what already looked good and the twelve biggest problems), then offered three directions: Lantern (dark caves lit only by the flora and the Ark's lamps, pale ceramic machines, a bone-white surface under an amber sun), Porcelain (everything light and airy) and Signal (the world as ARK's sensors see it). It recommended Lantern: the opening Kraig already liked, carried through the whole game.

• Lantern chosen, with Claude's defaults: modules at about half their size as one family of seed pods, the Ark drawn as a seed and power lines as roots; on phones the title and a compact screen, while play stays laptop-first.
• Four phases: 1, dark cave, bright machines; 2, the pod family and its unfold animations; 3, a readable screen; 4, alive caves and the surface.
• The rules: colour only where it means something, one accent colour per object, nothing on the ground that looks like interface, no pure black, and white only for lamps.

Oct 14, 6:27 pmdesignstoryplannedmilestone

Chapter Six, Reaching, and "build all the way to 6" (planned, not built yet)

“Yes build all the way to 6”Kraig, 6:27 pm

Claude: Claude drafted the finale in the same shape as the other chapters and proposed the build order to the end: the design upgrade first, so every chapter is built on the new look, then Chapters Two to Six back to back, each tested and put on the play link as it lands.

• Planned, stage 1, fuel and the launch site: rocket fuel from the flora's methane and oxygen from ice-cave water, rocket engines, heat shields fired from rock, fuel tanks and a launch pad at the city.
• Planned, stage 2, first launches: each rocket reaches the seed kit in orbit and brings one package down, each package speeds the terraforming, and the swarm hits launch sites hardest. Training run ARK 5.
• Planned, stage 3, the elevator: carbon fibre spun from methane for the cable, segments from the Heavy assembler, and the climb as the last great project.
• Planned, stage 4, the last stand and the win: the swarm's final assault as the air nears breathable, the elevator finishes, the whole kit comes down, "Air: 100% breathable", the land green to the horizon and colonists stepping outside without suits. The final card: "REACHING · The seed has grown."
• The order to the end: design phases 1 and 3, then 2 and 4, then Chapters Two to Six back to back. Kraig plays and redirects any time.

Oct 14, 6:39 pmdesignmusic

Music that follows the bloom

“Let's improve the music as well. Music should change based on phase or bloom.”Kraig, 6:37 pm

“Yes”Kraig, 6:39 pm

Claude: Claude proposed keeping the music generated live in code, driven by the flora's cycle, with cues for the big moments and a short theme of its own.

• Planned: rest sparse and dark, rising opening up, bloom the fullest, fading thinning downward; caves close and glassy, the surface wide and airy; a red grid thins the music.
• Planned: cues for the opening, the first fight, the bore, the breakthrough and the chapter card, a four-to-five-note HollowARK theme, and separate switches for music and sound.

Oct 14, 6:45 pmdesigninterface

Menus and settings, overhauled

“The intro, the menu for the game should get an overhaul. Like settings, keyboard bindings.”Kraig, 6:42 pm

“Yes”Kraig, 6:44 pm

Claude: Claude bundled it into one round: a living title screen, a pause menu, full settings, rebindable keys and save slots. Next after that: a feedback button, a sound-effects pass and polish for sharing links.

• Planned: a title screen with Continue, New game, Chapters, Settings, How to play and Credits; a pause menu on Esc.
• Planned: settings for sound, graphics, accessibility (text size, reduced motion, a softer breakout flash) and gameplay (difficulty, tutorial hints); rebindable keys; three save slots with autosave, export and import.

Oct 14, 6:53 pmdesignbalancetutorial

A tiny, cramped start, and a faster Chapter One

“The power cells should move closer because yeah, it drags a little slow. And just the times four speed boost is the only thing that makes that playable.”Kraig, 6:50 pm

“I expected the space that we start in to be even smaller, like really cramped and tight and just like the smallest little bioluminescent peeks through a crack and lights up the area. But what I saw was a big cavern with lots of space with like four or five blooms around. Digging your way out should be like a fun beginning, like reveal the map type of thing.”Kraig, 6:53 pm

Claude: Claude proposed moving the power cells right after the first smelting, so the red start lasts minutes, not 18, and a pacing pass aimed at about 25 minutes at normal speed, down from 38. The tiny start folded into the same pass.

• Planned: a starting pocket barely bigger than the Ark, lit by one small patch of flora glowing through a crack; everything else is rock to dig, so the first steps reveal the map.
• Planned: softer rock near the start, quicker early mining and repairs, cheaper early costs and less waiting.

Oct 14, 6:55 pmdesignart

A time-lapse for every chapter

“Yes, that works”Kraig, 6:55 pm

Claude: One of Claude's ideas for the design upgrade: every chapter-complete card can play a short sped-up overhead replay of the tunnels spreading and the base growing, set to the theme, made to be shared as a short video.

• Planned: light snapshots recorded during play, replayed in 20 to 30 seconds from the chapter card.

Oct 14, 6:57 pmdesign

Hold the launch until the whole game is played through

“I agree with the holding off till I play it all the way through and give my notes to do a launch. And also, like, that'll be a big part of what we develop, like the marketing, the initial batch of people we try to get to play and give feedback. But at first, we just want, like, as many people playing as possible and giving good feedback so that we can iterate on the process of making games in general.”Kraig, 6:56 pm

Claude: Claude suggested keeping these pages unlisted until the whole thin game can be won and Kraig has played it start to finish. Lessons from making this game are being kept for the next one.

• The pages stay quiet until the thin game is winnable and Kraig's playthrough notes are in.
• Then the launch becomes its own piece of work: getting the first players in and hearing back from them. The first goal is as many players and as much good feedback as possible, to improve the game and the way games get made.

Oct 14, 7:33 pmdesignartstory

An abstract title: the story as a blueprint

“The title screen is a spoiler for the opening. Not good. Let's make it more abstract for the menu.”Kraig, 7:32 pm

“What about a super minimal abstract blueprint style of the story leading up to the crash. Mother ship sending out seeds on an ancient ship. Light speed seed and drop out to see a planet. Close the animation on the planet.”Kraig, 7:33 pm

Claude: Kraig named the feel he wanted: ancient schematics. Claude's version: thin teal lines on dark, telling the backstory in a loop and never showing the crash or the cave.

• Planned: the mother ship releases its seed ships; one ancient seed ship; the jump to light speed; dropping out at a planet; closing on the planet as a slowly turning wireframe globe.