Design Notes

This Document

Obviously this is the history of this design. But it's real purpose is to demonstrate that this game was written by a real human. With all it's flaws and mistakes, I conceived of it over years; it became my passion and my vision and I relentlessly forced the AI to adhere to my vision rather than wander off on it's own (which it was very much wont to do).

The Auction

I've wanted to fit an auction mechanic in a game for more than a decade. I like auctions in board games but they take up a lot of time and effort. I thought it might be easier to do it in a computer game.

In real life Contracts (and, hence, bidding on contracts and hence AUCTIONS) have been associated with space since the very beginning. Most things in space were very expensive with nothing but prestige (and maybe some science) as a payoff so it's was heavily financed by government. And in America that means a putting a contract out to bid.

The Original Design

The original design, back in 2018, was private rich people going to space. At that moment in time SpaceX was doing amazing things and I thought "this is going places" and "what a great idea for a game".

I wanted players to be able to join and leave at any point in the game (another game mechanic I've wanted to do for years) and that meant I needed new players to be able to compete with established players.

So one player was best at getting material to LEO. Another could move that to the moon. Yet another could land supplies on the moon. Maybe the new player could do mining or prospecting. It was more of a cooperative game with everyone pulling together to "get the job done".

When I did my research I realized that there wasn't really anything ON the moon that would allow an economy to thrive. No heavy metals, no rare earth (would we call it "rare moon"?), no uranium, nothing. We might all cooperate but there was nothing to cooperate ABOUT.

I discovered that, when it comes to the moon, There is no There There (Gertrude Stein said this about Oakland, CA and I don't agree but it's a great quote anyway). So I abandoned it until 2024.

The Dungeon Game

I had some health problems in 2023 (an understatement) and I wound up doing dialysis several times a week. It was boring so I brought my laptop and coded up a dungeon game. It was a single player web game that ran in my browser based on json files.

I created code that would predict, based on where you were, what you could do. So that was the user interface. The game itself could be be regarded as a "finger exercise" but that user interface (but not the json files, ick) was a good idea.

The Second Design

In early 2024 we still weren't mining on the moon but SpaceX was going strong and it had some competitors. It was unclear to me whether any of their activities would pay off (although consumer communications looked promising) but I wasn't writing a simulation, I was designing a game.

I would be relentlessly optimistic and imagine that 4 people in a space station was "humans colonizing space" and that Space Tourism was a real business model. I would make this game.

First Code

Right from the start I had the concept of Tech->Assets->Contracts->Payoffs. I wrote the tech manager with dozens of techs loaded into a json file out of a giant hand-build spreadsheet. And a bunch of assets that could be built from the tech. And and an action manager that controlled a set of possible actions for the UI. And code that told the player what he could do in a simple table of check boxes.

I didn't have access to AI (it wasn't really a "thing") so I wrote all this and a player manager and the game code and the turn code and all the routes and all the initial HTML, etc, etc by hand. Some of that code is still in the game. The first playable was all me, no artificial sweeteners or GMOs (or AIs) were used.

Contracts

I started work on the contracts. My idea was that all the payouts would be contracts. There would be game generated contracts (from the government or from other non-players) that would allow the players to make money. And there would be contracts that would allow players to interact.

Maybe you had better tech than me but I had more money -- would you sell me some of your tech? Or maybe you were great at certain important capabilities (space tugs, reusable rockets, constellations, habitats, mining, etc) but not so great at others. Contracts were a way I could sell you tech and assets in return for other tech and assets.

This is actually pretty complex but pretty much what Date invented relational databases for. But all that json stuff had to go. What a dumb idea that was.

I started coding the Contract Manager and built the Database Manager.

Then I got access to an AI. Suddenly coding was a lot easier. It made short work of the remaining json stuff and hooked up all these different tables so a Contract connects to 8 other tables with all the different things that you could buy or sell with a contract.

I wasn't using the AI properly (it was all new to me) so it was not all that helpful and I read every line of code and still coded a lot of it by hand. This wasn't "vibe coding", it was AI assisted coding.

The Break

I got so I wasn't feeling well and I stopped working on the game for about a year. When I got "home" from work (I work remotely) I was exhausted every day. Low red blood cells or something. So I didn't work on the game. For a whole year. I thought about it a lot, but I didn't DO anything.

I used AI at work and I became a lot better at using it in this period.

I'm BAAAACK

Sometime in early 2026 I felt strong enough get back to the game. So every afternoon, after work, I would work away at it. I was now using a version of spec-driven development, although I didn't know it. I would write a Jira ticket that said exactly what I wanted and then have the AI code it. I started doing this in sprints. I built out the AWS server and got the game sort of running on it.

I used the AI for research and, sometimes, suggestions, but it didn't have access to the server and it only had access to the one Jira project and Confluence space relevant to the game. I didn't like what it wrote on Confluence so it was really read only -- I'd write a design and then it would pull it out of Jira or Confluence and write the code.

I wrote a lot of the initial game and a lot of other code in 2024 and 2025. But in 2026 I became a designer and tester.

As a designer I was responsible for holding the vision and not allowing the AI to change it and then, when it claimed it finished the code, I tested it. Nothing was ever released without testing. Nothing is in the game that wasn't aligned with my vision. The AI did not design any part of this game.

Through the first part of 2026, I worked out the whole game and got the AI to code it up. I keep finding things I want to add (better constellations, reusable rockets, ride sharing etc). But mostly it's done as of this day in May, 2026. Ready for Alpha (OK, "early alpha"; cut me some slack).

Time and again the AI would wander off and do something that wasn't part of the vision and I had to drag it back. For a few weeks parts of the game were real-time instead of turn based, for example.

In my view, AI slop comes from not having a clear idea of what you want or not relentlessly (!) checking that the final product is what you intended. Yes, I've done vibe coding and allowed the AI slop to remain. But not on THIS project.

The Future

There is a lot of testing that goes into a game. It's very likely the players won't like some of my "vision" and I'll have to change it. That's what play testing is for and I'm going to embrace it.

But at no time in the future will I allow the AI to redesign the game. That does not comport with my philosophy of how games should be written (which I came to in the 90s and, as with most old people, I'm not letting go of; "Hey you AI: Get off my lawn!").

This Document

It would be too ironic, even for me, to use the AI to write the document explaining that the AI didn't not design the game. So I wrote every word of this document. Like it? That's me. Hate it? Me again. Ironically that's what you get for depending on a human.

Other public facing documents WERE written by the AI. Almost all of them. I generally tell it what I want and ask for an outline and then ruthlessly edit that outline. And then it writes it and I ruthlessly edit the result.

Then, in all cases (including this document), the AI translates the text into the templates that become HTML in this kind of web app. It doesn't change the text but if you're worried about the sterile fingers of AI getting on the Art, well this game is full of sterile fingerprints. It's just too expensive to remove them.