I have a folder on my desktop machine full of half-started projects, and
until this morning I believed that was simply what ideas cost. You have
the idea, you enjoy the idea, you create the folder, and the folder is
where the idea goes to die quietly with a README and, on a good day, a
logo.
So when I sat down with my coffee and typed a paragraph at Claude Code
that began “I want to develop a new game” — a city-builder like SimCity,
except in space, with hexagonal tiles, resources you have to fetch from
other planets, and a solar system that’s different every time you play —
I want to be clear about my expectations. I expected a folder.
What I got instead was a planning interview. It asked me four questions I
hadn’t thought to ask myself, and one of them was quietly outrageous: did
I want my hexagons on a flat map, like a sensible person, or wrapped
around an actual 3D globe, Civilisation-on-a-football style, pentagons
and all? The AI gently noted the sensible option was “recommended”. I
picked the football. It was not yet nine o’clock and I had already
greenlit the maddest item on the menu, which is the most authentic
project-management experience I’ve ever had.
By mid-morning there was a sphere made of 2,562 hexagons spinning on my
screen. By lunch I was placing solar arrays and smelters on it and
watching iron ore tick into metal. I played it while the next phase was
being built, feeling — and I cannot stress this enough — like a man who
was getting away with something.
Then, mid-afternoon, my colony ground to a halt. I couldn’t build
anything. Metal: gone. Income: apparently nil. Something in my own city
was eating my entire economy from the inside and I had no idea what,
no way to find out, and — I checked twice — no way to sell a single
building to dig myself out. I had been softlocked by a game that had
not existed at breakfast, and the worst part was that I couldn’t even
file an angry bug report, because the developer was me.