Grand Prix

Grand Prix: Artist's Statement

Paul Ricard is one of my favorite Formula 1 tracks, so when I settled on a racing game, I wanted to race there. As a kid, I wanted to go watch the race there, and I've loved the track ever since. It's fast and wide, and it's fun to watch when they race. I never got to go, so in a way this game is how I finally get to that race: not in the grandstand, but on the grid. Grand Prix is a top-down racing game with two modes. You can run a three- or five-lap race against five rival cars, or a time trial against a ghost of your own best lap. The circuit follows Paul Ricard's shape: the twisty infield, the long back straight with its chicane, the flat-out right at the end, the long double right, and the tight run back to the pits.

The first decision I'd defend is making it about pure driving skill. I considered tire wear with pit stops and a limited overtaking boost, and I turned both down. I wanted getting faster to come only from driving better. So the car has a grip limit. If you turn too hard at speed, it slides wide and loses time. That means braking late and carrying speed out of corners is the whole game. The second decision was changing the track. The first version was too small and narrow, with no room to pick your own line or pass side by side. I asked for it to be rebuilt in the shape of Paul Ricard and made wider. It came back about 40% wider, with bigger run-offs, and that turned the race into a choice of lines and overtakes instead of just following the car in front. The third was including both modes. The race gives you rivals to beat. The time trial gives you only yourself, and your ghost shows exactly where you lost time.

We built it together, and the collaboration was real in the details. I didn't ask for Paul Ricard's blue-and-red striped run-offs, where the red bands slow you down more than the blue. I also didn't ask for the five red start lights with a reaction time and a jump-start penalty, or the purple and green sector times from real F1 timing screens. All of these were better than what I described. It wasn't always real. I first made an MMA game but didn't like it, so I switched to F1, and I threw away the first track entirely once I saw it was too small. Claude also renamed my track "Circuit Mistral" so the game wouldn't use the real circuit's brand. That was a decision made for me rather than by me.

Did I author this game? Less of it than I'd like. I chose the genre, the track, the width, the two modes, and the pure-skill direction, and I decided when it was good enough. Claude wrote every line of code and picked most of what you see on screen. I'd call myself the person who knew what it should be, not the person who made it.

Prompts and choices that shaped the game:

• “i was expecting it a lot better to be honest, what if we do an f1 game, could it be easier?” • “lets build an f1 game”

• When Claude asked me to choose the kind of race and the main tension: “Both” (a race against

AI cars and a time trial against my own ghost) and “Pure driving skill” (instead of tire wear or an overtaking boost)

• “make the track similar to paul ricard (french f1 track) and make the track wider” • “what else, i like the game, nothing to change” (when I decided the game was finished)

AI disclosure. I used Claude (Opus 5.5) to brainstorm, to build the game over two rounds (the first version, then a rebuild of the track), and to draft this statement, which I then edited. I also used Claude Code while working on the game. I first made an MMA game but abandoned it, and I had the track rebuilt because the first version was too small.