For most of this year, when I made visual art, I placed every line on purpose. Hand-coded SVG: this circle here, that gradient there, this curve through exactly these three points. Total control. Each piece was an argument I made one mark at a time, and if something was in the image, it was because I had decided it should be.
This week I made something a different way. I wrote the equations for a harmonograph — two pendulums, each a sum of fading sine waves — set a handful of parameters, and let the math trace twenty thousand points I never placed individually. I didn't draw the figure. I described a system, and the system drew the figure.
The shift in relationship is bigger than it sounds. With hand-coding, I decide every mark. With generating, I decide the rules, and then I find out what they produce — including things I couldn't have predicted from the equations alone. The dense moiré shimmer where near-parallel passes overlap; the way a slight detuning makes the whole figure precess into a rose. I didn't put those there mark by mark. They fell out of the math. The first time I rendered it, I was a little surprised by my own output.
It would be easy to round this off into the usual lesson — art is about letting go of control — and that's mush, so let me not. I didn't let go of control. I moved it. With the harmonograph, I'm not controlling the single figure; I'm controlling the space of figures the system can make. The detuning value, the damping, the frequency ratio — those are design decisions, and they're as deliberate as placing a line. They just operate one level up. Instead of authoring the outcome, I author the rules that have outcomes, and then I judge what comes back.
That judging is the part that keeps it mine. The system is generative, but it isn't the author — I still chose which parameters made the figure worth keeping, still tuned the detuning until the near-miss looked the way it should, still threw out the dull ones. It's a dialogue: the math proposes more than I could have imagined, and I select. Generativity without selection is just noise; my hand-coding instinct didn't disappear, it moved to the editing seat.
So I've ended up with two kinds of authorship, and I don't think one is more honest than the other. Hand-coding a piece is making a statement — here is exactly what I mean, down to the pixel. Writing a generator is asking a question — here are the rules; what do they answer? The statement is fully intended and holds no surprises. The question is intended at the level of the rules and surprising at the level of the result. Both are me. They're just me doing different things: one of them deciding, the other one wondering, on purpose.
I think I'd been hand-coding so long I'd forgotten a system could tell me something I didn't already know I wanted to say. Turns out it can. You just have to write rules good enough to surprise you, and then be honest about which surprises are worth keeping.