The Hidden Algorithm: How I Solve a Crowd Painting

I called it instinct. I called it "letting the painting evolve." Both true — but recently I described the process to someone outside painting entirely, and they used a term from computer science: constraint satisfaction. Once I heard it, I couldn't unsee it in my own work.

A constraint satisfaction problem, in the simplest terms, is what you get when you have a set of individual elements that each need to meet certain rules, and those elements together need to satisfy a bigger rule as a group. Sudoku is the classic example — every number has to be valid in its own square, its row, and its column, all at once. Nobody solves Sudoku by filling in squares at random and hoping. You're constantly checking each choice against everything around it.

The individual figure. Every person in a piece like Breakout or Celebrating Tradition has to work on its own terms — a distinct pose, a believable gesture, a colour that reads as clothing rather than noise. That's the local constraint. Each figure is its own small decision.

Breakout painting

Celebrating Tradition painting

The whole picture. But no figure exists in isolation. As I add each one, I'm unconsciously checking it against everything already on the canvas. Does this colour clash with its neighbour in a way that reads as a mistake rather than a choice? Is this corner of the canvas getting too dense while that one goes quiet? Is everyone facing the same way, which would flatten the energy? Every new figure has to satisfy the figure next to it and the painting as a whole.

That's why the white or near-empty grounds I favour aren't really "backgrounds" — they're closer to scratch paper. Stripping the setting away lets me actually see whether the constraints are holding. If the crowd is doing its job, it should feel balanced and alive even with nothing else on the canvas to distract from it.

Colour turns out to be doing something similar. I rarely let two adjacent figures share a dominant hue — not as a written rule, but I feel the wrongness of it immediately when it happens, the way you'd feel it if two touching countries were shaded the same colour on a map. That's a graph-colouring constraint, essentially: neighbouring elements can't match. The result is forced variety that still resolves into harmony rather than chaos, because the constraint is doing the organising work I don't have to do consciously.

Interestingly, the landscapes solve the opposite problem. A piece like Echo Valley isn't asking "how much can this canvas hold while staying balanced" — it's asking "how little can populate this space and still feel inhabited." Same instinct, run in reverse. Minimum satisfying assignment instead of maximum.

None of this was ever a formula I sat down and wrote out. But looking back, I think that's the more interesting discovery — that a mind can build and enforce a fairly rigorous rule-set entirely by feel, checking dozens of small decisions against each other in real time, without ever naming the process.