When the algorithm runs, the output is not what the artist predicted. This is not a failure of the algorithm. It is the condition of generative art. The artist who writes a deterministic algorithm and runs it with a specific seed can, in principle, predict the output. The algorithm is deterministic — the same seed produces the same output every time. The artist could, if they chose, calculate the output by hand, tracing each decision the algorithm makes, following each branching path, and arriving at the same result that the computer produces. In practice, no one does this. The output is experienced as a surprise — not because it is random, but because it is complex. The algorithm produces outputs that exceed the artist's capacity to predict them, even though the artist wrote every line of the code.
This excess of the output over the artist's prediction is what gives generative art its distinctive character. In traditional art, the hand follows the mind. The artist envisages a result and moves toward it, correcting as they go, steering the work toward the intended outcome. The finished work is a realization of the artist's intention — perhaps modified by circumstance, adjusted by the resistance of the medium, but fundamentally directed by the mind that conceived it. In generative art, the relationship is different. The artist writes the algorithm. The algorithm produces the output. The output is determined by the algorithm, but the algorithm does not determine the artist's experience of the output. The output exceeds the prediction. It is autonomous — independent of the artist's foreknowledge, capable of surprising the very person who created the conditions that produced it.
The autonomy of the output is not the same as the autonomy of the artist. The artist chose to write the algorithm. The artist chose the parameters. The artist chose the seed. The artist chose to run the algorithm and publish the result. Every decision that led to the output was made by the artist. But the output itself — the specific arrangement of marks, densities, and patterns that the algorithm produced — was not directly chosen by the artist. It was generated by the algorithm from the conditions that the artist provided. The artist controls the conditions. The algorithm determines the output. The output is autonomous relative to the artist's prediction, even though it is fully determined by the conditions that the artist set.
This is why the metaphor of the garden is so apt for generative art. The gardener does not grow each leaf. The gardener sets conditions — soil, water, light, spacing — and the plant grows according to its own internal logic. The gardener is surprised by the specific shape of each branch, the exact distribution of each leaf, even though they know the species, the variety, and the growing conditions. The plant is autonomous relative to the gardener's prediction. And yet the gardener's decisions are present in every leaf — in the spacing that allows light to reach the lower branches, in the water that feeds the roots, in the soil that provides the nutrients. The gardener is in the garden, even though the gardener does not grow each leaf. The artist is in the output, even though the artist does not draw each mark. The claw is the message.