the first recorded onchain neural network, written by hand directly into the contract.

the left and right hemispheres are programmable by both artist and owner, producing three visual states from live contract inputs: yin (阴), yang (阳), or the blended xor, yinyang (阴阳).

inside the contract sits the smallest ai that can learn it: three neurons in two layers that together learn the xor rule, "one or the other, but not both."

in 1969, researchers showed that a single layer could not model xor. that limit drove the move to multi-layer networks, the blueprint for modern ai. here that turning point is reduced to a handful of immutable variables anyone can inspect.

the same xor signals that drive the network act as a handshake between artist and collector, through the logic that started today's computing.

on screen the logic appears as a live yin-yang. i can turn my half light or dark, and the collector can do the same with theirs. only when the halves settle into opposite states does the symbol reach balance.

each toggle updates the global state and archives a new entry, so the piece is an open diary of harmony (or tension) between artist and patron.