PERSON
Johannes Kepler
The seventeenth-century astronomer who invented model-fitting by hand—spending a decade forcing the orbit of Mars to confess its ellipse—and whose double allegiance to beauty and to evidence is the discipline the age of pattern-finding machines most badly needs.
Four centuries before the first training run, Johannes Kepler performed the foundational act of machine learning alone, by candlelight, with quill and logarithm-free arithmetic. His decade-long war on the orbit of Mars—iterating model against data, measuring error, adjusting parameters, refusing to round away a discrepancy the data was good enough to resolve—is the ancestor of every
gradient descent that has ever converged. He found
three laws of planetary motion that reorganized the cosmos, published every dead end alongside the triumph, and still could not
surrender the most beautiful false pattern of his life—the
Platonic-solids model of the solar system—even after the data had conclusively refuted it. That double portrait—ruthless empiricist and incorrigible mystic, the man who killed the circle and could not kill the harmony—is exactly what a field drunk on
curve fitting and seduced by elegant architectures most needs to meet. The
[YOU] on AI cycle reads Kepler not as a distant ancestor but as