For more than half a century, swimming resisted scientific analysis. When Leon Lewillie opened the first International Symposium on Biomechanics and Medicine in Swimming in Brussels in 1970, he identified the obstacle precisely: a barrier of water had isolated our sport from rigorous measurement. This lecture explores how this barrier became permeable. Starting with Lewillie's balloon-borne electromyograms, it follows the development through Jan Pieter Clarys's five decades of towing-tank and muscle work, which demonstrated that technique rather than body form governs resistance, and on to my own nine editions of this symposium, where the focus shifted from developing instruments to understanding the properties of water itself. The second half turns to the force now reshaping the field: artificial intelligence. Computer vision can already segment races and measure the underwater dolphin kick, while wearable sensors can classify strokes and explainable models can rank the variables behind performance and flag injury risk before it occurs. However, these tools can produce inaccurate results, hide their reasoning and have not read much of our literature. The argument is simple yet demanding: machine learning provides scale and speed, while fluid mechanics provides cause. Only their collaboration — with the scientist firmly in the loop — yields understanding. Fifty-six years after Lewillie, the water still resists; the machine is powerful, yet blind; and the researcher must still ask the right questions and stand by the answers.
© Copyright 2026 XVth International Symposium on Biomechanics and Medicine in Swimming Proceedings. All rights reserved.