At the methodological level, Feyerabend defended genealogy as neither subversive nor vindicatory but heuristic, facilitating conceptual creation in the sciences and in philosophy. He found useful examples of what he called the “historical approach” in Ernst Mach and in Niels Bohr. At the descriptive level, he argued for a return to the historical approach, and found examples of it in the rising discipline of STS (in this regard he cites Bruno Latour, Peter Galison, and Andrew Pickering approvingly). Contrary to a popular misunderstanding, his own genealogical studies are not debunking, except of simplistic a-historical models of what was happening during key periods of scientific change. Or rather, even when his descriptive genealogies are debunking they are also vindicatory, as they show how certain methodologically proscribed elements usefully contributed to conceptual and empirical progress.
Cross-posted as a comment on this post discussing the genealogical method.