It’s sometimes useful for emerging fields where there’s no proper research. For example bodybuilders in the 80s got a ton of stuff right. And it took the research 30 years to catch up. But when good research exists I like to use it.