The fact that categories are often fuzzy does not make them useless as a general heuristic for finding people who are more likely to share similar experiences with each other.
Almost every kind of categorization system shares the problems you talk about: people can call themselves "programmers" for a variety of different reasons, even though they're working in wildly different environments and might be totally unsuited for each other's jobs. People self-identify with political parties, even though there are no strict rules about what a "Democrat" or "Republican" are, to the point where beliefs of members of those parties might overlap with each other or even be the opposite of what the party normally espouses. We separate games/movies into genres, even though genres are often a complete mess that don't always map well to player preferences or experiences. We take a light spectrum and we break it up into distinct colors, even though the number of colors and lines between those colors are completely arbitrary and meaningless.
It's good to recognize that categories are fuzzy human inventions that are prone to abuse, and it's good to be cautious about putting a lot of weight on them in every situation. It's good not to assume that everyone in a category is identical. But abandoning categories entirely would make it much harder (even impossible) to go through normal day-to-day life.
So it is similarly useful to be able to think about categories like "Black", "Gay", "Conservative", "Christian", "Rural", "Trans" -- even as we acknowledge that those categories are just human inventions and that they blur around the edges. Those categories are still be a useful heuristic for trying to find people who are more likely to share certain attributes or who have had certain life experiences.