Tags are not hierarchical categories
I disagree with Mat's Mug where he says:
if minesweeper is a "subset" of game, then tagging with both game and minesweeper would be redundant - using the most specific tag correctly labels the post, there's no need to specify both
The purpose of tags is not hierarchical categorization of knowledge, like categories in Wikipedia. The purpose of tags is to make it possible for people who are experts (or just interested) in a subject area to find posts in that area (for example, in order to answer them).
Suppose that you are an expert in games and like to answer game-related questions. If so, then at present you can visit the unanswered/tagged/game page from time to time and you're likely to find many unanswered questions about games. But now suppose that enthusiastic taggers start removing the game tag from game-related questions that have some more specific tag. Now you have to visit the unanswered/tagged/game+chess+minesweeper+sudoku+tic-tac-toe page, and who knows what games you are missing because you haven't itemized their more specific tags in your search? This situation does not seem like an improvement to me.
Paradigm tags can be meaningful
I'm also not convinced that paradigm tags are meaningless. You can program Haskell in a functional style, with lots of higher order functions and function applications. But you can also program it in imperative style, with lots of monads and
let a = b in c (see for example this in-place quicksort question).
So paradigm tags make sense when the paradigm is unusual or something that the OP wants to draw attention to.