There is some state that is global. Current user for instance, it is used globally, so why create a complex way of passing it around instead of just making it global?
And then that is not a global state, my point is that there is global state, and trying to go through all kinds of academic exercises in "proper software design" is unneeded complexity. It's a tool, don't use it for the wrong job, but don't dismiss it just because other people use it the wrong way.
One reason is that, if you have to pass that information to functions which use that information, you can easily see which functions rely on this global state, and which don't. The usefulness of this and the annoyance of passing it manually depends on the situation, of course -- read-only "state" like the current user, and write-only state like some global logging interface, is the sort where one function's use of the state doesn't materially affect another function's behavior, so it's more OK access directly than other state. (Another example: the database connection configuration info might be unchanging and read-only, but it's real useful to pass it around manually, because then you see exactly which callees are opening database connections.)
