The former ubiquity of Perl made most of these common knowledge. For example, $? works in both Perl and bash (actually, I don't know which did this first...). So in the context of shell scripts, where Ruby started, having these magic globals makes a lot of sense.

IMO they can make code easier to understand, once you've learned what they mean. $2 conveys the same idea as matches[2] without having to think about array indices, and why the matches aren't zero-indexed like a normal array. This is obviously totally subjective.

Especially when Ruby has no way of scoping these variables... unlike Perl:

  {
      # level 1
      local $\ = "[1]\n";
      some_print_operation();
  
      {
          # level 2
          local $\ = "[2]\n";
          some_print_operation();
      }

      # level 1 again
      some_print_operation();
  }

  # $\ back to what it was before all this :)

True, but I have seen -- $:.unshift File.dirname(__FILE__) -- used quite a bit in the wild, especially in MacRuby projects.

It's also rare to see them heavily used in Perl nowadays. Lots of recent code doesn't use them at all (e.g., with autodie, no need for $!)

If you write your shell scripts in Ruby they do come in handy. I certainly wouldn't use them outside of that context.

I use $1, $2, $3, etc. quite often. $' and $` are also useful for throwaway scripts.