No. I get it. I do understand. Maybe I should have said needlessly complex. Ie you can have keyless entry to a car and start the engine with a press of a button. Or -- instead -- you can search for a key in your pockets (put the bag in your hand on the ground first) insert it into the lock, turn the key and then pick up the bag, take the key and put it into the ignition, turn it and start the car.
So you tell me the first solution is dumb and the other one is smart? I would say that exactly opposite is true.
You are combining two features of the language that don't really have anything to do with each other (references, and core type sigils), and conflating them in a way that's irrelevant, which is why you are getting some pushback on your critique.
Perl uses sigils. You can think of them like inflection in spoken language. They provide hints as to what is being referred to, to help make intent clear.
Perl uses references. In some respect, almost all languages use references, and it's just a matter of whether it's explicit or implicit. Perl makes it explicit, which means the programmer has to understand it early to make some things work (nested data structures). Other languages may have it as implicit, but that doesn't mean the programmer can ignore it, it just means they may be able to ignore it for a short while when first learning the language. Sooner or later they will encounter and have to deal with whether they are working with a copy of data or the original data. If they are lucky, this happens before it's a nasty bug.
So you tell me the first solution is dumb and the other one is smart? I would say that exactly opposite is true.