When we talk about the meaning of words, what exactly do we mean? What is this thing that we refer to as "meaning"? This is something of a paradoxical question. If there was consensus on meaning, then there wouldn't be any need to ask the question. But if there isn't consensus then it is unclear exactly is being asked! There seems to be some kind of infinite regress or circularity.
However, that doesn't mean that we are stuck. We want meaning to mean something and we want the meaning of meaning to be itself. If we write this as a function, we get the follow:
meaning("meaning") = meaning OR meaning(quote(meaning)) = meaning
If it weren't for the quotes, then it would be idempotent. Instead we'll call this quasi-idempotence. Given these requirements, we can't say that meaning is something absurd like a banana, as that wouldn't define a function. Other potential definitions of meaning will be ruled out by quasi-idempotence as we are about to see.
Let's suppose we say meaning is purely descriptive, as opposed to prescriptive. That is, to find the meaning of a word, we should go out into the world and see how people use it in the language games they play and call that the meaning. Well, then we should do the same for the word "meaning". And from what I can gather, in some language games people are in fact just trying to get things done and say the things that need to be said to play a particular language game. And in other language games, there's some kind of centralised or decentralised authority and people are trying to use the word the same as them. This occurs in science or other technical fields. So we said meaning was descriptive, but then when we calculated meaning(meaning) we saw it was sometimes descriptive and sometimes prescriptive. So this definition ended up undermined itself.
On the other hand, let's suppose we said that the meaning of a word was its Platonic Form. It is easy enough to show that this is quasi-idempotent:
meaning("meaning") = form("meaning") = meaning
So, this definition is consistent, but it doesn't mean that it is correct. There's no reason why there can't be multiple theories that are quasi-idempotent.
Another possible theory would be to start off with the frame of meaning as use, head out into the world, see how people actually use meaning, then take that as a new definition of meaning and recurse. Eventually, this might hit a fixed point, at which point we can stop.
So, given that we end up with multiple theories, how do we decide between them? Well, this just comes down to whichever theory best meets our aims. Words were created by humans to meet human needs. They are part of the map, not part of the territory. To understand this better, I would strongly recommend this post on conceptual engineering.