Symbols are arbitrary, but if the association between the symbol and its meaning is consistent, then a convention can be built using it. But what are conventions good for? Even if A always leads to B, so what?
Well, A=B is not that useful (for now), but A=T.R.E.E. allows me to write the word “tree” with 1/4 the letters. The usefulness of a convention is the ability to substitute something more practical (A) for something less practical. We might call this “compression”, same meaning, fewer letters, less space, less wood, etc. Writing “tree” or “A” is way easier than attaching an oak tree to the page, for many reasons.
Something interesting is that although the tree will never actually be in my mind, the symbolic elements that I am using to construct the representative model, those ARE in my mind, and, in fact, I have no symbols for them. They are the actual thing. “A” is not a representation of an A. I don’t have a symbolic representation for A, but maybe I should. Otherwise, how will I be able to symbolically represent them in my symbolic representation of symbolized representing.
Another way of visualizing this is to imagine making a doll, which is a toy person, out of wood. The doll is not a real person, but the wood is real wood. Now, try building toy wood out of real wood the same way you built the toy person. It would be like trying to build toy people out of real people, and it presents a . . . curious little kink in our symbolic modeling process.
We might ask: can symbolic things and literal things interact to accurately model the world, or does the model need to be entirely symbolic or entirely literal? Symbols are made of literals, but can literals be made of symbols? Isn’t that the purpose of critical modeling, to create something “real” out of symbols?