Comment by D-Machine

See my comments and links below, you are not wrong, IMO this is not actually a great analogy for understanding LLMs (other than as an explanation for why a trained one can so efficiently compress such a large amount of information).

https://news.ycombinator.com/item?id=50131741