What do you not understand? The stackoverflow post you mentioned is as clear as possible. Let's say you come up with your own, very simple language, call it Tiny ML for a tiny metalanguage (they don’t actually do this, this name belongs to ML).
Then you program the interpreter, say, in C, for this tiny language. When this interpreter is up and running, you take this interpreter, which can run Tiny ML-code and write another interpreter in Tiny ML, which can handle a much more complex version of Tiny ML, let me call this language Hyper ML. There you have it, your last interpreter should be an illusory metacircular interpreter.
source
share