Martin & Rose (2007: 156):
In order to make sense of discourse, one thing we need is to be able to keep track of who or what is being talked about at any point. When we first start talking about somebody or something, we may name them, but then we often just identify them as she, he or it. By this means our listener/reader can keep track of exactly which person or thing we are talking about, i.e. which participant in the discourse. There are many other ways of introducing participants into a discourse, and keeping track of them as we go, that we will explore in this chapter.
To be clear, Martin's system of identification confuses the instantiation of experiential potential by a speaker (talking about persons and things), with the comprehension strategies of a listener (making sense of discourse, keeping track of what is being said), and presents the confusion as the instantiation of a speaker's textual potential.
That is, it is the speaker who introduces participants, but it is the listener who has to keep track of them. Systemic potential is a model of what speakers can mean, not a model of how listeners understand.
Moreover, participants are construals of experience (experiential meanings), not textual statuses assigned to experiential or interpersonal meanings (textual meanings). Introducing participants is introducing experiential meaning; tracking participants is tracking experiential meaning.
As will be seen in later posts, this metafunctional misunderstanding derives from confusing the textual meaning of referring with the experiential meaning of the referent.