The distinction I'm alluding to is simply "the map is not the territory." In other words, in the same way that you can have something which represents a tiger without actually being a tiger (eg. the word "tiger") you can have something which represents "the boldness of this text" without literally being the boldness of that text. The double asterisk notation in Markdown, and the <b> tag in HTML, are symbols for the boldness of the text they annotate.

What I'm saying is that untrained users have trouble with the idea, fundamental to markup languages, that some text can represent attributes of other text. To such users, an asterisk means "the text has an asterisk in it." They cannot reinterpret asterisks as markup without relearning the way they look at text itself. They're used to having the territory laid out in front of them, and all of a sudden you're handing them one map and asking them to write another, but they don't know what a map is.

The distinction is exactly that between

    <p><b>some text</b></p>

and

    <style type="text/css">
      .emphasized { font-weight: bold; }
    </style>

    <!-- [...] -->

    <p class="emphasized">some text</p>

One specifies "this text shall be bold"; the other specifies "the semantics of this text require that it be emphasized", with precisely what 'emphasized' means handled elsewhere, in the CSS which defines how the class 'emphasized' shall be rendered -- which, given appropriate media queries, might appear entirely differently depending on whether the document is rendered to a large screen, a small screen, to print, &c., &c.