The reason that may still be unclear is that a Controller object, by definition, has no directly accessible interface through which the Actor can reach it. Providing that interface is the responsibility of the Boundary object. It's about clear division of responsibility really, which promotes more optimal coding practices and greatly enhances maintainability.
This common problem is modeled in the MVC design pattern (Model, View, Controller, see Design Patterns; Erich Gamma et al). In practical terms for ASP, the design plays out this way:
The user interacts with a form page (ASP/HTML with formatted content and controls), which POSTs to another ASP (this time, only containing the script logic, with no interface), and during that processing it may interact with a data store, and when done it calls another presentation page (ASP/HTML) to format and present the output.
The flow of control in MVC terms would go something like: Actor -> View object (ASP/HTML) -> Controller object (ASP) -> View object (ASP/HTML). The Controller object may/may not access a Model object (data store) during that interchange.
BTW, the MVC terminology is analogous to the ICONIX terminology of Boundary (View), Control (Controller), Entity (Model). If you look at EA's Analysis diagram toolbox, you'll find these objects available for modelling using the ICONIX terminology.