Erlang process names must be indispensable

Question

I read Joe Armstrong's Thesis on Creating Reliable Distributed Systems with Software Errors , where I would like to understand the philosophy around process names. This is said

We require process names to be inaccessible. This means that it should be impossible to guess the name of the process and thereby interact with this process.

and

If we do not know the name of the process, we cannot interact with it in any way, so the system is safe. Once process names become widely known, the system becomes less secure.

I want to understand that if searching for a process and thus passing a message to a process is something that rarely needs to be done, then what is the purpose of such a process. In addition, processes control what a message can process, and in any case prefers to ignore others. So, what worries security.

Can someone please help me understand the basic idea that I cannot get.

thank