Re: [Model-t] Definition of an intermediary

[Martin Thomson <[e-mail]>]

On Fri, Dec 24, 2021, at 03:20, Carsten Bormann wrote:
> On 2021-12-23, at 15:57, Vittorio Bertola 
> <vittorio.bertola=[e-mail]> wrote:
>> 
>> between the fingers of the sending human being and the eyes of the receiving human being (or of the single human being involved, in human-to-server communication) is an intermediary. 
>
> … which demonstrates that the term intermediary is always relative to a 
> perspective.

This is the key observation, I think.  If we're to make contributions that are of use to protocol designers, then we need to think about perspectives other than the human one.  Once the effect on the true end-to-end system of communication between people has been duly considered, there might be work to do.  Engagement at an appropriate level allows us to provide concrete and actionable guidance.

> Very important for me is whether 0, 1, or 2 endpoints agreed to go 
> through this intermediary (or even had the initiative to ask for or 
> select the intermediary).

0, 1, or N :)

I generally agree here, though when you say "agreed", that gets interesting very quickly.  The notion of consent has some real limitations here.  Part of what I'm talking about in the draft is ensuring that these interjections by intermediaries are deliberate, that participants are identified, and the extent of their capabilities understood.  That is all in service of a more comprehensible and robust system.