Proxy ports versus full ports in SysML, there has to be a better explanation

[mse]

After looking at the latest SysML standard and reading about proxy ports and full ports I was rather disappointed how the standards committee authored that paragraph. They start off okay by talking about some differences, but then it's like they dismantle their entire reasoning and finish off with the statement:

Modelers can choose between proxy or full ports at any time in the development lifecycle, or not at all, depending on their methodology

With such a statement, why have a standard at all? Other websites (https://mbse4u.com/2013/09/23/sysml-full-ports-versus-proxy-ports/) don't do well either. For example, in an article on SysML full ports and proxy ports, the author goes to length to explain the concept, a little more than in the standard, and even showing some examples, the author finishes however with:

I recommend to use only proxy ports and to ignore full ports.

If anyone can cite a better source than the standard, I would appreciate it.

[RobCDeJong]

All standards are based on consensus of several stakeholders. You do not have to use the complete standard in your project. Best thing to do when you start a project, is to document which modelling language constructs you should use and which constructs are not used by the project.

As Weilkiens states in his mbse blog, he recommends only to use proxy ports. I agree to this as we also only use proxy ports in our projects. But if you have a project where you buy a component as a connector between other components and you are not interested in the internal behavior of the COTS component, you can model it as a full port. This is the trade-off you have to make!