The concern is that an apparently user-defined or custom stereotype keyword «port» appears in some diagrams for the SysML and SysPhS specifications, even though no such stereotype keyword officially exists in UML-2.5.1 or SysML-1.6 (not even non-normative). This can be confusing for those not fully familiar with the underlying metamodel machinery, and I've seen a lot of people redundantly copying the practice.
For example, in SysML-1.6:
In the SysPhS-1.1 spec the keyword «port» appears in some diagrams but not in the body text.
The only justification for it might be to distinguish between a Property as an Association end and a Port as an Association end in "associative" modelling in Class Diagrams or Block Definitions Diagrams (for special educational or illustrative purposes), but typically naming is sufficient for that anyway, and Ports can always be simply listed in compartments.
What can make it even more confusing is that SysML-1.6 DOES support special stereotype keywords for the extending stereotypes ProxyPort «proxy» and FullPort «full». Note however that these have not always been popular with some users: