I do not know what "porn" is and I have a lot of experience with porn-filtering technology, so I had to define porn using some sort of heuristics for practical business purposes (not scientific research, but business) and did OK in that task, but the whole exercise demonstrated that it is basically impossible to define it well.

That leads me to believe that "most of us" who have not had that kind of experience are even less equipped to define "porn".

Note that the business needs of porn filtering differ dramatically from the research needs, because for the business purposes it is OK to err on the side of being conservative - at least to some extent - and business needs do not and should not entail the level of rigor inherent in scientific research.

In other words, a sloppy definition is OK for business purposes but not OK for research.