The term "philosophy" can be defined as:
Philosophy implies discussion
references to empirical foundations
ref: Paul KT Persson (2013) vetenskapsteori.se
An earlier (2012) proposed definition was "Philosophy implies discussion of topics
without references to empirical foundations". A softer definition is "discussion about how something can be, or could have been" (as opposed to how something is shown to be).
Why? - The contents of the definition of philosophy
The term "discussion" is included in the definition above in order to distinguish between philosophy and an arbitrary statement. The demarcation stresses that philosophy should include some type of argumentation for what the discussion wants to highlight. Only the phrase "a stone can be gray" should therefore, in my view, not be covered by the term philosophy, while a discussion of rocks and rock colors probably may be encompassed by this term.
If a discussion is based on references to any matter of fact, it is categorized outside the area of philosophy. Some examples are history of philosophy (when the discussion refers to what other philosophers have expressed) or natural science (when the discussion refers to physical objects). Conclusions based on combinations of such references are still within the same areas.
The discussion about rock colors, according to paragraph "...discussion..." above, hence should not consider the real appearance of rocks, because it would in that case be categorized, not within the field of philosophy, but within geology.
See also below in the section "A complication for philosophy".
With "empirical foundations", unlike "observations", is meant something that is reported in such a way that risks of errors are decreased. For example "I felt that..." is not, in my opinion, an empirical basis, although it may be said to be an observation.