> I don't see any specific evidence given any that good textbook is "inconsistent," merely vague accusations cast without evidence. A straw man.

Evidence from your link:

   Human–Computer Interaction, Preece et al. (1994, p. 6): The authors explicitly define perceived affordances as being a subset of all affordances, but another meaning is used later in the same paragraph by talking about "good affordance."

   Universal Principles of Design, Lidwell, Holden & Butler (2003, p. 20): The authors first explain that round wheels are better suited for rolling than square ones and therefore better afford (i.e. allow) rolling, but later state that a door handle "affords" (i.e. suggests) pulling, but not pushing.