Online platforms often witness harmful communication. While the previous research has mainly focused on the linguistic aspect of such communication, the role of the authors' stance toward the discussed topic remains unclear. To address this gap, we first construct a unique dataset from an online platform containing self-disclosure annotations about authors' stances. Our analysis of the constructed dataset indicates that anyone can be a target of toxic communication, whether a real person or a virtual entity. We, then, demonstrate that toxic communication is not limited to haters, but also fans. Writing toxic comments does not necessarily indicate a negative attitude towards the subject or individual being discussed.

PDF