Turns out that the theory that online behavior turns negative because of anonymity was based on some 1980s experiemnts that were (surprise!) simplified in the popular retelling. Recent research does not support the theory. To the contrary, there is now a body of research that indicates that revealing true identities increases harassment of target groups, even when the harasser is not anonymous.
Or, put another way:
Some people are jerks.
It is often factors such as fear of punishment/penalty that moderate behaivor, as well as acclimation to and sensitivity around social norms.
Anonymity does not appear to meanginfully contribute to these factors to the point where removing the cloak of anonymity by itself alters behavior.