For example, when you're interacting in a customer service context on, say, steam, you are not anonymous. Valve can change its TOU to make it so that you forfeit all claims to redress if you use abuse during a customer service interaction, and can further scale that up to include additional consequences (banning you from participation in certain ways or from certain benefits).
Additionally, "in band" consequences don't have to compromise anonymity. On twitter or facebook, for example, you can have abuse reports that are verified result in different levels of account restriction. You can make it so that the account's posts are no longer visible in other people's timelines as replies, for example (e.g. they are only visible to people who specifically follow that user or when specific posts are linked to directly). You can restrict accounts in other ways and even block access to accounts for short periods of time. Or you can ban people forever. All of these techniques have been used for years and years to prevent abusive behavior, the only reason they haven't been used effectively against this particular sort of abuse is because of apathy on the part of most platform owners, but that can change easily.
And the same dynamics can occur with individuals hurling abuse online. Those individuals could experience consequences for their actions (from twitter, valve, potentially the criminal justice system in some cases, etc.) from institutions as well as from society as a whole. Telling a game developer "I will rape you!" should be an action that dramatically curtails the perpetrators career options and results in losing friends, but today it doesn't.
Sure, people hold dark impulses inside them routinely, everyone understands that. That doesn't mean they are uncontrollable. This is a recent phenomenon, which itself is proof enough that it is not due to some uncontrollable universal urge.