I feel skeptical about using Dunbar's number as a standard for online interactions.
It was originally observed in connection with things like nomadic tribes and sections of the Roman military. There's a study that shows that it may be applicable to online social networks, but if we purposely use it to limit ourselves when creating new types of social network, then that choice affects the network's growth. It could be a self-fulfilling prophecy.
How do we know what network size is selected for by "natural" evolution in social network structures that haven't been conceived of before?
I haven't done that much research on this yet. Just a thought.