Because that's how they _be_. Even if we created a monotonous people without any identities, different persons with different abilities and interests will diverge from others and cluster with similar individuals, forming various intersecting groups, and lead to a complex system of group and personal identities. (edit, forgot: Such groups bring together diverging dialects, first diverging metaphors, then words, phrases, etc., so on maybe until mutual intelligibility is lost.)
Your XMPP analogy does not hold because XMPP was popular among some tech geeks whereas even my granny who does not understand the concept of video and blushes and hides when changing her clothes when the TV is on gets to use WhatsApp, with help from my mom who thinks her WhatsApp is locked because of the recent encryption thing they introduced and the yellow little message the app displays related thereto.
In much the same way, consider the step backwards we've taken in messaging. We had a lingua franca - XMPP. We gave it up so that various powerful actors (google, facebook) could lock us into their unintelligible dialect.
Was this a good thing?