The issue is anyone with a web browser (even ones provided by the ones with walled gardens) can use the HTTP protocol to see hypertext documents. But Facebook users aren't using a public protocol in the Facebook messenger, so they can't easily communicate with Twitter users, who also don't use a public protocol.
It's not the end of the world, but allowing more people to connect more easily is a good thing. The larger issue is that people are less likely to connect once they find their comfortable niche. You don't use Facebook or Twitter but someone who does and would be interested in your thoughts are less likely to stumble across you.
I think there are practical concerns too that push away from openness and towards silos - it isn't just about money.
Open protocols can be slow to improve because of the very nature of their openness - they have dependencies and need to remain interoperable.
Closed protocols can adapt faster because one entity controls the entire thing. In the case of chat XMPP was in the lead for a while, but the mobile situation was terrible - connections were constantly dropped and even though there were clients that supported multiple accounts (Meebo) it was a pretty bad software experience. The current messaging products are not open, but they are better.
This is a shame I think - because it'd be a lot better if the open products were actually better, not just better because they are open.
Maybe the trick is to hack on a better open protocol, but even then it could similarly be outpaced. Maybe the path to this is defining an open 'social' protocol and fixing the internet identity issue at the same time. Maybe keybase.io can do this?
It's not the end of the world, but allowing more people to connect more easily is a good thing. The larger issue is that people are less likely to connect once they find their comfortable niche. You don't use Facebook or Twitter but someone who does and would be interested in your thoughts are less likely to stumble across you.