[erikb]

The question with social networks are never technical. As you can see by nearly all real life examples, the technology can suck, can exploit, can be this or that engine/framework/language. What matters is who is in there. If you build a social network based on a technology you will never achieve a significant size, especially if you choose a chat platform that is less and less often used.

Just make it work and then get influencers to use and talk about it. That's how you win with a social network. Or if you care about reasonable income without winning big scale then go the Hipchat way and integerate well with some special usecase (like they integrate with all the other Atlassian products and development tools).

[SwellJoe]

Is XMPP "less and less often used"? I just started using it seriously relatively recently. It's quite nice and there are multiple excellent open server implementations, and I've been thinking on ways to integrate it into some of my own projects. I don't really follow trends in this area, so I don't know what people are doing with it...I do know that Facebook and Google have abandoned XMPP, but that's to be expected. They want to own their users, and they have enough power to do so, regardless of the ethics of the question.

[erikb]

It's a technology that you need to be rather firm in to use it. There are many tools, but often they are a little "hacky" to non programmers. That's why it gets more and more replaced by the FB chats, Whatsapps, Skypes, and Snapchats of this world. The same goes for email. Email will die a slow death. Maybe even your children can still build a career out of email know-how. But it's more and more replaced by other technologies. Just look at China were people all didn't know the internet in its youth. The first contact to internet most of the Mainland Chinese had was with mobile solutions like QQ (their ICQ copycat). For them email is like a harder way to do the same thing that you could do just as well with Wechat/QQ. So even the most serious of corps in China have Wechat/QQ support and phone, but they might not even have an email support. This will come to us, too. Just slower because people know email already.

[digi_owl]

The crazy thing is that FB and Google used XMPP (Google still may, but had disabled their servers from accepting outside traffic) earlier.

I used to have Pidgin running hooked up to Facebook so i could say a bit in touch with family.

But with the economic downturn everyone has gone back to building silos, in an attempt at maintaining profits/growth...

[toyg]

I don't think it was the downturn that did it. Companies look at successful companies for hints on how to succeed. What has been the most successful company on the planet for more than a decade? Apple.

Apple are not shy about "owning" their platform, and they aggressively leverage their walled gardens. The market did not punish them; in fact, users raved about their "refined" experience. Other companies took note and ran with it.

[erikb]

That Apple thing might have been a reason. Another one might have been that if you don't follow a standard it's easier to debug your code and add the features management craves so much. I've seen that at work projects before where the developers pretty much enforced removing standard interfaces to get more freedom to resolve their tasks. It's a pity when not even developers see the advantages of interoperability of their services, but it happens just as often.

[sethhochberg]

At my day job we recently migrated away from XMPP for our internal chat rooms and IM (all on Slack now). XMPP worked fine, but, its one less service for us to maintain ourselves and everybody using a more consistent client by default instead of the mix of Adium / Trillian / Pidgin / etc has been nice from a feature standpoint. I think many other companies are in the same boat, based on Slack and Hipchat and the other chat-as-a-service providers having huge growth.

[marquis]

Hipchat uses XMPP so you answered the parent's question perfectly. Slack, I believe, created a custom protocol.