I'd be surprised if they were really completely dropping it. It probably has more to do with rush to market on latest features than some attempt to slam doors on people's feet. XMPP for voice really is not all that great, I've implemented it myself.

We've supported Googleifyed xmpp jingle in FreeSWITCH [http://www.freeeswitch.org] since 2006. Its never really looked like something that would scale since the signaling protocol was over an already high level transport designed for chat. XMPP offers one really good feature, you can reach users directly using a user@domain.com style address and there is no NAT or any other networking or lookup issues to reach that user. The downside is it won't scale due to the fact that xmpp servers are heavily rate limited and not designed at all for tons of messaging at heavy rates. So Google luckily had the best super cluster of xmpp services in town and that allowed them to build on that for the voice stuff but I bet it became clear to them quickly the challenges with trying to exponentially keep up.

I would gather more info and look for the whole story before passing judgement. If they have some goal to use some fancy new audio and video services, there is a chance they can focus on that first and make sure it scales and it should be trivial to gateway that back to xmpp for existing topology.

Coming from a telephony background, I am more concerned with the tunnel vision towards paradigm shift at any cost that threatens people who still use telephones from being disenfranchised by this attempt to reinvent communication too drastically too quickly. Balance is key when it comes to legacy vs new wave in communication technology.

I have been working on the open source softswitch FreeSWITCH http://www.freeswitch.org/ for almost 6 years now.
During that time I have seen SIP continuously evolve to try to cover its own shortcomings which all stemmed from the simple concept of "If we base it on HTTP, we can use proxys and never have to worry about media" Of course this is not true and the amount of complexity that is put into each SIP device is much too great which is probably why Cisco prefers its own lighter "skinny" protocol. Sadly they own Sipura and Linksys and have SIP on their devices using countless hacks that make interop a nightmare. I am sure you can do many of these same attacks on any brand of phone. There are much better reasons out there to curse Cisco for being involved in VoIP. =D
 

We work on an open source softswitch called FreeSWITCH http://www.freeswitch.org/
Blocked ports and content filtering can mess up Voice over IP traffic running on your broadband line which can be used as a free alternative to the "Digital Phone" services many providers offer. Some entire countries already do this type of thing like China for instance. There are ways around it using secure packets so the payload cannot be sniffed and other workarounds but it would be a huge pain if we had to do that inside the US.

We have an Open Source project called FreeSWITCH http://www.freeswitch.org/ that allows you to do this sort of thing with any computer running Windows MAC or most UNIX. It can be paired with traditional phones with a small analog adapter or a hardware telephony card from Sangoma http://www.sangoma.com./ But you could just get a software phone for free as well and play around with it.
 

When you create closed source code you have a much higher chance of flaws because your code can not tested nearly as much as open source can. As the leader of an open source project, FreeSWITCH http://www.freeswitch.org/ , I am fortunate to have a very large crowd of beta testers who help ensure our releases are as stable as they can be. If you are selling the application and never letting anyone see the source you run a very high risk of missing something in Q/A and releasing buggy software. When people pay for it the will get angry so I am not surprised such a suggestion is being made but I find it unpractical to enforce since if it "works right" is hard to judge in some cases besides maybe medical equipment or other situation where human lives are at stake. Blue screens of death are hardly an excuse to sue anyone.

Our project (FreeSWITCH) uses the MPL for the main application and BSD for satellite libraries that we create that can be used by other projects etc.

Once you decide to have open source code, it's more logical to stick with the fact that at least the core code is FREE and come up with ways to develop a product on top of it if you want to have something to sell. Otherwise it sounds like an "open source tax" and businesses do not like uncertainty. If they choose to use a code base they need to know it will always be available.