In a recent story about how Google creatively destroys markets, I said that only Google will run the servers for Wave, its re-think of e-mail. I was wrong about that, as Google reps took pains to tell me. I want to set the record straight. What Google is doing with the Wave communications architecture is important enough that it merits its own story, not just a strikeout in the original.
Google has said it will "federate" Wave. That means it will make it possible for anyone to operate their own Wave server and have it communicate with other Wave servers. This is just how e-mail works today: Anyone can run an e-mail server that can send messages to and receive messages from any other e-mail system. The Internet routes messages from server to server.
I was told that anyone will be able to "build their own Wave server without involvement from Google." That means corporations and governments will be able to deploy their own instances of Wave inside their secure firewalls if they like, and decide how or if they want to open up their servers to the outside world. For businesses with strict data retention and auditing requirements (e.g., all public companies, governmental agencies, health care businesses, etc.), this also means that they'll be able to write in software to meet their needs; or that other companies will be able to create and sell Wave servers. (If, that is, business gets behind Wave at all.)
However, sources familiar with the intricacies of building a real-time synchronization engine, which is what Wave is, tell me that it is incredibly challenging to make such a system work "at scale." Showing off a Wave demo is one thing. But a successful, well-performing, wide-scale rollout requires advanced technology that few companies have the chops to write, and it's hard to keep performance up as the size of the user community grows. This may be why the original planned release of Wave to developers outside Google has been delayed by several days.
Wave will be based on (and extend on) the existing messaging standard, XMPP (eXtensible Messaging and Presence Protocol). So a lot of Web developers will be able to get started quickly on writing Wave extensions and apps, as well as building their own servers to run Wave or back-end services for their Wave plug-ins.
Wave may represent new thinking about packaging real-time communication, but it's not based on 100 percent new or proprietary technology, so we might see some interesting third-party extensions to Wave very shortly after it starts rolling out to the public.
I stand by the rest of my story, especially my main thesis: Google destroys entrenched markets. That's not a bad thing, though. Especially when it comes to e-mail.