The Ramifications of Google Talk

It's now all quiet at the Google Talk front. It's as if all the initial buzz that Google Talk generated was snuffed out with the eBay-Skype merger. However, the ramifications of Google Talk may actually be bigger than even the initial buzz.

The problem with many analysts is that they continue to mischaracterize the nature of Google. Although Google makes almost all it's revenue from Advertising, it is not an advertising company. It's a disruptive force picking apart the advertising industry piece by piece. Google is essentially an infrastructure company that leverage it's brain trust to build massively scalable infrastructures that's econmically feasible. That is what google does, the fact that it makes a lot of money is more of side-effect rather than a well thought out plan. Google Ads one may argue was an afterthought and fortunately with a very lucrative side-effect.

So when Google sets out to get into the voice telecommunications space by its release of Google Talk, it is a strategy to first on building massively scalable voice services rather than contemplating how to make a buck. Google doesn't count its eggs before its hatched. It figures out how to out engineer everyone else and then reaps the economic benefits of technology prowess (if the opportunity avails itself). For example, despite the incredible usefulness of Google maps, it's still unclear how Google monetizes its technology.

Now let's dig deeper into Google Talk's architecture to reveal something that may turn into a true juggernaut. Google Talk is based on the architecture prescribed by Jabber (a.k.a. XMPP) protocol. The architecture of Jabber is analagous with email, that is multiple servers cooperate to provide an end user experience. Contrast this with protocol underlying Skype. Skype's architecture is a overlay peer-to-peer network, where any client connected to the network can provide server like functionality.

The common notion of achieving massive scalability is to off-load work into clients. There are a disproportionate number of clients versus servers, and if one could tap that unused capacity then one may have sold the not only a technical problem but an economical one as well. However, by transposing processing responsibility to a client one ultimately yields ubiquity.

The question one needs to ask is whether a network architecture like Skype can is feasible within a device restricited mobile device. Technology will obviously improve over time. However Google understands the value of the being pervasive before anyone else. That's why it has invested in novel services like Dodgeball which is based on ubiquitous simply technology like SMS. Google isn't going to wait to integrate with the carriers' location based services.

Extremely limited clients put the onus on servers. Anyone can develop servers, however developing server farms that can economically scale is Google's forte. Google intends to leverage this expertise to roll out ubiquitious services without waiting for the masses to upgrade their mobile phones. It's a pragmatic strategy.

Telecom industry continues to harp on the idea of converged services with visions of communication occurring in multi-modalities. That is voice, music, video all present in the same conversation. This use case however is not something that exists in some far flung 3GPP future. It exists today in the form of Instant Messaging (IM). IM clients continue to add new and novel ways to interact. It is the prototypical converged application, the interface of our mobile future will lend its ancestry from an IM client.

Therefore, the company that is able to ubiquitously port their IM client to mobile devices first would be the one that wins the whole ball of wax. Interestingly enough, Jabber based J2ME implementations already exist today. J2ME already is available in a vast majority of cell phones today. The one technology road block is a good J2ME based codec. Google could potentially work out this difficulty by providing codecs on its service in combination with J2ME MMAPI. That in combination with the dark fiber that it has been rumored to accumulate one could then speculate that a J2ME enabled Google talk is in the near horizon. Just as we've seen Google take lowly javascript to render highly interactive email clients and map viewers, one could see Google do the same with voice with what already is ubiquitous on mobile phones (i.e. J2ME). If the mobile phone manufacturers don't enable this capability, then expect to see Google to take action themselves (see: Andriod acquisition). Now that's something that should give everyone the willies.

[update] "Google's WiFi Service" now extend this to WiMax. Scary proposition.

[update] CNET has a news story along similar lines.