Howard has posted a manifesto titled Mobile and Open, so I’m going to take this chance to jump in and reframe some of the stuff I’ve said along these lines before. I think it’s a great topic, and most people misinterpret what I mean when I say that there must be more openness to the mobile infrastructure. So now that we have a great touchpoint in the writeup from Howard I want to run thru a few things.
First is the Mobile Data Services and Innovation post. The idea here is that innovation normally comes from the bottom up, not the top down. The example that Howard gave of Tim Berners-Lee is excellent. This is not a universal truth I don’t think, but very often new solutions have come from people who were able to put together a solution to their own problem without having to make sweeping changes in order to do so. It’s normally only after the solution has been out in the wild for a bit that users of the current dominant system catch on to the utility of the new system. I won’t belabour the reasons behind why the providers of the previous system might even be hostile towards the new system, we’ve covered that before. I personally believe that the ability of open source to avoid this problem (the current install base being controlled by a hostile entrenched provider) is one of the major reasons for long-term success of open source. The same thing is very true in the mobile environment. Once a provider assembles their “stack” of solutions they become resistant to changing it. That’s just what their business model requires them to do in most cases. When the providers become resistant to change the environment as a whole stagnates and innovation slows down to a crawl. Keeping the environment open, and allowing the users of the system to potentially act as innovators without requiring them to get the help of providers, is the way to make sure that progress moves at the pace of the market. Otherwise the market only moves as fast as the telcos want it to.
The points about open innovation commons and the ability to organize ad-hoc networks are pretty closely tied together in my view. I’ve commented on the need for making mobile services more end-to-end in the post already linked above, as well as the need to evolve beyond the established online equivalents in some cases. The flexibility of the layered system that is TCP/IP is what has helped keep the Internet the thriving diverse environment that it is. No one needs to rip out routers and switching equiptment on the Internet in order to deploy a new application, all applications exist on top of a common basic set of capabilities. And as much as possible the equipment between the two ends of an exchange need to know nothing about the content and meaning of that exchange. This is not true in the mobile environment. Although some services in mobility run over TCP/IP style infrastructure, there are plenty that are proprietary and explicitly embedded in the network. I am 100% convinced that the long term benefit of mobility itself would be best served by adopting an infrastructure that models this end to end principle seen in the architecture of the Internet. Does that mean straight TCP/IP? No of course not, service layers need to be added to the base TCP/IP stack to provide the kinds of features that mobility makes possible. But those features should be added levels on top of a common base, and not modification of the fundamental layers. I think this would also spur innovation in mobility the same way that we saw an explosion in online activity in the early 90s. Right now we’re seeing pale reflections of online services for the most part. True mobile specific applications seem to be almost accidents (take SMS for example). The efforts to translate online activity into mobile activity and to evolve one mobile activity into another are just small bits of what could be done if the community was allowed to be self serving. I know there are fantastic applications out there waiting to be discovered. The current infrastructure doesn’t allow them to be deployed as is, and the existing incumbents won’t understand them until they see the applications in action. The only way to break the deadlock is an open environment.