I think an interesting question though is what ought to be the best method of delivering apps on mobile in the future in order to increase innovation and adoption, overcome distribution/discovery problems etc. In my view, the constructs of local versus network are historical and artificial, notwithstanding the very real problem of network access that can still kill network-only apps in many situations.

I think that richer web programming models are required for mobile, such as local device access APIs via the scripting engine, persistence etc. Should browser apps have a comms pathway to MIDlets? These are interesting questions to be pondered.

For me, the future of mobile apps has to support multimodal apps (real-time comms triggering and triggered by web events, or think of using the browser as the UI for comms apps) and mash-ups. Most of the interesting evolution of mobile apps will require such approaches, I am fairly convinced of this.

At the same time, will side-loading become more important? After all, this is how the iPhone works and no one is complaining. It depends on whether or not the user will tend to stick to a solid set of apps, unlike the web where we are more fickle and like to sign-up almost daily for new services. User behaviour in this regard is still largely unknown.