The "HTML5 will catch up" argument is a common fallacy. It assumes native apps are standing still, when in fact they are moving rapidly forward. If anything, the gap continues to grow. The web isn't even in their cross-hairs, there's just so much effort for each native platform to keep surging ahead of the others.
If you think Android and iOS are "done", consider where they are headed and you can see they are barely getting started. Consider input such as speed, physical gestures, breathing, brain activity. All manner of form factors. Integration with the "real world" such as Nest and Square. Wearables, cars, home appliances ...
Whatever early-stage web APIs exist for these things are vastly, tragically, fragmented across the various HTML5 runtimes.
The "other web", that of HTTP and REST and JSON and cloud services, is a freaking wonder of engineering and a giant productivity boost to developers of any client. But the UI side of the web is continuing to trail for now. Competition among tablets and mobiles will help native apps as much as it will help the web.
I think what will happen is what happened on the desktop. There was a time when every application you wanted on the desktop had to be installed natively. This was because of the following reasons: web didn't exist yet, web existed but the tools sucked, web existed but networks were slow. Over time the networks got fast enough and the tools got good enough that the vast majority of applications could be built as a website. There are still cases where building a desktop application makes sense such as when performance is critical, but those cases are in the minority now. Building a website also has the distinct advantage that you only need to build it once and it works everywhere (mostly).
When it comes to mobile, I think we are in the "web exists and the tools are almost good enough but the networks suck" phase. When the mobile networks get fast enough, I suspect most applications will just be websites. Of course there will always be some applications that are built for native because of performance or they have some fancy touch-based interaction.
> It assumes native apps are standing still, when in fact they are moving rapidly forward. If anything, the gap continues to grow.
I'm sorry, but there are only so many ways you can display a CRUD application. I don't see the next versions of Twitter and Instagram pushing the hardware much more than they are today..
People said the same about terminal-based applications when windows came along. It just takes some imagination Do you really think Instagram and Twitter will be running on the same form factors and look+feel the same in 10 years?
Twitter and Instagram as photo-heavy apps could eventually support 3D models of the world, drone photography, wearables etc (Twitter already has a Glass app).
Same for their notifications - all sorts are possible on watches, speech, car dashboards.
And that's just assuming they stand still functionally.
As for CRUD apps, there's loads of innovation happening. Look at Amazon's new Dash device as one of many examples that change the interaction style from clicking on a web page to speech + real-world scanning. I'm guessing it's not powered by HTML5.
I see no reason why it wouldn't catch up. What makes a good UI on a native app? Fast responses and clean animations? Fancy transparency and 3D effects, perhaps. There's no reason a web browser couldn't do these things. Web UIs may always be slower than native but with fast enough processors it will eventually reach a point where you just wont notice the difference.
There has been serious improvement in many areas of mobile web/browser support in the last 12 odd months; for what I've seen. Thanks to Android, Chrome, Firefox and Dolphin.
If you think Android and iOS are "done", consider where they are headed and you can see they are barely getting started. Consider input such as speed, physical gestures, breathing, brain activity. All manner of form factors. Integration with the "real world" such as Nest and Square. Wearables, cars, home appliances ...
Whatever early-stage web APIs exist for these things are vastly, tragically, fragmented across the various HTML5 runtimes.
The "other web", that of HTTP and REST and JSON and cloud services, is a freaking wonder of engineering and a giant productivity boost to developers of any client. But the UI side of the web is continuing to trail for now. Competition among tablets and mobiles will help native apps as much as it will help the web.