Fast forward seven years, and some of the biggest names in software (Instagram, Lyft) have completely abandoned the web for "desktop" (really, native) software.
That so much code ships directly to mobile devices says a lot about the power of centralized app stores and truly sandboxed installations where one app can't trash the entire system. We've come a long way.
Despite this, I'm inclined to agree that the web is threatened. The desktop may be defined by local files but the web wasn't defined by remote ones. It was more about the culture. More about anyone-can-publish. To me, that's the baby at risk in the mobile bathwater.
That, and any aspect of personal computing that can't be performed on a phone. Which is most of them. I'll stick with the hemorrhoids and circadian disruption, thanks.
My wife and I were talking about this the other night: It's not that people who learned to compute on the desktop are abandoning it for mobile. It's that most people couldn't handle desktop computing in the first place. It's too abstract (dare I say difficult?). So the mobile revolution is really about the unwashed masses computing for the first time. Perhaps that's why mobile apps and operating systems eerily remind me of soap operas and reality TV.
Totally agree. For most everyday scenarios, using Windows is like flying a Boeing 737 for your daily commute: it's a huge overkill, requiring way too much setting up and maintenance and when you incidentally touches one switch the whole thing stops working.
Windows 8 didn't take off because what MSFT did was in addition to all the panels of the 737, they put an iPad in the cabin. For those familiar with Windows that's no big deal: this is why we see people shrug off all the Win 8 hate, saying "it's just Windows 7 plus touching interface". But for those had trouble coping with Windows in the first place, it's now doubly confusing.
The cycle takes 10 years or so but once you've seen a few you'll realize that the same things come round again and again. The only progress is through better hardware. Software guys, with their (our) fascination for the new and shiny and disdain for "legacy code" means we just reinvent the wheel over and over again.
So, clearly the plan is: 1. Write a good database (as Oracle say "the database is the application") 2. Put a dummy interface on it (RESTful HTTP anyone?) 3. Write a native client for each platform (or maybe a web based one if you want endless testing and pain) 4. Profit? (Or perhaps running costs... and no profit)
Weirdly, they recycle this concept again and again as something ground-breaking and wonderfully new but this is 99% of apps, as you say.
As you implied, the nitty-gritty is in the db for most applications. And I've come to accept that much of the most interesting and innovative stuff happens in game programming.
And I'm not even including non-web apps like Instagram, Vine, Snapchat, Uber, Lyft etc. here.
The only place where I can see we would avoid to do native apps is for mobile apps. Because in the future mobiles will be faster but they stay limited by their small screen. So you will use higher level tools to do your "simple" app that doesn't require thousands buttons and stay more productive.
But don't those software only work by connecting to their cloud servers? Could you use Instagram if Instagram Ltd turned off their servers? That means it's inherently web software.
The web approaches native, with asm.js and webgl. It is not just a document anymore.
Native approaches web, just look at what Apple, or Dropbox are doing(they use native code for giving you network capabilities).
Web software basically runs anywhere and doesn't have to be updated; it has minimal platform dependencies.
Native code, on the other hand, is a 25-year-long story of install/uninstall, sophisticated package/dependency management, chroot jails, and viruses.
I fear that, if web software continues its crawl out of the browser sandbox, we'll get all the disadvantages of native code (security exploits, dependency problems) without any of the advantages (speed, high-performance graphics)