Take a look at what's currently available in desktop web browsers: maps, document and spreadsheet editing, highly interactive charts, video conferencing etc. (things previously thought impossible/impractical for a web browser to do) If you need huge amounts of data to even run your application, the argument to go native makes some sense, but the browser (even on mobile) isn't as inherently limited as the article describes.
Lets not try to limit the web, but try to make it better and then let the users decide which they would prefer to use.
I'm not limiting the web. You're not limiting the web. The web has limits. Actual, hard physical limitations (along with confidentiality and privacy limits); limits that something running entirely locally, directly with your actual operating system simply suffers, and will always suffer, to a much, much, much lower degree. Your opinion (and my opinion) is meaningless when confronted with actual, physical reality.
Besides having to transfer data to run the application in the browser, what limitations are inherit in a web-based application? (I acknowledged the data transfer limitation in my previous comment).
Browsers (especially mobile browsers) are unnecessarily limiting what can be done on the web. Adding additional permissions could enable browsers access to more system resources and components, especially on mobile.
Could you provide some examples of applications that don't require a lot of data that couldn't be run in a web-based environment, because of physical (and not current browser or software) limitations?
Doesn't even necessarily have to transfer data with web workers, service workers, caching, etc.
Google IO 2015, I believe, stores it's data so that when you load it up it loads from a cache that it's service worker will reload once you have it open. Allowing it to work offline and removing the cost of transfer at the beginning.
Lets not try to limit the web, but try to make it better and then let the users decide which they would prefer to use.