These days I do not code a website to begin with. I create web services(REST/JSON) first and then code a native app to access it if my use case fits mobile better. If I need a website, then I implement it by calling the same web services. I believe the web as we know (of front end data handling) will surely die. It will be replaced by headless (UI less) services web. As users migrate to mobile/tablet devices, and new mobile capabilities (touch, location, context etc.) are accessible, many traditional web use cases need to be redefined to optimize for new form factor and to exploit the new capabilities. This will make many new start ups follow a mobile first approach (a la instagram). Web sites will be required for few use cases that requires heavy data entry and extremely complex UIs. And probably the 80:20 rule will apply. Websites has been built on duct tape technologies for long, and today's mobile development platforms are way too sophisticated and easy to design, code and debug. Yes you don't get the cross platform compatibility - but is it really that important as it is made out to be?
If I am a start up, I will try to iterate on a single platform ( a la instagram) and get my product right. Then I can move to new platforms - It takes more effort to get the product right than implementing it on several platforms. If my product is successful, supporting that on a new platform will take 1% of the money I could raise and effort. Who cares a hoot if you have a half baked product released at once on 100 platforms?
If I am a start up, I will try to iterate on a single platform ( a la instagram) and get my product right. Then I can move to new platforms - It takes more effort to get the product right than implementing it on several platforms. If my product is successful, supporting that on a new platform will take 1% of the money I could raise and effort. Who cares a hoot if you have a half baked product released at once on 100 platforms?