1. Twitter - by far the social network that I use the most - is struggling to break 300 million monthly active users and is not hitting revenue targets. (Contrast with Facebook's 1.44 billion monthly actives.) Even investor Chris Sacca has warned that he's going to start making "suggestions".
2. Instagram - still a newcomer in many peoples' eyes - is beginning to send re-engagement emails in response to flagging user growth.
3. The 2016 US election is apparently going to be huge on Snapchat. Translation: Snapchat is over. The next generation of young users are already looking for something else. Snapchat was released in September 2011.
4. The Document Object Model - core to how web pages are manipulated inside the browser - is slow, and may never catch up to native apps. We've known that responsiveness matters for engagement for over a decade.
5. It's possible to build more responsive web apps by going around the DOM. But these JavaScript-based web apps are harder to parse and often can't be seen by search engines (unless you provide a fallback, which requires a lot of extra programming time).
6. Push notifications - which are core to apps like Snapchat, and possibly the future of Internet applications - are not available on the open web. Browsers like Chrome are implementing them on a browser-by-browser basis.
7. Facebook has no HTML fallbacks, renders almost entirely in JavaScript and lives off push notifications. Twitter has HTML fallbacks, is very standards-based, uses push notifications but also SMS and email, and is generally a good player (with respect to the web, at least, although it's less good at important features like abuse management). Facebook is kicking Twitter's ass.
8. The thing that may save Twitter? Periscope, a native live video app, which is highly responsive and live-video-heavy.
9. Users have stopped paying for apps, and instead opt for free apps that have in-app purchases, so they can try before they buy. We're a long way off having a payments standard for the web.
10. There's no way to transcode video in a web browser, which means uploading video via the web is effectively impossible on most mobile connections. (Who wants to sit and wait for a 1GB file to upload, even on an LTE connection?) Meanwhile, the web audio API saves WAV files, rather than some other, more highly-compressed formats you may have heard of. Similarly, resampling images is difficult. In other words, while the web has been optimized for consumption (albeit in a slower way than native apps, as we've seen), it has a long way to go when it comes to letting people produce content, particularly from mobile devices.

What does all of this mean?

I don't mean to be pessimistic, but I think it's important to understand where users are at. The people making the web aren't always the people using it, and there's a serious danger that we find ourselves trying to remake the platform we all enjoyed when we first discovered it.

Instead, we need to make something new, and understand that if we're building applications to serve people, the experience is more important to our users than our principles.

All of these things can be solved. But while we're solving them at length, native app developers are going off and building experiences that may become the future of the Internet.