It is a natural consequence of Web 2.0.
During Web 1.0 people hosted their own servers and content was shared freely amongst people as to what they themselves held for important rather than what some automated system decided for them. But most notably, websites were largely read-only. You could only see whatever the site owner would put on the site himself. If there was something that you posted onto a site, it was through contacting the owner himself directly. With the rise of 2.0, websites arose where people would post content onto an automated shared platform hosted from a singular server held by a singular user. This content allowed these places to make a pretty decent profit off of people willingly posting content on the site through ad-revenue. And with rising content creator count there was also rising viewer count and thus also a rise in advertisement revenue: what you end up with is a positive feedback loop that results in the website expanding and growing its influence far further than what one would expect from a simple read-only website.
The death of the internet is due to rapid and persistent centralisation, which as a natural consequence of this centralisation also leads to the consolidation of power we see today. Enter legislation and share-holders into the mix, creating the stagnant and lifeless internet you see today. The formation of a central point is what allows for legislative bodies and corporations to be able to attack and influence what people are and aren't allowe to see. Unlike torrenting which to this day is undefeated no matter how hard governments have tried to clamp down on things like TPB and Limewire due to those things being largely decentralised, Web 2.0 sites and "cloud" computing services as a whole are incredibly easy to intimidate and control. And yes, among Web 2.0 is also the very website we are on right now, but at the same time there is also the community surrounding it: if one imageboard dies, people will move onto the next. We already have the means by which to find each other even if we do not know the identity of any individual user, there are commonly accepted cultural conventions that allow us to find one another in the event of site death. We simply look to a different imageboard hosted by another user in a nigh-identical format. It's decentralisation by what could be described as a disjointed webring. This type of community agreement makes it a lot closer to the 1.0 websites of old than something like Facebook or Youtube, even if the barebones website itself is still Web 2.0.
The solution is simple but at the same time it requires an active effort, making it more difficult than the 2.0 passive consumerist internet lifestyle. It is to return to the decentralised roots of the web: people hosting their own content and sharing it amongst themselves. Not relying on a central node but rather relying on a distributed if not completely decentralised network. And it may seem outlandish to propose to go back to read-only websites but you see that same Web 1.0 sort of revival through Tor. All those "weird" places you see on the deep web? Those are the same type of websites people were running during 1.0, just in a more modern image and using new technology.