The web is so fast-moving and iterative there aren't really 'generations'. I'd say the biggest growth I've noticed in recent years is CSS. The past ~4 years have seen CSS become a lot more capable, not just for design but also for interactivity, and it doesn't seem to be slowing down.
Single-player internet? At some point, LLMs can probably simulate enough of the entire internet to just fake it on localhost, and any URL you type in will generate a whole website right then and there, customized to your personality, genes, and politics.
if you tell an LLM it's a webserver, and send it http requests, it'll respond in kind for whatever url you feed it. makes for a fun way to spend some time and run up your OpenAI bill.
The next big web disruption is commercialized, self-hosted, personal servers. Everything is already there except a branding and easy IPv6. Think how disruptive the that will be for businesses based on ad revenue and traffic funnels.
You are describing the www of the 90s. Ok, we didn't host at home but basically you "built your own website". There were no "big businesses based on ad revenue and traffic funnels".
Of course (compared to now) it sucked. Discoverability was hard. Quality was (let's be generous) erratic. There was no interaction (no comments, chat et al - thats what email and news (nntp) were for.
If you look at the big businesses based on ad revenue (Google, YouTube, facebook) they evolved out of that landscape specifically to solve problems. Those problems remain regardless of where the sites are hosted, so moving the hosts won't affect them.
Due to how Google forgets everything from a year ago, I'd argue that this space is ripe for disruption. Discoverability is now even worse due to too much corporate blog spam, with either stolen content or LLM generated content that makes no sense but still leads to better SEO rankings.
If searx wouldn't be so overengineered and if people had an easy way to integrate bookmarks into their search engines/linklists, I think this could have lots of potential. Maybe RSS or the new federation protocols might be a nice base for this.
I think this is untapped potential of peer to peer networking, where your (subscribed) social circles reflect discoverability of content. Funding for maintenance and development should be seen as operational costs of the network, not as a startup exit strategy.
That is not even remotely similar. A website is just content. Owning the entire service provides full ownership of content and data and it dictates to whom you will make that experience privately or publicly available to. This was the semantic web vision of Tim Berners Lee that never materialized.
I'm not sure about other countries, but in my country ISPs usually block port 80 (so I can't host my personal home page myself). Would such a revolution require a browser that defaults to another port?