One of the frustrating things about web3 is that people who talk about it seem to be lumping together a bunch of things - intrinsic decentralisation, crypto, disillusionment with corporate social media, virtual reality - which don't really have much in common. The term 'web3' is obviously meant to make this iteration of the net seem as inevitable as Web 2.0 did in hindsight. When I was thinking about this post I was going to go on a rant about how both Web 2.0 and web3 are social changes disguising themselves as technological shifts, but that's not really true.
It seems easy to connect Web 2.0 with technical changes like complex frontends and the app-ification of everything, but I think that puts the cart before the horse. The websites which allowed the internet to be a space where more than a minority of users were able to upload content - MySpace, and the first iterations of Facebook and Twitter - were, technically, more web 1 than web 2.0. Things like single-page apps and complicated frontends and so forth were a response to those platforms taking off, and to the adware which followed that ascent. It should always be remembered that Facebook didn't get where it is by being especially good at technology: it had to get good at technology in order to keep up with the demand created by its initial success.
And the typical Web 2.0 stack is so complex, now, because it's a pile of hacks, many of them ingenious, for serving fairly complex applications over good old HTTP(S). The backend, similarly, has changed, to keep up with demand and the need for flexibility - hence CDNs, the cloud, containerisation - but those machines, however abstracted they are away from the old box humming in a server rack, are still, essentially, in the business of sending content over the network to be reassembled and run in the browser.
Web3, on the other hand, to the extent that it's anything at all, promises to reimplement all of this at a low level, which is one of the things which makes me very sceptical about it. The nearest thing we've had to a revolutionary protocol would be Bittorrent, which was a genuinely grassroots peer-to-peer network which in the jurisdiction where I live got shut down by the ISPs, though those who know where to find these things are still using it, I imagine. To the extent that it exists at all, even people who are relatively positive about it make it sound extremely jerry-rigged and cumbersome, with entry- and exit-points which are even more centralised than the web already is.
A few months ago I got a new iPad to use as a control surface for my SuperCollider synths, and when I was browsing around the app store it was remarkable how 2010 it all felt, and I realised that at least part of the eagerness for web3 is that the industry needs a next big thing, whether or not there's any demand for it. The thing about big revolutions, though, is that you don't necessarily get another one just because you missed out on, or are nostalgic for, the one which happened twenty years ago.