"The web" is a collection of independent networks providing a means to traverse hyperlinked text documents across any network that is addressable.
Not only are servers not a problem, host-based addressing is what makes the web work at all, as numerical addressing by itself would have killed the web's growth long ago, and users would have no reasonable way to address content.
You're ignoring the data silo problem. Of course the Web is decentralized in terms of accessing web pages, but-- the majority of data is not composed of pages, and the majority of actions on the Web are not composed of hyperlink traversals. For the applications on the Web to be decentralized, the data and behaviors have to be decentralized as well, and they are not.
The majority of data is composed of pages, but the biggest uses of the web today are not. The web isn't the web anymore. It's a shitty application platform.
In this sense, the web works exactly how any application platform does. You can't get content out of old apps, or apps that no longer run, or on systems that no longer run. This isn't data siloing, this is just legacy applications.
With web apps the data is siloed, but it doesn't have to be, and didn't used to be. The Internet Archive is proof. Getting the data out isn't that difficult, IF it 's not hiding behind a web app. The difficult part is to convince people to stop writing applications which are siloed and do prevent easy access to content that the web used to provide.
Peer to peer networks are not a solution to incompatible legacy applications. It's like a vehicle which is immobile when it runs out of gas. Instead of building the vehicle so it can still be used when it lacks power, they're changing the way the roads work. It's ridiculous.
You can't just wish that people would build apps a different way though, right? You have to change the system somehow. So, this is how we're doing it. No legacy support, but it does have interesting new use-cases that can disrupt the legacy.
It isn't.
> but why?
"The web" is a collection of independent networks providing a means to traverse hyperlinked text documents across any network that is addressable.
Not only are servers not a problem, host-based addressing is what makes the web work at all, as numerical addressing by itself would have killed the web's growth long ago, and users would have no reasonable way to address content.