[cookiengineer]

While I agree with you I think that most of the described effects do only exist because the way the internet works is centralized, and humans have to remember domains which they aren't very good at.

...otherwise the most googled term would not be facebook, just to click on the facebook.com link.

If there would be something like decentralized trackers (similar to the torrent architecture) you could have lots and lots of specialized communities that provide meta information about those websites and urls.

This would also allow different sources of traffic and updates if the discovery aspect of similar semantic content would be provided by something like a tagging system or a search field.

[kiwidrew]

> Humans have to remember domains

Humans don't have to remember domain names (and email addresses and URLs) because we have address books, contact lists, and bookmarks. We do have to recognize domains (and email addresses etc) when we see them, but that doesn't seem to present any issues in the real world.

Domains are absolutely everywhere in modern life: business cards, restaurant menus, outdoor advertising signage, and even in people's conversations with each other. People do, in general, understand how to use domain names.

Of course I do agree with your overall point that we need better protocols for content discovery, but I just don't think that domain names themselves are a stumbling block here.