We hardly even use the globally consistent names we have. Between humans its always "hey did you check out that website I sent you?" And never "hey did you check out https://foo.bar.biz"
Sharing and endorsement can just be part of the protocol. My reference to the site may as well be in terms of which of my friends think its trustworthy so I'm better enabled to make that determination myself (some of my friends have better taste than others).
Knowing that Alice.brownierecipe.2 and Bob.favoritebrownies refer to the same underlying document with an unfriendly global name is not so gnarly of a UX problem that we need to tolerate single points of failure to avoid it.
What irony that this article is about crawling content-addressed data.
I'm not sure where IPFS went wrong, but I think our web will continue to degrade until we figure it out. (It'll continue to degrade after that also, but then we can let it burn since we'll have a replacement to switch to).
If we trusted digital signatures on content instead of server names, we could have a model where a single bit of server load propagates to millions of interested parties. As it is the server must do something distinct for each interested party.
CDNs mitigate this only partially, because mutable data means they have no good cache invalidation strategy.
There's got to be a solution that doesn't involve heaping even more burdensome requirements on those who would dare to publish.