Aaron Parecki

it's a bit of a shame that protocols in URLs have been hijacked to belong to single applications (cough Apple) rather than for interoperable ones, so that to add, say , hyper: to a browser you have to fork the whole browser.

it is a lot harder to set up a server on an arbitrary machine in some ways than it was in 1989, but I think the answer is to use the URL insight from the original web and define URLs for the new protocols rather than have them try to replace DNS entirely

I remember having this conversation with timbl a while back when talking about OAuth in general - he didn't see why I needed to tell one server about another one when I could do it in the client.

I think there's a world in which your private key is also backed up but some trusted peers vs. just a single fragile file... But agree there's work to do

I think we need to explore a more peer to peer version of the web, away from client/server, master/slave and other non-parity models.

Having it tied to a private key is a pain in lots of ways too (I say this having lost my scuttlebut config while changing machines recently)

That's one opinion yes. There are good arguments on both sides.

Ideally though, an identity shouldn't have to be tied to a server, even if I own it?

Yep although the WordPress plugin requires some active effort by the user. At least it’s just installing a plugin and not dealing with markup though.

That's very true, and the Wordpress plugin makes the case as well.