Egregoros

Signal feed

Timeline

Post

Remote status

Context

4
@nyx I skimmed it (I'll read it more thoroughly later) but there's a lot to like here, especially like your prescriptions for the future of this network. I respect and admire what you're doing and your thought process behind it. I don't have any critique or anything I agree with it. we need to get out from under DNS because a lot of censorship has already happened here and in the future is going to ramp up as people notice this space.
@sun thanks, yeah I would like at some point to write a more thorough piece specifically focused on things I think could be done to improve fedi (esp ideas I've had that are more technical) but this was already long and bordering on disjointed rambling. DNS was one thing I thought of as well but that's a really hard problem to solve without effectively doing a netsplit and using/making something like opennic lmao
@nyx @sun we already do this kind of netsplit in effect when we use a new host name or address. the reason netsplits aren't end-alls is because of federation -- irc is already "federated" and dns is already "federated" but this is a different form than fedi being "federated". to federate https you use a web browser which inherently extracts the authority component of the address, and we have countless https authorities. yes dns is rooted in iana but you just need a path to the resource which allows alt roots, and also alt roots can still align or agree with the iana consensus in probably the majority of cases where there isn't a reason to disagree (censoring authoritative nameservers, although it is more popular to censor widely used dns forwarders like google and cloudflare because recursively resolving authoritative dns records for yourself from the iana root is a lost art)

so long as we use "names" to refer to things, we need a "name server" to dereference those names. but those names can be anything -- dns names, https names, did names, it's all the same at the end of the day, just different protocols and content formats for describing resources. the focus shouldn't necessarily be to avoid dns but rather to avoid dns *censorship*, which can be done without throwing out the baby with the bath water.

one cool thing about uri space is you can always put names in a namespace as long as everyone agrees/understands what that means. our root namespace is the iana registry for uri schemes, but many people ignore the scheme and start at dns names (assuming https: or whatever). even so, a dns: uri scheme exists. dns:foo.example and dns://1.1.1.1/foo.example and dns://192.168.1.1/foo.example might all resolve to different resource records. so we can route around censorship if only we had an agreed-upon network consensus for a convention that lets us delegate this way. or we run communal dns resolvers that aren't google/cloudflare, like a community-wide hosts file almost. "the network" becomes whichever resolver you use as the entry point. (the same thing could be done with https: instead of dns: if you pick one or more "identity services" which do nothing but mint ids and map/redirect them to current locations)

Replies

2
@sun @nyx these problems aren't insurmountable, it's just a question of whether they are worth the effort. neither are alternative technical solutions necessary if the problem is actually social (unless it makes it technically impossible to enforce those social constraints in a way that is cheaper than undoing the social constraints or routing around them)