I was talking to a friend who was thinking about the internet we want to have, decentralized, less silos, a bit like the nineties where it was possible to have static pages, host email, write your own CGI scripts, and it was all step by step easy and possible if that was what you wanted. And we got talking about the kind of things we need to today to get this back. Do you have reading suggestions? Blogs to read? Projects? People to follow?
@kensanata my feeling on this subject is that asymmetric access networks are really at fault here -- many DSL and DOCSIS networks are painfully asymmetric -- 16Mbps down, 1Mbps up. 200Mbps down, 4Mbps up! just barely enough to sent TCP ACKs to sustain a passible ingress speed.
the only solution to that foundational problem, is cheap virtual machines available from the plethora of hosting companies. but that just reinforces the client/server model further, imo.
@aag @kensanata while i agree that the asymmetry sucks, in absolute terms the upstream is probably still bigger than the symmetric speeds most of us had in the 90s. A T1 was 1.5Mbps, iirc?
@telent @aag @kensanata nothing else comes close to NAT in terms of damaging free software's ability to put power in the hands of regular people.
as far as I can tell the only real progress being made is in SSB-land, which looks to me like a somewhat shaky technical foundation.
in my opinion, the only reason NAT is damaging is because of an inherent flaw in IP -- addresses are both locators and identifiers.
besides, with the processing power available to *every* network connected device, anything useful is going to be an overlay. who really cares about the underlay network?
so I'm back to "asymmetric bandwidth to the end user" is the biggest problem
@aag @telent @kensanata sure; I'll agree that asymmetry in the network is the root cause. but bandwidth availability is only incidentally related to that.
well, of course. but we didn't have machines that could process, or software that required, much more bandwidth than that in the 90s.
NAT destroying the end to end principle was unfortunate, but many a protocol designer has over come that a few times over (with varying degrees of success)
@aag @kensanata ISP firewalling and (especially) NAT obviously don't help, though I can see they have valid reasons (spam, address shortage) for wanting to