Anarchy Eroded: Project Efnext

Ray Dillinger bear at sonic.net
Sun Dec 31 10:14:30 PST 2000




On Sat, 30 Dec 2000, Eric Cordian wrote:

>Jim Choate writes:
>
>> So much for belief in free markets. You realise that there is nothing
>> that requires servers to install this, or cease using the old network?


>Note that the two things IRC really needs, end to end encryption and
>authentication, are not even on the list of "improvements" these people
>are working on.

I think that if you have authentication, what you wind up with is not 
really IRC-like.  I would like to see pseudonymous authentication (ie, 
each nick would have a key for signing and be able to prove they were 
the same person who last had that nick) but if you just say 
"authentication" these powers are going to think in terms of keeping 
out anybody whose True Name they don't know.  

I can see crypto helping keep things between the clients and the 
servers with an asymmetric encryption scheme; somebody would provide 
the server's public key when logging on, the server would use their 
public keys to send them stuff, and nobody could suss out the network 
packets.  Of course, people could still just run clients if they 
wanted to know what folk were saying, but with crypto they couldn't 
packetsniff to backtrack to the source very easily.


(Jim Choate Quoting Adam J Herscher)

,So what are my
>options at this point? Well, I can link to their network, or I can decide
>not to. If I decide not to, I will remain with a group of unwanted leaf
>servers with no hubs. 

So the question becomes, how difficult is it for someone to set up an 
IRC hub?  The answer is not very.  I've got the software on my SuSE 
box -- I shut it down when I was starting to harden it, but when I 
first connected it to the net it was ready to function as an IRC host -- 
and if I'd typed the names of other IRC hosts into a config file it 
would cheerily have acted as an IRC hub.  

The old-style IRCies are going to need to set up a few new hubs, but 
I don't think this is going to kick them off the network entirely. 
It's just going to create a new IRC-like protocol and convert some 
existing IRC nodes to run it.  

The danger of course is that programmers are going to abandon "normal" 
IRC protocol.  If they quit developing new software for IRC hosting and 
linking, or if the software for the new protocol is substantially easier 
to use and slicker, then people who make Linux distros are going to quit 
packaging old-style IRC hosts, people who keep download sites will take 
them down and put up "the new version", etc.

>It would indeed be unfortunate if all controversial IRC traffic ended up
>being carried by isolated IRC servers, akin to remailers, whose admins
>were under constant attack, and which came and went on a daily basis.

Ugh.  I think that's where this plan is pointing.

>I anticipate that if Efnext pulls off this "Conform or be
>Delinked" exercise, people will be setting their sights on Usenet as the 
>next thing that needs "fixing."

I anticipate that if IRC and/or Usenet are "fixed", then there will 
be a much stronger motive for people at large to create analogous 
protocols to IRC and NNTP with encryption and strong guarantees of 
privacy.  But they're going to have to be pure peer-to-peer 
protocols, so there is no "server backbone" that concentrates 
power in a few coercable hands.  Fortunately, I think the connectivity 
model is robust enough now -- it wasn't, back when NNTP was created -- 
to go fully peer-to-peer on netnews transfer.

Hmmm.  There may even be a niche in this new ecology for a network 
standards body composed completely of pseudonymous individuals, to 
help create and standardize network protocols for the underside of 
the net.  

				Bear







More information about the cypherpunks-legacy mailing list