> How are gateways chosen?
Someone volunteers.
> What is to stop gateways from being taken down (using the "flaw" that you
> talk about) isolating a potentially large group of nodes?
Nothing. A gateway is a public node. Currently there is nothing to stop a
public node from being taken down. The cluster will be isolated, but not
shut down. The alternative is that they all get shut down. Some trading
can still occur just within the cluster until they find a new
gateway. Also, there can be multiple gateways for a cluster, so shutting
down one gateway wouldn't necessarily isolate the cluster.
> How do people reliably find another Freenet node which they trust, and
> which can trust them?
Since this must occur out-of-band, the particular mechanism is
situational. The same ways that you form any sort of persecuted group. You
start a small cluster with people you know. As you meet more people that
you think you can trust you add them and hope that none of them are
agents. If you can't find people to trust then you can take your chances
with running a public node or you can give up.
> Why won't the gateway become a bottleneck if all requests for information
> not stored within the local cluster must pass through it?
First of all, it is likely that because of the isomorphism between a group
of people that know and trust each other and the cluster, it is likely
that there will be some overlap between requested files. This means that a
greater percentage of requests will be serviced by the local network than
in a normal public network.
However, ignoring that, the gateway acts as a bottleneck going out of the
cluster but not going into the cluster. (I'll post a separate post to
substantiate this claim.) So yes, from inside the cluster if you want to
request something outside of the cluster, you have to go through one of
the gateways. There can of course be multiple gatways. But it is of course
an issue that if you only have a few gateways and a lot of nodes in the
cluster that the gateways will become overloaded and things will get
slow. If that happens, you can always run a public node or try to find
another gateway to balance out the cluster's load.
It is important to note, however, that while it is possible that being in
a cluster could suck (in which case become a public node if it sucks so
much that it's worth the risk to you), a cluster will not degrade the
performance of the public network. People in clusters should be allowed to
degrade their performance if they feel like it but of course should not be
able to degrade the performance of the public network, which they won't.
> Explain how someone efficiently attempts to do a DH key-exchange with
> 65,000 ports on a machine which could be at the end of a 56k modem?
If you have a lot of machines and a lot of bandwidth dedicated to just
this purpose, you can do this. I don't think this is an unreasonable
resource for an attacker to have if we're considering governments to be
potential attackers.
> No shit. The point is that I think that you are wrong. Also, you
> suggested this before - what was the general opinion then - I seem to
> recall that it resulted in Oskar making some disparaging remarks about
> your understanding of how Freenet works. Can you remind me when you last
> suggested it so that I can check the archives?
Everything I say results in Oskar making some disparaging remarks about my
understanding of how Freenet works. Let's critique my proposal further and
not my character. And I have no idea when I last suggested this.
_______________________________________________
Freenet-dev mailing list
[EMAIL PROTECTED]
http://lists.sourceforge.net/mailman/listinfo/freenet-dev
