Hey,
I've been thinking. The problem is also that the centralized nodes are that they would
be loggable by others. So if you have new freenet user getting logged on at these
centralized nodes their ip could be logged and the freenet network be charted as a
result.
My penny and a half,
Chaviv
*********** REPLY SEPARATOR ***********
On 2-7-2001 at 3:36 Brandon Wiley wrote:
>> This is the bit where I like the Gnutella system a little better.
>>
>> The way Gnutella works is that you find an web-site containing the seed
>node
>> information. You download this and use it.
>
>This has absolutely nothing to do with Gnutella. It would be more accurate
>to say "This is how Gnutella users do it." However, this also happens to
>be how Freenet users do it. That's what inform.php is, a web-site
>containing seed node information. Although there is a difference in how
>that information is updated.
>
>The Gnutella community has pretty much explored all of the different ways
>that seed addresses can be propagated. The absolute best for users is to
>have defaults pointing to trusted nodes included in the distribution. You
>have to trust the distributors of the software before you use Freenet. So
>you can probably trust the nodes that they point you to. If you are weird
>and don't make that trust connection, you can override the defaults to
>point to your friend Bob's node instead. Letting the distributors assign
>trusted seed nodes is a pretty good amount of centralization. You're not
>adding a weak point. The existence of distributors of the software is
>already a weak point. If they take out sourceforge and hawk and Mr. Bad
>and Sebastian then they've pretty much screwed us. We might as well admit
>this when deciding on our seed node propagation system. Sure, other
>distributors could arise in the case of our demise. In that case the task
>of assign seed nodes would fall to them.
>
>However, this system begs the question of how the list of trusted nodes
>gets updated. There is the current inform.php model in which anyone puts
>their node in the list. There is the mostly similar model used by the
>Gnutella web sites in which they just run a node and report announcements
>that they get. Both of these are obviously attackable.
>
>The only way that I can think that we could have a list of trusted seed
>nodes where we know the seed nodes to be actually trust-worthy is to have
>the developers themselves run nodes that just do announcement. This of
>course does not scale very well. This is because trust itself does not
>scale very well.
>
>
>
>_______________________________________________
>Devl mailing list
>[EMAIL PROTECTED]
>http://lists.freenetproject.org/mailman/listinfo/devl
_______________________________________________
Devl mailing list
[EMAIL PROTECTED]
http://lists.freenetproject.org/mailman/listinfo/devl
