On Thu, Apr 26, 2001 at 01:28:26AM +0300, Kalle A. Sandstrom wrote:
> 1) Node impermanence. The network may be large (I've had 50-HTL requests
> go unanswered with a timeout - this has to be a sign of *something*...),
> but if a large percentage of the nodes are run by Napster refugees (i.e.
> running Windows, a 1-minute attention span, "this sucks! I can't download
> anything!", kill node), data and references are bound to disappear from
> the network faster than they appear. I have no solution to this problem;
> perhaps a larger number of really persistent, well-connected nodes would
> help.
I think you've hit the nail on the head.
Perhaps instead of encouraging everyone who drops by the Freenet homepage
to run a node, curious visitors should be encouraged to download a client
and discover whether Freenet is useful to them. If they find that it is,
they should be encouraged to run a node and give something back to the
network, *provided* that they have an OS that doesn't go down every 8
hours, an ISP that lets them stay connected for more than 2 hours, and a
high-bandwidth connection.
(I'm not saying these criteria can or should be enforced, but there's
such a thing a politely turning people away - ask Oskar to show you.)
The idea of turning people down when they offer to run a node may sound
elitist, but it seems like what Freenet needs right now is more
stability, not more nodes at any cost. Several people have mentioned that
they have multi-gigabyte nodes that aren't full. I suggest we stop
encouraging casual users to run nodes until we really need the space.
Michael
_______________________________________________
freenet-tech mailing list
[EMAIL PROTECTED]
http://lists.freenetproject.org/mailman/listinfo/tech
