comshaft strikes again

[Dan Ritter]

On Tue, Sep 23, 2008 at 11:49:34AM -0400, Bill Bogstad wrote:
> On Tue, Sep 23, 2008 at 10:01 AM, Dan Ritter <dsr-mzpnVDyJpH4k7aNtvndDlA at public.gmane.org> wrote:
> > leafnode is your friend.
> 
> Hmm, how does one find people willing to provide USENET feeds (as
> oppossed to access) for individuals?
> I had heard of leafnode, but didn't see where to get a feed.

leafnode can turn any reader-level access into a feed. That's
the appropriate thing to do for small sites (small being,
basically, anything under a hundred interested readers, these
days).

> My comment about bittorrent was intended (poorly) to suggest that some
> kind of distributed hash function (DHF) might allow a USENET
> cooperative using spare cycles/bandwidth/disk space to do something
> akin to the way that bittorrent makes centralized ftp servers much
> less important.  For compatibility with standard clients, a NNTP
> gateway to the 'cooperative' would be probably be needed.  Each
> node/gateway would cache copies of everything read locally via NNTP
> as well as it's share of the general content.  As more people become
> interested in a particular newsgroup/article itt gets replicated to
> more nodes in the network.
> Not sure if you would do DHF by newsgroup name, message-id or what...

I assure you that NNTP is more efficient than that. Also, that
binaries in Usenet are an abomination.

Don't re-engineer everything according to the latest fad. The
Usenet model is an excellent method of distribution given the
requirements:

- tens of thousands of nodes which generate writes
- latency is less important than completion
- not every node subscribes to every channel
- each node changes subscriptions daily
- each node supports 1-100,000 clients
- millions of reads, no updates

These are not the requirements for P2P.

-dsr-
-- 
http://tao.merseine.nu/~dsr/eula.html is hereby incorporated by reference.

You can't defend freedom by getting rid of it.