PGP took this cautious approach with its elegant web of trust model, but
look at how well verisign is doing in the SSL and even PKI areas... I think
the worries may end up hindering the adoption of the tech if we go down
everything decentralized NNTP style (what happens when I deceide to spam the
NNTP group. From dejanews? Everyone under that feed get cut off... That's
rediculous.)
August
-----Original Message-----
From: Bill Kearney <wkearney nospam at gilman.edu>
To: 'cdindex nospam at freeamp.org' <cdindex nospam at freeamp.org>
Date: Tuesday, March 09, 1999 4:13 PM
Subject: RE: Reality Check and Ideas
> Sorry to piss in your bucket but
>
> NNTP is a flood fill scheme with no expiry other than time,
> no replication,
> no control of distribution pattern
>
> DNS is [buzzword hat on] a highly scalable replicating
> database with extensive
> caching including optional negative caches fault tolerance
> and load balancing
Ok, so take a protocol like NNTP and make it include a provision for
expirations. I don't see anything wrong with a flood-fill scheme. We're
talking a lot less data than news. No control or distribution pattern
sounds like a GOOD idea if you want to be sure the system can't be co-opted
into a commercially controlled venture like CDDB.
However, provisioning for some kind of "this query brough to you by X
company" might be worth considering. Realistically, if someone wanted to
'give' bandwidth to a server they might want something in return. Allowing
the clients to choose to use such a server is worth providing. I'm not
opposed to this mechanism if I can setup and use a server that shares the
data input/output load and doesn't require the advertising spam. Heck, I'd
be willing to leave the "brought to you by" in the query records if they
came from a commercial upstream server. In practice, however, I'd omit such
servers from my server search path. That is, if such a server existed. But
at least I'd be ABLE to do this. The current CDDB doesn't allow for this.