From: Gregory Alan Bolcer (firstname.lastname@example.org)
Date: Sat Apr 22 2000 - 13:46:06 PDT
Adam Rifkin -4K wrote:
> Tom wrote:
> > Napster's central server tech is its death, pure and simple.
> So what the heck are they doing looking for a VP of Biz Dev???
> Clearly they are planning for the longterm. They got NINE MILLION users
> in SIX MONTHS. This has gotta be the fastest rampup in history.
> > Read the specs on the protocol, lots of gnutella is http:)
> They use a "mini HTTP" protocol for downloading files from a server, but
> they don't use HTTP for anything else. Imagine the possibilities if
> they had used a MAGI server or a full HTTP+DAV server on every desktop
> instead: worldwide distributed filesystem, baby. [Insert random bitter
> Beberg comment about distributed filesystems here. :]
Here's what I want from Napster. Why don't they just cache the metadata
at the cache site? What the world needs now is a Dublin Core for
Mp3s. The other item it needs is to build up a body of knowledge
about the songs other than embedding the metadata into the name of
the file. Right now I can take random trash, put an MP3 extension on it,
put it up on my Napster server as "Nirvana/The End" and have everone
go and believe that the copy of Adolescent's "Kids of the Black Hole"
is really Nirvana covering a Doors song live in Gent, Belgium, in 1991. 
How do you know other than the community of Napster users have adopted
the same naming convention that works most of the time?
Content Intellectual Property Instantiation
----------- --------------------- -------------
Title Creator Date
Subject Publisher Format
Description Contributor Identifier
Type Rights Language
Shouldn't all that stuff be properties on the file or even Xpointers  to
metadata about other files/links?
I want a recreatable event stream that in addition to the following metadata,
there's properties like influences, albums, band members and an associated
dated event stream such as "Joe left the band for NewBand", or "The group broke up",
"they played a concert in Detroit and the WebHeads opened for them". If you
have a bunch of these little event snippets lying around all over the place,
you can use a search-like engine to recreate the whole history of the band including
all their influences, which songs they covered and collaborated on, who was in the
band at a particular time. You start to make headway on an Internet-scale, traceable,
body of knowledge that includes IP issues. It's sort of like a cat's cradle in that
if you pick it up by the band member string, you search on Jonny Bandmember and get
precise metadata search on the whole history of bands he's been in, songs he's written, dates and times
he's left, where he's played, etc.; if you pick it up by the band string, you get
all the band information views.
The abundance of music, the dynamic nature of the music industry, band members,
music writers, and popularity will always gurantee that no single person, place,
or computing thing will ever be able to manager all of these things into one
big central repository that's complete (although the particular subset may
Anyways, just some random musings...
> Gnutella could really use a chat system too. It's really lame that
> people have to talk with each other using search queries.
> While I'm making requests, I'm still waiting for the Gnutella/Freenet
> gateway. And, what the heck, let it interface with www.scour.net too...
> let's really blow our brains out.
> Would I be a good messiah with my low self esteem?
> If I don't believe in myself is that blasphemy?
> -- Bloodhound Gang, "Hell Yeah"
-- Greg Bolcer email: email@example.com web: http://www.endtech.com work: 949.833.2800 cell: 714.928.5476 fax: 603.994.0516
This archive was generated by hypermail 2b29 : Sat Apr 22 2000 - 13:46:41 PDT