[2627] in cryptography@c2.net mail archive
URNs heirarchical vs decentralised
daemon@ATHENA.MIT.EDU (Adam Back)
Mon May 4 11:19:47 1998
Date: Sun, 3 May 1998 11:52:13 +0100
From: Adam Back <aba@dcs.ex.ac.uk>
To: eternity@internexus.net
Cc: cypherpunks@cyberpass.net
TAZ servers[1], and eternity USENET virtual URLs[2], and URNs[3] are
all ways of creating indirection, allowing URLs to move, or to be more
conveneintly quoted if they are long (being encrypted in the case of
TAZ/Rewebber and URNs also)).
[1] http://www.cs.berkeley.edu/~daw/cs268/
[2] http://www.dcs.ex.ac.uk/~aba/eternity/
[3] http://www.acl.lanl.gov/URI/URC_proposal/public.html
TAZ servers and URNs serve the same function in providing an
indirections. However in [3] the author plans to add other less
desirable functionality such as copyright protection, access control,
all in a hierarchical, DNS like system.
It seems that URNs should be an opportunity to build in eternity like
functionality, in that proposal[3] above allows a URN to map to
multiple URLs.
The designer are mostly thinking in terms of heirarchical design,
basically duplicating the DNS system for URN -> URL mapping, rather
than domain name to IP address. Also building in fascist copyright
protection, and top down hierarchical control. (ie the government
decides it doesn't like your content, they pull your URN, some
megacorp decides that your URN conflicts with their trademark, they
pull your URL.
Also the designer talks about Seals of Approval (SOAPs) signatures
rating a web page, and Author Institutions, implying authors must
belong to some heirarchically defined institution in order to publish,
and _still_ tying pages to specific hierarchically controlled
locations even if there is an indirection.
I think this is an area that needs some attention, the IETF have I
think, or soon will have a working group on it. It must be subverted,
or ensured to be flexible enough to allow decentralised URN spaces to
thrive in place of their heirarchical world view. One danger would be
for example if the decentralised URN space were a sub division of URNs
by being given a organisation of `alt' like USENET alt groups. If
lots of content which governments and megacorps find objectionable are
then published in the alt URN heirarchy, they have ability to pull the
whole heirarchy, or impose controls upon it.
This must be avoided by ensuring that sufficient parts of the system
is decentralised.
I like the idea of first come first served name ownership, first
person to publish the hash of a URN gets the URN, this can be
decentralised (with a reliable network time stamping services).
Eternity USENET URLs are based on first to publish virtual URL hash.
I think Eric Hughes was/is working on a domain system based on hashes.
I would be interested to hear comments from people who have read the
URN proposals in more detail on the above, and also on approaches to
build decentralised name spaces resistant to attacks by the rich and
powerful by design.
Adam
--
print pack"C*",split/\D+/,`echo "16iII*o\U@{$/=$z;[(pop,pop,unpack"H*",<>
)]}\EsMsKsN0[lN*1lK[d2%Sa2/d0<X+d*lMLa^*lN%0]dsXx++lMlN/dsM0<J]dsJxp"|dc`
