Regarding David Berlind's "DNS inventor says cure to net identity problems is right under our nose", http://www.business-standard.com/ice/story.asp?Menu=119&story=20692

The DNS system is but one case of a HUGE class of problems, that includes all directories, as well as product lists, etc. Resolving an identifier into a meaning. Mockapetris should google around subjects like EDI code lists. or recall the X.500 wars, and Microsoft's campaign of over 10 years to replace DNS with various directory schemes. Observe the requirements documents that gave birth to UDDI or ebXML RegRep. And observe how far THOSE have gotten. :-(

We all want convenience, robustness. But directory information is far too valuable to trust a single address resolver or to publish in a single monolithic form, accessible to all.

Clay Shirky writes about IM protocols, phone, fax, email, urls etc: "people have noticed how valuable it is to own a namespace..." http://www.shirky.com/writings/dns.html

We're already too dependent on the IP numbering scheme and DNS. DNS is a POS. It is a single global namespace, yuck, what if there are 2 people named "Smith"? Go away Mockapetris, and all engineers of top-down hierarchic control. DNS is great but is not the only game in town.

We need a healthy diversity and competition in addressing schemes and resolvers. We need more work on the applications layer of the stack, to give users complete, sovereign control over their end-to-end conversations. And we need work aligning the information vertically within the stack, so that the identity of the ULTIMATE owner of the conversation is not repeated and reified and abstracted and translated, fifty different times up and down the stack,

Todd Boyle http://www.ledgerism.net/P2Paxes.htm
contributor, ebXML Core Components Technical Specification



Reply via email to