[OpenID] why is xri so obtuse?

Drummond Reed drummond.reed at cordance.net
Tue Jan 2 03:46:08 UTC 2007

>> Drummond Reed wrote:
>> 1) XRI is an open, public, royalty-free identifier standard developed at
>> OASIS. It is every bit as free as the URI or IRI specifications from IETF
>> (http://www.ietf.org/rfc/rfc3986.txt and
>> http://www.ietf.org/rfc/rfc3987.txt).
>James A. Donald wrote:
>Anyone, including Microsoft, can declare open public royalty-free 
>standards for interfacing to their proprietary for profit services, and 
>their proprietary for profit software, and with great regularity 
>Microsoft does exactly that.

While I don't dispute that, James, it doesn't seem fair to lump everyone in
the world who publishes royalty-free standards in with Microsoft. There are
hundreds of open standards that were created simply to solve problems on the
net and not accomplish proprietary lock-in. RSS and microformats are just
two examples.

As Kaliya's post explained, the long lineage of the XRI specs started with
proprietary technology and then was all turned over to XDI.org (originally
XNS.org) as an international public trust organization. Yes, this means
XDI.org is similar to ICANN, but it was started independently of ICANN and
with a great deal of forethought about what it would mean for the community
of all users to actually OWN AND CONTROL the root of the community. You can
check out www.xdi.org for more details about its roots and goals.

>> 2) The use of XRI does NOT require the use of XRI global registry
>> Although such registries are supported by XRI architecture, they are only
>> one option. You can also use XRI infrastructure with a) DNS names, b) IP
>> addresses, or c) complete p2p identification and resolution as described
>> XRI TC co-chair Gabe Wachob in his blog entry at
>> http://blog.wachob.com/2006/12/xri_is_a_networ.html. 
>James A. Donald wrote:
>Which says that anyone can be a root if others treat him so - but every 
>XRI name space requires a single root, who thus has monopoly control of 
>that namespace.

Is that not true of all namespaces? That's the core concept of delegation --
whoever controls a namespace at one level can delegated control of the next
level, and so on. Everything from file systems to DNS work that way.

>But what we really want is a namespace where no one is a root, or a 
>single namespace that is multirooted with as many roots as people who 
>feel like being roots, and are treated by other as root.

That's exactly what Gabe explains can be done with XRI architecture using
XRI cross-reference syntax. Did you read his post at

>Now in fact it is hard to implement such a system in a user friendly 
>manner, but XRI is *not* such a system.

If, after reading Gabe's post, you still believe that's true, could you
explain why you believe XRI is not such a system? It would be helpful to me
to understand.



More information about the general mailing list