[tdwg-tag] LSID Sourceforge URL & LSID Best Practices

[Jonathan Rees]

On Sun, Aug 30, 2009 at 12:29 AM, Bob Morris<morris.bob at gmail.com> wrote:
> On Fri, Aug 28, 2009 at 11:52 AM, Jonathan Rees<jar at creativecommons.org> wrote:
>> ...
>> The fact that ICANN and DNS work as well as they
>> do prevents anyone from working on an administratively decentralized
>> alternative.
>>
>
> Umm, I would say that DNS is a giant success story about
> administratively decentralized technology, but my parsing of this
> sentence makes me believe that you think it is not administratively
> decentralized but should be.  I suppose only the TLD servers have
> their DNS records administered of necessity by a single agency, and
> those provide substantial redundancy.

Sorry I was not clear. Yes, DNS/ICANN is a success story for
decentralization. I was not referring to the system as a whole but
rather to individual domain names. If the owner of a domain disappears
or reorganizes, then all users of its URIs are screwed - the
well-known 404 problem. This phenomenon is what I've been calling an
"administrative single point of failure" resulting from
"administrative centralization", in contrast to "technical single
point of failure". No amount of technical replication can address this
vulnerability. Just saying that a URI is "persistent" does not make it
so, and we know that domains go south in spite of the best intentions
of those who originally create and disseminate its URIs, and in spite
of the availability of technical replicas (at other locations) of the
data that users need.

The only alternative to DNS/ICANN is some alternative to it (sorry) -
say, if domain D goes away but a copy of the needed information exists
at E, then configure clients somehow to resolve D to E instead of to
what ICANN tells you. This is what I've been calling "administrative"
redundancy, which has a distinctly different character from mere
technical redundancy. My point was just that even though such consumer
choice would be wonderful, in principle, and resembles the way that
historically robust systems such as the Linnaean system and libraries
work (and that would be required in order for many kinds of URN to
work), it is a fantasy - it's very unlikely to come about, because
DNS/ICANN works so well. (Same argument applies to handle system.)
Consumers are left with no power, and putatively-persistent-URI
creators, in selfless service to consumers, have to voluntarily take
on the burden to just "try very hard" to make domains used in URIs be
well-behaved in perpetuity.

Jonathan