Re: Persistent Documents and Locations

Larry Masinter (masinter@parc.xerox.com)
Tue, 22 Aug 1995 00:31:28 PDT


To: uri@bunyip.com
In-Reply-To: "Karen R. Sollins"'s message of Mon, 21 Aug 1995 08:21:18 -0700 <199508211521.LAA09705@lysithea.lcs.mit.edu>
Subject: Re: Persistent Documents and Locations
From: Larry Masinter <masinter@parc.xerox.com>
Message-Id: <95Aug22.003139pdt.2763@golden.parc.xerox.com>
Date: Tue, 22 Aug 1995 00:31:28 PDT

> 				If I let out a reference to
> something for which I have sole responsibility and that I intend to
> keep that way, then I have the option of causing that thing to be no
> longer available. Et voila!  the potential for dangling references.

The immovable reader meets the unstoppable author! (Or is it vice
versa!)  Publishers may wish to keep readers from keeping copies of
things, but can they?

> If one takes the position that dangling references are a "bad" thing,
> one is saying that this sort of situation is a "bad" thing.  Either I
> can never let out the references or the things can never be withdrawn.

I'm trying to avoid unnecessary value judgements. My position is that
dangling references are "bad" when they are unwanted by the parties
involved.

When dangling references are unwanted:

* A system of long-term preservation is necessary to prevent dangling
  references. 

  As long as references are to last longer than the lifetime of
  the author or copyright holder, some institutional means of
  preserving the digital objects associated with the information
  is necessary.

* A system of long-term preservation is sufficient to prevent
  (unwanted) dangling references to digital objects.

  Presumably the long-term preservation system would have some
  mechanism for cataloging the things it was preserving. This
  need not be of any global design or optimization, but merely
  both satisfy the URN requirements (long-term handle) and also
  the engineering requirements of the preservation system.

  Given such a handle, one could use it as or in addition to
  a reference.

I'm not claiming that this will solve all 'dangling reference'
problems. Services move as well as digital objects, and objects get
updated in ways that finding the latest version of something requires
additional infrastructure.

And I withdraw the claim that 'naming without preservation isn't
useful'. Of course it's useful, and the case where the author doesn't
want to relinquish control is just one instance.

But it is both necessary and sufficient for solving a class of
dangling reference problems, where documents disappear merely because
the author stopped maintaining the site.