On Wed, 12 Feb 2003, Robert Spindler wrote:

> This exchange... illustrates the difficult tension archivists feel these
> days between preservation and access. The scholarly research community
> has profound opportunities to improve the speed and availability
> of very current research results through electronic publishing and
> self-archiving. These things benefit the scientific community in very
> direct and immediate ways...

This is a fair statement of the current situation regarding digital
content in general. But it misses the point about the basic fact that the
refereed-research corpus (20,000 refereed journals-full) is a
*toll-access* proprietary corpus today. *That* is the corpus with the
access problem, and it is also the corpus with the preservation
problem.

The institutionally self-archived corpus is merely a *duplicate* of a
small portion of the above corpus -- a partial backup, if you like, for
immediate use. It makes no sense to slow the momentum for providing this
immediate back-up, now, by saddling it with the preservation concerns
that should be addressed to the primary (toll-access) corpus! (This
is the token that has not yet dropped for the well-meaning digital
library/preservation community -- with respect to the very special case
of the self-archiving of the refereed research literature.)

Having said that, one must add that it is very likely that the
growing self-archived "backup" secondary-content may actually drive
and accelerate the momentum toward the preservation of the primary
corpus. For as researchers come to rely more completely for their daily
access on the open-access versions (all linked by their OAI-compliance
and interoperability), they will not brook any loss of access, for any
reason, not for a moment. The collective eyes of the research community
will be relentlessly trained on this precious resource of theirs, and
corpus-wide access-failure "alerting" systems will no doubt be created,
to act on the first sign of access-failure.

But none of this can come until self-archiving occurs, and generates the
open-access versions of the toll-access content, and research-community
reliance on them. Until then, the preservation community should focus its
attention on the (real) preservation problems of the primary, toll-access
corpus, and not do or say anything to slow or discourage the growth of
its self-archived back-up in any way.

> On the other hand there's the back end of research replication,
> criticism and revision, citation tracking/analysis and the history
> of science.  Archivists think about this stuff quite a bit, perhaps
> more than the scientific community does at this time since perhaps
> they have not experienced significant/relevant loss so far.

Researchers think about and practise replication, criticism and revision,
citation tracking/analysis and the history of science, plenty! There
is no trade-off between steps taken to secure immediate open-access
through self-archiving and all these other benefits. They are additive
(perhaps even multiplicative). And new benefits, once savored, tend to
become entrenched...

> We used
> to think about catastrophic loss events, we know now that loss is
> likely to be more subtle - the gentle corruption of over time from
> software incompatibilities, character set incompatibilities, loss
> of formatting, addressing failures (its out there but in a place
> you can't find), linkage failures (between digital images and their
> metadata for example), hidden viruses. Clifford Lynch has noted that
> our tools for detecting corruption are very blunt. The subtlety of
> loss makes preservation advocacy very difficult because loss is not
> catastrophic until it reaches a critical mass.

The critical mass that is needed is a critical mass in self-archived,
open-access content, duplicates of the existing toll-access content of
the 20,000 refereed journals. That critical mass of content will in turn
generate a critical mass of users (most of them both the authors and the
self-archivers of the literature itself and also one another's
readers). It is the daily usage and vigilance of that critical mass of
increasingly reliant users that will drive the real preservation efforts,
not the fragmented initiatives on the fragmented literature today, most of
it inaccessibly behind proprietary toll-booths. But for this to happen,
it must not be rendered still-born by insisting on preservation before
content-provision! The content is already there, behind the toll-booths.
Focus preservation efforts on that for the time being, while giving its
self-archived open-access backup the chance to reach critical mass!

> One of the things I've been trying to pay attention to in this
> environment is: "What advice should we be giving to document creators
> to help them minimize the potential for loss?" Can we influence the
> process of document creation to maximize the potential for *real*
> archiving without slowing the dissemination of research? The OAIS
> reference model is very helpful in thinking about these things,
> specifically in terms of submission of one or two or several forms
> and formats of the content (the archival information package, the
> distribution information package, etc...) Simply uploading files
> will not suffice if long term preservation is desired.

What authors -- still reluctant and uncertain about self-archiving,
and mostly not doing it at all  -- need now is not *more* demands on
their time and resources, but the simplest, most minimal ones possible:
"Pop your current word-processor text (TeX or PDF or PS or HTML or XML
or a Word-generated version of one of these, plus the Word document too)
into your institutional Eprint Archive along with the metadata (author,
title, date, etc.)." That's all. (Forget about OAIS for now! The
OAI-compliance of the Eprint Archives is enough for now.)

> The thread about "toll-access" content vs. self archived content is an
> important piece. Stevan places a great deal of trust in the commercial
> publishing industry for long term preservation of the "toll-access"
> content, and yet publishers seem unlikely to make the timely and
> continuing preservation actions necessary to retain electronic
> content unless there is sufficient market revenue to support the
> preservation costs.

You miss the point! I don't place trust in the toll-access sector. I
simply point out that the primary content is exclusively in the
toll-access sector! So preservation efforts should be directed *there*
and not at the budding attempts to get immediate-access to it by
self-archiving open-access back-ups! If there is a current, urgent,
preservation problem for this literature, address it at its source, not
at the site of an attempted solution to *another* problem, which is
immediate *access* to the toll-access literature, not its
*preservation.*

> I can still hear Kevin Guthrie of JSTOR asking
> the group at CNI in December, "Which of your institutions is willing
> to help fund the public good?" (paraphrased regarding who will pay
> the cost of preservation?)

And because no one was willing to shoulder the cost of the preservation
burden for the primary literature itself, you think it is a good idea
to discourage nascent access-efforts by loading the preservation-burden
onto them?

> Another perspective on this thread is
> that there may be significant differences between the self-archived
> version and the commercially published version that demonstrate the
> influence of reviewers, new research by others, etc.  Both versions
> may indeed be archival!

Authors are free to self-archive as much as they wish. Some think
the unrefereed preprint is enough. Most think the refereed postprint
is necessary. Many also self-archive drafts in between, as well as
post-publication corrections and updates. The OAI-compliant Eprints
Archives are set up to track the versions.

But this is all misplaced fastidiousness! The canonical version is
"archived" as well (or badly) as it always was, in the publisher's
proprietary corpus. We are talking here about fulfilling the needs of
those wpuld-be users whose institutions cannot afford access to the
toll-access version. Are you suggesting that they should be deprived of
any access at all, if it cannot be guaranteed that they have the verbatim
final text at this time?

See the Preservation, Authentication, Corruption, Certification, Version
Control, and perhaps even the Peer Review and Copyright FAQs of the
self-archiving initiative to see how these endlessly repeated worries
are without any grounds:

http://www.eprints.org/self-faq/#1.Preservation
http://www.eprints.org/self-faq/#2.Authentication
http://www.eprints.org/self-faq/#3.Corruption
http://www.eprints.org/self-faq/#5.Certification
http://www.eprints.org/self-faq/#23.Version
http://www.eprints.org/self-faq/#7.Peer
http://www.eprints.org/self-faq/#copyright1

> In the end its pretty clear that unless the scientific community
> values preservation of their work at some level close to the value
> they place in fast dissemination, the archival perspective will be
> very difficult to sell.  We'll need to make preservation as seamless
> as possible if we're going to expect the scientific community to
> participate in saving their own memory.  Retrospective repair is a
> fool's game no one can afford.

The corpus to repair right now is the toll-access corpus. The main
problem facing the open-access initiative now is researcher unawareness
and sluggishness in doing what is in their own interests. The existing
free self-archiving software packages have taken enough of the simple,
obvious preservation-ready steps to make them ready for immediate,
full-speed self-archiving, now. Once the critical mass of open-access
content is self-archived, it can be retro-fitted for more rigorous
preservation. And the designers of the free self-archiving software
packages are always ready to listen to constructive and realistic advice
on preservational features. But what is *not* needed at this time is
any implication that self-archiving should worry or wait, even for a
microsecond, for preservational reasons.

Stevan Harnad