Re: How long should items be able to be accessed from a repository?

From: Arthur Sale <ahjs_at_OZEMAIL.COM.AU>
Date: Sat, 11 Mar 2006 09:17:09 +1100

There are a couple of issues/arguments in this posting that deserve
teasing out further.


Arthur Sale


> -----Original Message-----

> From: Repositories discussion list

> Behalf Of Danny Kingsley

> Sent: Friday, 10 March 2006 22:37


> Subject: How long should items be able to be accessed from a
repository? How long is

> long term?


> How long is long term?


> There is currently some discussion (on this list) about how long items

> should be able to be accessed from a repository. It is an interesting

> question, and opens up larger ones such as what responsibilities do

> have for ensuring the archiving and perpetuality of papers they have

> published in a digital-only format? If libraries are simply licensing

> subscriptions (which are for a finite period), and not maintaining a

> copy in their libraries (the LOCKSS principle) then long-term archiving

> issues must be addressed by the publishers.


> The logical forward argument from that is that the open access copy of

> paper sitting in a repository is only needed until the electronic copy

> the publisher's version goes open access (as increasingly they are

> some embargo period, often 1-2 years, sometimes as short as six months

> the publication date).


It is important to note the huge fallacy in this argument. It is
unnecessary to have open access to have archival Preservation. Even
journals that *never* make their journals open access still have
Preservation responsibilities, and they generally meet them. The archival
Preservation need is satisfied by every journal article as soon as it is


The emergence of delayed access (making historical articles freely
available only after a period) is to be strongly deprecated. Six months
is still an unconscionable delay in open access. Two years is as good as
infinity in most fields. Again, except for providing a lazy way of
providing long-term access by previous subscribers to old issues/papers.
It has nothing to do with open access.


> If that is the case, 'long-term' responsibilities of an institutional

> repository only exist for this short timeframe, and energy expended on

> longer archiving roles would be wasted.


It is wasted anyway. Open access repositories containing mainly published
articles have no long-term Preservation roles. This may not be true of
repositories with a larger "Digital Library" agenda.


> But this cuts down to the even bigger question of what the

> repository exists for. Even once the decision is made to separate the

> of a 'digital library' and an 'institutional repository', there is a

> required distinction. If we define the institutional repository as

> soley for preprints, postprints and supporting datasets, there are two

> choices for the required length of archiving, depending on the purpose

> the repository.


> The first instance is where the institutional repository is simply in

> as a tool to facilitate open access to the publication output of the

> institution. In this situation, the items in the repository are only

> required to be there and accessible until they become freely available

> time, as publishers open up access to back copies of publications. This

> even more pertinent in disciplines where papers have a short half-life,

> as biomedicine. Some papers are obsolete within a very short

> which means their citation rates will stop altogether, obfuscating the

> for them to remain in the repository other than as an historical


This ignores the persistence of URLs, which citers are entitled to expect
for a much longer period. Broken links are the bane of every researcher,
and of citation tools.


> The second instance is where the repository is used as a record of

> output by the institution for reporting purposes, such as for DEST

> in Australia. This is a likely scenario in the future, as one way of

> ensuring material is deposited in the repository is to tie the process

> that of grant applications.


This is not a way of "linking process with grant applications", but a way
of satisfying governmental reporting requirements (and formula-driven
institutional funding allocations). As such it will have the effect of a
"mandate" in Australia in the long term: non-compliance will mean lack of
research infrastructure funding to the school/university, so you had
better comply. The same is true in the UK with the RAE, and elsewhere.


> Again, the question of the definition of the institutional repository

> becomes essential before decisions can be made and time invested in

> to solve problems that might not really exist.


And this is precisely right. If a definition of what a repository is
diffuse and poorly thought out, expect (a) lots of wasted time arguing
over non-essentials and (b) a blow-out in costs of 2x to 5x at least.


Arthur Sale



> Danny Kingsley

Received on Sat Mar 11 2006 - 04:01:07 GMT

This archive was generated by hypermail 2.3.0 : Fri Dec 10 2010 - 19:48:15 GMT