Re: Eprint versions and removals

From: David Goodman <>
Date: Tue, 14 Dec 2004 22:02:09 +0000

Marie Meyer wrote:

> How do I know that the self-archived version I am looking at is the same
> as the "official" version that appeared in the prestigious journal -
> that no important but subtle changes have been made?

Dear Marie,

I think a fuller answer can and should be given. You have no guarantee
except the integrity of the author and the alertness of the users.
Further, nobody actually knows the extent. It is clear that any problem
can be minimized by the editor requiring the peer-reviewers to clear
up any ambiguity and see to any necessary rewriting. The copy-editor
is responsible for the presentation, but should not have to correct the
science or rewrite the English. Inadequate editing will always produce
poor journals. Examining for differences of the sort you have in mind
requires much patient work by those expert in the subject--it is not a
job for automatic text-matching programs, as we are not concerned with
variants in spelling and capitalization. (Traditional bibliography
refers to such aspects as accidentals. Before copy-editing, the printer
did as he pleased. ) Stevan has said he expects differences in content
to be very rare; I am less confident. Some day someone will do the work,
and we will then have factual knowledge.

Authors vary somewhat in what they consider the preferred version. I
will, for example, soon have an article in Learned Publishing. I have
not posted the preprint because subsequent events in the UK have caused
me to rethink my numbers. The editor was wise enough to alert me, and
with his help I managed to get a note to that effect inserted in proof,
but I could not redraw the figures. When I have finished redrawing
them, I intend to post it as a revised version of my article, making
reference to the original, which will have my original estimates. (LP
has a very generous copyright agreement that permits me to do this.) I
could have hid this, and just posted my revised version as if it were
the original. That would give the sort of situation you describe, and
I think all of us would consider this wrong. With other --possibly
better-- manners of publication, the journal itself could have made the
substitution, keeping both versions on record, in the manner of arXiv.

Even before electronic journals, authors have been known to improve
their data in a subsequent publication, without mentioning that they have
earlier published it in a less accurate form. This is fundamentally the
same situation, and the protection against this is the vigilance of the
peer-reviewers and the editor. As we all know, this needs back-up from the
readers, who are really the final peer-reviewers, and will detect what the
official peer-reviewers may miss, just as they have detected other frauds.

Our hope is to get an OA publication system that will build on the
strengths of the present one, and be fully equitable and accessible as
well. It would be good if OA could improve other aspects, such as accuracy
and speed, but it should not be judged against the standard of perfection.
The criticisms that have been made against the various forms of OA,
if applied to the conventional system, would reject it soundly.

You are however right that there will inevitably be weak points, and
all scientists have a common interest in improving them. The more OA,
the more who can join in.

Dr. David Goodman
Associate Professor
Palmer School of Library and Information Science
Long Island University

-----Original Message-----

> In the OAI/OA world the reader discriminates *exactly* the same
> way the reader selects in the "reader-pays world": On the basis of
> the name of the journal in which the article appeared. Publication
> name is one of the OAI tags.
> MODERATOR'S NOTE: Again, to save time and space, please see:
>Pertinent Prior Amsci Topic Threads:
> "Text authentication, plagiarism, and degree-of-authorship"
> "Eprint versions and removals"
Received on Tue Dec 14 2004 - 22:02:09 GMT

This archive was generated by hypermail 2.3.0 : Fri Dec 10 2010 - 19:47:44 GMT