On Sun, 3 Mar 2002, Arthur P. Smith wrote:
> I hope Andrew Odlyzko
> was misquoted on the "do the same thing for $100,000"! Perhaps he'll
> explain himself...
I believe he was referring to archiving it (actually, 20,000 papers of
it), not to publishing it.
It is agreed that the archives are parasitic on publisher-funded
peer-review until/unless that is no longer covered by subscriptions
(and hence needs to be covered per outgoing paper out of
subscripton-savings). Let us also agree that the rest of the added-value
(on-paper version, publisher's PDF, other enhancements of the
peer-reviewed draft) need to be unbundled from peer review and
archiving and sold separately, so it can be tested whether they still
have a market.
So, what Andrew said was: $100,000 for archiving 20,000 papers
(i.e., $5 per paper) + $200-$500 per paper for peer review + whatever
the add-ons (paper, PDF, enhancements) cost.
> Anyway, on a somewhat related topic, I have a question for Stevan
> and the others who propose continuing to use journal publication
> as a stamp of approval on the free literature: What mechanism
> is proposed for ensuring the validity of the "published in ... journal"
> stamp? Currently the arXiv approach is a bit of a mix, but I believe
> mostly the journal publication information is supplied by authors
> rather than by the publishers. Is this a potential problem?
I will put it very simply: There are many ways that this problem, if it
is a problem at all, can and will be solved. Much more significant is
the fact that this obviously surmountable and far-fetched "problem"
should be raised at all (other than as a technical issue for the OAI
committee). I think raising it is another symptom of the Ptolemaic
extremes one must go to in trying to save the Gutenberg cost-recovery
model for this anomalous literature.
http://www.eprints.org/self-faq/#2.Authentication
http://www.eprints.org/self-faq/#3.Corruption
> We've
> talked a bit about the issue of minor differences between the published
> and "arXiv" versions - but what about the potential for abuse? If the
> information is author-controlled then an author could easily pad
> his "publication" list with invalid references to a few obscure journals
> that almost nobody subscribes to any more. Or perhaps even
> "valid" references to papers with slightly or completely different
> author lists (the published version missing this particular author...),
> slightly different titles, etc... The traditional argument I believe is
> that such abuses would be caught, but would they, if most users trust the
> arXiv rather than going to published journal sites? And even if
> caught, could it take years, long enough for a malfeasant author to
> accumulate significant grant funding, research salary, etc.?
I leave it to the reader's imagination, noting only that one could have
raised the same worry about the potential for abuse in switching from
barter to "gold" or currency or credit-card payment or online-payment,
from eye-witness oral interaction to paper or telephony, from mail to
email, etc. etc. I would note only that counterfeits are much more
easily detected by online digital means than by making one's feet do the
walking. (And that scholarly rigor will vary from individual to
individual, as it always did, and cumulative research will continue to
be self-correcting. And let's trust the rigors of hiring/salary
committees to fend for themselves...)
> The ultimate solution is to ensure that the journal references
> are placed by responsible parties - either the journal publishers
> themselves, or perhaps secondary abstracting/indexing publishers
> (though those can also be guilty of egregious errors in their databases),
> or perhaps by groups of established and authenticated research libraries.
> Would journal publishers actually be motivated to publicly place
> even that much data in a free system? We've done some things along
> these lines (various agreements in astrophysics and bio-medicine for
> example) but normally we expect some sort of payment or quid
> pro quo for exchange of our publication database information with
> another party. What's the benefit to publishers here?
Authenticating the publisher's certification (i.e., the fact that such
an article by that author did indeed appear in such-and-such a journal)
is a metadata question. We've already noted the quid-pro-quo for peer
review ($200-$500 per paper). How much of a surcharge per paper do you
reckon the added service of certifying this certification should
amount to?
> Would there
> be any benefit for secondaries? And as for the third option - would
> research libraries be up to the tedious data entry tasks (SLAC's SPIRES
> has done this for high energy physics, but I'm not aware of other
> examples)?
SPARC does it centrally, for the whole corpus. Institutional Eprint
Archives would only need to do it locally, for their own contributions
to the corpus.
> Or is just allowing authors to insert journal publication data sufficient?
For, now, for sure. Once it's (1) all up there and free for all through
self-archiving, and (2) we have a chance to see whether the transition
to downsized, open-access journals, funded by submission instead of
subscription, actually takes place (it might not, and a "double
economy" might stably co-exist, if Arthur is right, and all the other
add-ons could continue to have enough of a market-value to pay for it
all), we can find plenty of ways to optimize it.
Stevan Harnad
Received on Mon Mar 04 2002 - 11:16:06 GMT