On Wed, 15 Nov 2006, Jan Velterop wrote:
> What Beckett and Inger do amounts to market research. A sound practice
> in any line of business, and always flawed, because what is measured is
> intentions. What Harnad wants is an analysis afterwards, dismissing
> market research, saying that it is not the same as evidence (though it
> is, albeit of intentions).
> That is also flawed, in that it amounts to destructive testing.
I think Jan has not understood my critique at all. The profound
methodological flaw I pointed out was not that the PRC survey was asking
about librarians' intentions, but that it was asking about the *wrong*
intentions, in the wrong way. As a result, the librarians' expressed
intentions give us no indication whatsoever about the causal effect of
self-archiving on cancellations: neither about (1) whether there will be
such a cancellation effect at all, nor about (2) how much self-archiving
would cause a cancellation effect, nor about (3) how big such a cancellation
effect would be, nor about (4) how soon such a cancellation effect
would happen, if it were to happen.
All we have from this study is the confirmation of a banal fact that
anyone could have stated in advance: All things being equal, everyone
would prefer a free product to a priced product and would prefer the
product immediately upon purchase, rather than after a delay. (I invite
anyone who thinks otherwise to explain clearly to everyone how the PRC
survey answers any of questions (1)-(4).
The PRC survey, to repeat, did not ask acquisitions librarians about
what they think they would *cancel*, in exchange for what, under what
conditions. It asked what product they would prefer to *acquire* over
what product (where among the properties of some of the "products"
was the fact that they were free!).
And the author's final, peer-reviewed draft of a journal article,
self-archived by the author to make it free for all, is not a "product"
that acquisitions librarians acquire (or cancel!) at all. Acquisitions
librarians acquire (and cancel) journals -- priced journals (i.e.,
not OA journals either!).
The "Share of Preference model," in other words, is trying to generate
cancellation predictions by comparing apples and oranges. We get from
it not a hint of a reply any of the relevant questions to which we do
not currently have an answer:
(1) Will self-archiving generate cancellations?
(2) How much self-archiving will generate cancellations?
(3) What and how much will be cancelled (and in favour of what)?
(4) How soon will it be cancelled?
The answer to all four questions before the PRC study was: We do not
know, because we have no data. The answer after the study, to all four
questions, is exactly the same.
(For those who felt that our prior knowledge -- that everyone would
rather have something for free rather than paying for it -- already
answered (1) - (4), they too learned nothing new from this study that
they did not feel they already knew before!)
> Surely it must be better to try and find out
> beforehand if there is a(n entirely logical) change to be expected in
> the climate amongst librarians (the ones who would pay, after all, in a
> self-archiving mode), than to find oneself in an irreversible situation
> struggling to cope with the consequences. It is a question of whether
> one applies the precautionary principle or not. One thing is clear, this
> is not a discussion about open access per se, but about the ways to
> achieve it.
For those who think librarian's open-ended second-guessing of hypothetical
contingencies would answer questions (1)-(4), the answers to these
questions might possibly have given a hint:
"What percentage -- and which -- of the journals to which you
currently subscribe (100%? 80%? 60% 40%?...) would you cancel
(and in favour of what, if anything) if the author's version of
*all* the articles in all those journals were accessible free
online (80%? 60%? 40%?...)? If they were accessible immediately
(after 6 months? 12? 24?...)?"
More specific, hence marginally more credible, but much less predictive
for (1)-(4), would be:
"Would you cancel journal X if 100% of its articles were
accessible free online (80%? 60%? 40%?)? If they were accessible
immediately (after 6 months? 12? 24?)?"
Still more specific, hence more credible, but utterly useless for
predicting anything to do with (1)-(4):
"If 100% of X were immediately available for free online and Y
was not, and your users needed X and Y equally, and you could
not afford both, and you currently subscribed to X and not to Y,
would you cancel X for Y?"
Yet the surveyed librarians were not even asked that. They were
asked which options they preferred least and most out of 3 competing
hypothetical "products" (consisting of journal contents), for example, (A)
a half-priced journal, 100% of its contents available immediately, vs. (B)
a quarter-priced journal, 80% of its contents available immediately,
vs. (C) a free journal, 100% of its contents available after a delay of
6 months.
Then the "Share of Preference model" was used to make projections from
a matrix of such most/least preferences among competing *products*
for *acquisition*, without taking into account that free articles,
anarchically self-archived by authors, in their own IRs, are neither
products for acquisition, nor are they acquired by acquisitions
librarians!
Questions (1) - (4) are questions about *cancellations*, under various
conditions, not about acquisitions among competing products; and they
are not answered by the "Share of Preference model" and its projections.
I do not know, from the fact that a librarian (like any consumer)
prefers free things to priced things, what they would cancel (in favor
of what) under what conditions. Yet that is what this is all about.
> It is worth finding structural ways to achieve and sustain open access.
> What weakens the current publishers' position (independents as well as
> societies) is that not everyone gives the choice of open access to the
> authors for all -- or even any -- of their journals.
Today, about 15% of authors self-archive spontaneously, at no cost (except
a few keystrokes). Unless Jan imagines that being given the opportunity
to pay a journal's asking price ($500 to $3000+) to do those keystrokes
for them would make a substantial difference for the growth of OA, I
rather doubt that having this added option will make much difference.
But the added option is very welcome anyway -- as long as it is not used
to try to prevent OA self-archiving from being mandated by institutions
and funders -- or to try to have the paid option mandated in its stead.
Note, though, that it is not an institution's acquisitions librarian,
buying incoming products for its library, that would be paying the
publishing costs for the institution's outgoing research articles! (That
certainly wasn't in PRC's product survey either!)
However, there *is* a way to square this circle. Read on:
> What weakens the self-archivers' position is that they desire the
> 'goods', but expect those to pay for it who would have more of a logical
> reason *not* to pay (i.e. librarians, through subscriptions).
The only goods that researchers want to get are the ones they give:
their own peer-reviewed research (conducted, written, and peer-reviewed
by them, but with the peer review administered by a reliable and
answerable 3rd-party fair-broker: the journal whose name and
track-record certifies the outcome of the peer review).
The bill for (i) the peer review, plus (ii) the paper edition, (iii)
the online edition, and (iv) other added-values, is currently being paid
by institutional subscriptions. If the market for (ii) - (iv) eventually
shrinks because of self-archiving to a point where it no longer covers
costs, costs will be cut, (ii) - (iv) and their expenses will be cut,
and the costs of (i) will be paid out of the institutional savings
(OA publishing).
Jan seems to prefer to have it all paid for now -- *all* of it --
providing OA to peer-reviewed research (i) only on condition that
it is bundled with (ii)-(iv) and their costs, regardless of whether
the authors or users want (ii)-(iv); and he sees the self-archiving
of just the unbundled peer-reviewed draft by the author as some sort
of free-loading. (I wonder why he doesn't see it as the only way to
"unbundle" (i) from (ii)-(iv), so that we can have immediate OA and let
(ii)-(iv) fend for themselves, as long as they still have a market?
Or as the way to redirect subscription savings on incoming products
and services that may one day no longer be needed toward paying for a
service -- peer-review -- on outgoing institutional research, that will
always be needed?)
> What is needed is a better appreciation of what 'publishing' in a
> journal actually is. It is not the 'making public', as that can be done
> easily enough without a publisher of any kind. It is more like 'making
> public under a plausible and creditable banner'. That is both more
> difficult and more costly than it might seem, superficially.
But the "plausible and creditable banner" is merely peer review (i),
and that journal-name's established track record for peer-review quality
standards.!
> Managing that process, which is a service to the scientific community,
> carries costs. Paying for those costs via subscriptions is a historic
> relic and flawed in an internet world, whether these subscriptions are
> sustaining the 'mother' journals of self-archived articles or not.
Which costs are we talking about? Those of (i), or the whole kit and
kaboodle, (i)-(iv), take it or leave it?
> Publishing is part of research, and therefore the cost of publishing is
> part of the cost of research.
I am not sure where institutional subscription budgets are coming from,
but that's the money we are discussing. Call it a part of research if
you like. It is being spent, today, on institutional journal
subscriptions. As long as those subscription revenues keep paying for
(i)-(iv), author self-archiving is pure benefit (to research and
researchers). If/when self-archiving should ever cause cancellations (the
empirical question under discussion here), and if/when the cancellations
should ever be big enough to make cost-recovery via institutional
subscriptions no longer sustainable, then the institution subscription
windfall-savings themselves (no longer covering the costs of (i)-(iv))
can be used to pay for just (i), unbundled, on the OA publishing
cost-recovery model that Jan favors (prematurely).
*If* -- and if so, *then* -- not now. Now we just self-archive, enjoy the
immediate benefits of OA, and let nature take its course to drive the
unbundling of (i) from (ii)-(iv), if that's what's in the cards. The
PRC survey under discussion is certainly not telling us whether or
when it is indeed in the cards. And Jan's suggestion to simply pop for
the asking price of the bundled (i)-(iv) a priori, right now, over and
above the funds already being spent on subscriptions, because "the cost
of publishing is part of the cost of research" sounds good as a slogan,
but it is a non-starter today, when all those research funds are still
committed to paying the subscriptions.
The logic and contingencies are all fully mapped out here, and have
been ever since 1999, but as far as I can tell, no one has ever stepped
through these simple contingencies seriously and attentively enough even
to grasp them (trivial though they are):
http://www.publications.parliament.uk/pa/cm200304/cmselect/cmsctech/399/399we152.htm
> An article processing fee model is the way to sustainable open access.
Perhaps. But what size fee? and a fee for what"? for (i)? or for the
whole (i)-(iv) bundle? and levied when? now, when the funds are still
tied up in subscriptions, for the whole bundle? paid for, now, out of what
funds, then? and why now, when we can self-archive and wait for (ii)-(iv)
to sink or swim without tying up access and impact for researchers for
another decade, and without double-dipping into scarce research funds
that are already paying for subscriptions?
> The so-called BOAI-1 [green OA self-archiving] and BOAI-2 [gold OA
> journal-publishing] are not symmetrical or even dimensions in the
> same space. (Self-)archiving is part of OA publishing; OA publishing
> is not part of self-archiving. Beckett and Inger, and most of their
> respondents, have understood that.
Researchers self-archive their own peer-reviewed articles, where the
costs of managing (not performing) the peer review (i) have been covered
out of part of the journal subscription revenue from the sale of the
bundled product (i)-(iv). That's not OA publishing, that's ordinary,
subscription-based publishing.
Researchers do not ask for royalties on the sale of their writings,
nor do they charge for providing peer review. They let the journal sell
subscriptions in exchange. (If and when journals should no longer be
able to make ends meet that way, they can and will adapt, in the
direction Jan envisages. But there is no adaptation without adaptive
pressure...)
So, Jan, your point is what? Pre-adaptation? Intelligent Design? Figure
out a way to redirect what's currently being spent on subscription
costs to pay for peer-review (i), unbundled from the sale of the paper
edition (ii), the online edition (iii), and any other add-ons (iv).
Once you have an intelligent design for that, and an intelligible way of
getting us from here to there, I'm with you.
Till then, time's a'wasting and access/impact is a'losing, daily, and
cumulatively, and needlessly. And OA self-archiving mandates can and
will remedy that, once and for all. So while publishing sorts out its
economics, let research sort out its access and impact.
Stevan Harnad
Received on Fri Nov 17 2006 - 04:10:25 GMT