Definition of Open Access

Stevan Harnad amsciforum at GMAIL.COM
Sat Dec 7 08:52:15 EST 2013

The Green/Gold distinction (which is based on who provides the access, the
publisher or the author) is more important now than ever, as publishers
fight to retain control of their content. The distinction resolves
confusion rather than creating it; and it is simple to understand (but then
needs to be adhered to).

The OA movement should resolutely push for Green OA; Green OA mandates
should be formulated to ensure that *compliance is by the party bound by
the mandate* (the fundee, if a funder mandate, the employee, if an
institutional mandate). On no account should mandates rely on compliance by
a 2nd party, the publisher, who is not bound by the mandate and has every
interest in maintaining control over the content.

There is a 3rd way in which articles can be made OA of course, other than
by the author (or the author's assigns) (Green) or by the publisher (Gold):
It can be made OA by a 3rd party, either a user or a rival publisher or
service provider. This is partly what the kerfuffle is about,
and it will no doubt spread to ResearchGate, Mendeley and the like. (It
also concerns versions, because Green OA usually involves only the author's
final draft whereas 3rd-party OA often involves the publisher's proprietary

My advice to those who are up in arms about Elsevier's take-down notice for
3rd-party service providers is to redirect their resentment to doing
something legal and feasible, namely, mandating and depositing the final
draft in their institutional repository immediately upon acceptance, and
making it OA as soon as they can (or wish).

The term "OA" (and the goal of the OA movement) should also continue to be
reserved for  *immediate (online) access*. The inverse of Open Access is
Access Denial. Access is denied by Access Tolls (subscriptions, licenses,
pay-to-view), but, just as surely, it is denied by Access Embargoes. Hence
it is a contradiction in terms to call Embargoed Access "Delayed Open
Access." It is *Delayed Access (DA),* just as Toll Access is *Toll Access
(TA)*, not "Toll Open Access!"

And, as I mentioned in my preceding post, a one year access embargo is now
the real target to beat (as publishers already know all to well). Access
delayed for a year is not a victory for the advocates of Open Access; nor
is it a solution to the Access/Impact problem in the online era. A 1-year
delay might be a convenient unit for doing bibliometric measurements on the
growth and latency of Green and Gold Access (and a welcome compromise and
marketing ploy for the publishing industry), but "Open Access" should
continue to be reserved for immediate, toll-free (and permanent!) online

(The same applies whether the Open Access is Gratis or Libre, by the way.)

*Stevan Harnad*

On Sat, Dec 7, 2013 at 3:50 AM, Bo-Christer Björk <
bo-christer.bjork at> wrote:

>  I fully agree with Stevan on the need to define a clear standard for
> what we measure and when, by I have a different view of some details.
> Why not simply talk of "Immediate Open Access" and "Delayed Open Access",
> both provide open access. I'm also getting more and more hesitant about the
> use of the terms Gold and green since there is so much confusion in actual
> usage. The term subsidized open access is kind of misleading. The only
> subsidy a lot of OA journals, in particular in the social science and
> humanities, and journals published elsewhere than in the US, UK, are
> getting is the usage of a university web site, the marginal cost of which
> is almost nil. Or in Latin America etc. the use of Scielo, which is very
> low cost per journal and hence only a small part of their resource use.
> Other than that its mainly voluntary work by academic communities. Remember
> that the universities of editors, reveiewers, etc already "subsidize"
> society and commercial publisher journals.
> The open archives term (for delayed open access) that Elsevier invented is
> downright silly. Most people who think of this as getting e-access to
> articles published many years and decades ago.
> I agree with Stevan that perhaps their could be a three month delay border
> for the definition of immediate Open Access, to allow for a slight delay
> for authors putting up manuscripts of non-embargoed journal articles. As
> for delayed OA I would suggest going for just one minumum period in broad
> studies and I would put it at slightly over a year, perhaps 15 months. This
> has to do with the increasingly common 12 month embargo periods, and again
> the fact that many authors following such embargoes may post a couple of
> months later. Also it is very common for academics to post articles to IRs
> for their full last year production in January, February the next year when
> they have to report meta data to their universities for book-keeping, which
> means that for some article the delay will be slightly over a year.
> If a study in particular wan't to study how green OA increases as a
> function of the delay (6, 12, 24 ect) that is naturally fine, but in most
> reporting in the popular press (including journal like Nature) they
> simplify the message to single figures.
> In practice it is difficult in mass studies based on sampling of say
> Scopus meta data to determine the exact delays for each article (which
> would also entail also finding out when the copy was posted). All you can
> do is run the googling at one point in time (or a relatively short period).
> In order to have a big enough delay it is often convenient to use the
> scopus or ISI data of articles published in the year before the last one.
> One last item which somehow would need to be sorted out (and which was
> raised in connection with the recent Science-Metrix study) is that
> automated searches also catch what I would label "promotional OA", for
> instance the practice of many publishers to have the first issue of the
> last year open using a rolling scheme (that is if you google a year later
> the articles are no longer available). Dependent on the time lag of the
> study, but in particular for delays between a few months and say a year and
> a half, counting such articles in could raise the overall OA prevalence
> with as much as five percent. Also due to the fact that in such studies
> googled hits are sometimes classified as gold OA, based on the journals in
> question being in DOAJ, such hits will then misleadingly  be classified as
> green OA.
> Bo-Christer
> On 12/7/13 2:01 AM, Stevan Harnad wrote:
> On Fri, Dec 6, 2013 at 2:00 PM, Bo-Christer Björk <
> bo-christer.bjork at> wrote:
>> The Elsevier study on OA prevalence study was part of broader report. The
>> methods are just shortly mentioned so its a bit problematic to comment in
>> detail.
>> The global gold OA share found is 9,7 % of scopus articles, consisting of
>> 5,5 % APC paid and 4,2 others (not just 5.5 % as Stevan noted below). The
>> global hybrid share is 0.5. The green global share could be assumed to more
>> or less be the sum of preprint versions of 6.4 % and accepted versions 5.0
>> %, adding directly to around 11 %. In particular if their method only took
>> the first found full text copy and then classified it
>> The big flaw of the study seems to be in the sample used, since it
>> consisted of equal numbers of Scopus articles that had been published 2
>> months, 6 months, 12 months and 24 months before the Googling. If the hits
>> are simple added up for all the sampled articles this means that a major
>> share of selfarchivied manuscripts are ignored, due to embargoes or author
>> behavior in for instance selfarchiving once a year. For instance half of
>> the copies in PMC would not be found in this way. Equally the very low
>> figure for "Open Archives", 1.0 %, could be a result of this method. Our
>> own results for delayed OA are around 5 %.
>> So all in all the figures are much lower than if one includes articles
>> made OA with at least a one year delay, which we find is the method we
>> would recommend for studies claiming to give overall OA uptake figures.
>> Whether this methodological choice was a conscious one from the study team
>> or just an oversight is difficult to know. But if they would have adhered
>> to a strict interpretation that only immediate OA is OA, the sampling
>> should have been different. Now it's somewhere in between.
>  Bo-Christer is quite right. Elsevier's arbitrary (and somewhat
> self-serving) 6-category classification system (each of whose categories is
> curiously labelled a "publishing system") leaves much to be desired:
>  1. Gold Open Access
> 2. Hybrid
> 3. Subsidised
> 4. Open Archives
> 5. Green Open Access: Pre-print versions
> 6. Green Open Access: Accepted Author Manuscript versions
>  It is not just what Elsevier called "Gold Open Access" that was Gold
> Open Access, but also what they called "Subsidised." The difference is
> merely that what they called Gold was publishing-fee-based Gold and what
> they called subsidized was subsidy-based Gold.
>  Elsevier also neglected to mention that "Subsidised" did not necessarily
> mean subsidized either: There are also subscription-based journals that
> make their online versions free immediately upon publication; hence they
> are likewise Gold OA journals.
>  What Elsevier called "Open Archives" is also not what it sounds like: It
> seems to be *Delayed Access* articles, accessible only after a publisher
> embargo, either on the publisher's website or in another central website,
> such as PubMed Central, where publishers also deposit, sometimes
> immediately, sometimes after an embargo.
>  The two Green Open Access categories are also ambiguous.The pre-print
> versions are (correctly) described as pre-refereeing drafts (but it would
> take a lot closer analysis to determine whether the pre-prints differ from
> the refereed version. It is easy to determine whether they were posted
> before the official publication date but far from easy to determine whether
> they were posted before refereeing. (The date of the letter of acceptance
> of the refereed draft is often one that only the author and the editor know
> -- though it is in some cases printed in the journal: did Elsevier look at
> that too?)
>  The post-refereeing author's drafts are presumably what they are
> described as being, but it is not clear by what criteria Elsevier
> distinguished them from pre-refeeeing drafts (except when they were in an
> institutional repository and specifically tagged as unrefereed).
>  So, as Bo-Christer points out, there are many methodological questions
> about the data without whose answers their meaningfulness and
> interpretability is limited. I would say that the timing issue is perhaps
> the most important one. And to sort things out I would like to propose a
> different system of classification:
>  *Open Access (OA):* The term OA should be reserved for immediate OA,
> regardless whether it is provided by the publisher (Gold) or the author
> (Green). A reasonable error-margin for OA should be* within 3 months or
> less from publication date*. Anything longer begins to overlap with
> publisher embargoes (of 6, 12, 24 months or longer).
>  *Delayed Access (DA): *The term DA should be used for delays of more
> than 6 months. And besides the usefulness of separately counting 6, 12, and
> 24 month DA, DA should also be analyzed as a continuous variable, reckoned
> in months starting from the date of publication (including negative delays,
> when authors post the refereed draft during the interval from acceptance
> date to publication date. The unrefereed preprint, however, should not be
> mixed into this; it should be treated as a separate point of comparison.
>  So there is *Gold OA* (immediate), *Green OA* (immediate), *Gold DA* and *Green
> DA* (measured by 6-month intervals as well as continuously in months.
>  If a separate distinction is sought within Gold, then fee-based Gold,
> subsidy-based Gold and subscription-based Gold can be compared, for both OA
> and DA. The locus of deposit of the Gold is not relevant, but the fact that
> it was done by the publisher rather than the author (or the author's
> assigns) is extremely relevant.
>  For Green OA and DA it is also important to compare locus of deposit
> (institutional vs. institution-external). See mandates below.
>  In all cases independence and redundancy should uniformly be controlled:
> Whenever a positive "hit" is made in any category, it has to be checked
> whether there are any instances of the same paper in other categories.
> Otherwise the data are not mutually exclusive.
>  If desired, all the above can be further subdivided in terms of *Gratis*(free online access) and
> *Libre* (free online access plus re-use rights) OA and DA.
>  Tracking Gold has the advantage of having clear unambiguous timing
> (except if the publication date differs from the date the journal actually
> appears) and of being exhaustively searchable without having to sample or
> check (if one has an index of the Gold OA and DA journals).
>  Tracking Green is much harder, but it must be done, because the fight
> for OA is rapidly becoming the fight against embargoes. That's why Green OA
> should be reserved for immediate access. It is almost certain that within
> the next few years most journals will become Gold DA (with an embargo of 12
> months). Hence 12 months is the figure to beat, and Green DA after 18
> months will not be of much use at all.
>  And the best way to push for immediate Green OA, is to upgrade all Green
> mandates to require *immediate institutional deposit*, irrespective of
> how long an embargo the mandate allows on DA. Requiring immediate deposit
> does not guarantee immediate OA, but it guarantees immediate Almost-OA,
> mediated by the repository's automated copy-request Button, requiring only
> one click from the requestor and one click from the author.
>  The immediate-deposit requirement plus the Button not only fits all OA
> mandates (no matter how they handle embargoes of copyright), making it
> possible for all institutions and funders to adopt it universally, but it
> also delivers the greatest amount of immediate access for 100% of deposits:
> immediate Green OA for X% plus (100-X)% Button-mediated Almost OA. And
> this, in turn will increase the universal demand for immediacy to the point
> where publisher embargoes will no longer be able to plug the flood-gates
> and the research community will have the 100% immediate Green OA it should
> have had ever since the creation of the web made it possible by making it
> possible to free the genie from the bottle,
> *Stevan Harnad*
>>         On 12/6/13 5:31 PM, Stevan Harnad wrote:
>> Elsevier has just conducted and published a study commissioned by UK BIS:
>> "International Comparative Performance of the UK Research Base – 2013<>
>> "
>> This study finds twice as much Green OA (11.6%) as Gold OA (5.9%) in the
>> UK (where bothGreen OA repositories<>
>>  and Green OA mandates<> began)
>> and about equal levels of Green (5.0%) and Gold (5.5%) in the rest of the
>> world.
>> There are methodological weaknesses in the Elsevier study, which was
>> based on SCOPUS data (Gold data are direct and based on the whole data set,
>> Green data are partial and based on hand-sampling; timing is not taken into
>> account; categories of OA are often arbitrary and not mutually exclusive,
>> etc). But the overall pattern may have some validity.
>> What does it mean?
>> It means the effects of Green OA mandates in the UK<> --
>> where there are relatively more of them, and they have been there for a
>> half decade or more -- are detectable, compared to the rest of the world<>,
>> where mandates are relatively fewer.
>> But 11.6% Green is just a pale, partial indicator of how much OA Green OA
>> mandates generate: If instead of looking at the world (where about 1% of
>> institutions and funders have OA mandates) or the UK (where the percentage
>> is somewhat higher, but many of the mandates are still weak and ineffective
>> ones), one looks specifically at the OA percentages for effectively
>> mandated institutions <>, the Green
>> figure jumps to over 80% (about half of it immediate-OA and half embargoed
>> OA: deposited, and accessible during the embargo via the repository's
>> automated copy-request Button, with a click from the requestor and a click
>> from the author).
>> So if the planet's current level of Green OA is 11.6%, its level will
>> jump to at least 80% as effective Green OA mandates are adopted.
>> Meanwhile, Gold OA will continue to be unnecessary, over-priced,
>> double-paid (which journal subscriptions still need to be paid) and
>> potentially even double-dipped (if paid to the same hybrid
>> subscription/Gold publisher) out of scarce research funds contributed by UK
>> tax-payers ("Fool's Gold<>
>> ").
>> But once Green OA prevails worldwide, Fair Gold<> (and
>> all the Libre OA re-use rights that users need and authors want to provide)
>> will not be far behind.
>> We are currently gathering data to test whether the immediate-deposit<>
>>  (HEFCE<>
>> /Liege<>)
>> Green OA mandate model is indeed the most effective mandate (compared, for
>> example, with the Harvard<> copyright-retention
>> model with opt-out, or the NIH<> model
>> with a 12 month embargo).
>> *Stevan Harnad*
>> P.S. Needless to say, the fact that the UK's Green OA rate is twice as
>> high as its Gold OA rate is true *despite* the new Finch/FCUK policy<> which
>> subsidizes and prefers Gold and tries to downgrade Green -- certainly not
>> because of it!
>> _______________________________________________
>> GOAL mailing listGOAL at eprints.org
>> _______________________________________________
>> GOAL mailing list
>> GOAL at
> _______________________________________________
> GOAL mailing listGOAL at eprints.org
> _______________________________________________
> GOAL mailing list
> GOAL at
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <>

More information about the SIGMETRICS mailing list