Fast-Forward on the Green Road to Open Access:
The Case Against Mixing Up Green and Gold

Stevan Harnad


This article is a critique of:

The "Green" and "Gold" Roads to Open Access:
The Case for Mixing and Matching

Jean-Claude Guédon
Serials Review 30(4) 2004

Open Access (OA) means: free online access to all peer-reviewed journal articles.

Jean-Claude Guedon (J-CG) argues against the efficacy of author self-archiving of peer-reviewed journal articles -- the "Green" road to OA -- on the grounds (1) that far too few authors self-archive, (2) that self-archiving can only generate incomplete and inconvenient access, and (3) that maximizing access and impact is the wrong reason for seeking OA (and only favors elite authors). J-CG suggests instead that the right reason for seeking OA is so as to reform the journal publishing system by converting it to OA ("Gold") publishing (in which the online version of all articles is free to all users). He proposes converting to Gold by "mixing and matching" Green and Gold as follows:

First, self-archive dissertations (not published, peer-reviewed journal articles). Second, identify and tag how those dissertations have been evaluated and reviewed. Third, self-archive unrefereed preprints (not published, peer-reviewed journal articles). Fourth, develop new mechanisms for evaluating and reviewing those unrefereed preprints, at multiple levels. The result will be OA Publishing (Gold).

I reply that this is not mixing and matching but merely imagining: a rather vague conjecture eculation about how to convert to 100% Gold, involving no real Green at all along the way, because Green is the self-archiving of published, peer-reviewed articles, not just dissertations and preprints.

I argue that rather than yet another 10 years of speculation
what is actually needed (and imminent) is for OA self-archiving to be mandated by research funders and institutions so that the self-archiving of published, peer-reviewed articles (Green) can be fast-forwarded to 100% OA. The direct purpose of OA is to maximize research access and impact, not to reform journal publishing; and OA's direct benefits are not just for elite authors but for all researchers, for their institutions, for their funders, for the tax-payers who fund their funders, and for the progress and productivity of research itself.

There is a complementarity between the Green and Gold strategies for reaching 100% OA today, just as there is a complementarity between access to the OA and non-OA versions of the same non-OA articles today. Whether 100% Green OA will or will not eventually lead to 100% Gold, however, is a hypothetical question that is best deferred until we have first reached 100% OA, which is a direct, practical, reachable and far more urgent immediate goal -- and the optimal, inevitable and natural outcome for research in the PostGutenberg Galaxy.

All highlighted quotes are from J-CG's article:

Recent discussions on Open Access (OA) have tended to treat OA journals and self-archiving as two distinct routes.

From the day it was coined in 2001 by the Budapest Open Access Initiative (BOAI),  'Open Access' has always been defined as free online access, reachable by two distinct routes, BOAI-1, OA self-archiving ("Green") and BOAI-2, OA journals ("Gold"):

To achieve open access to scholarly journal literature, we recommend two complementary strategies.

    I.  Self-Archiving: First, scholars need the tools and assistance to deposit their refereed journal articles in open electronic archives, a practice commonly called, self-archiving. When these archives conform to standards created by the Open Archives Initiative, then search engines and other tools can treat the separate archives as one. Users then need not know which archives exist or where they are located in order to find and make use of their contents.

    II. Open-access Journals: Second, scholars need the means to launch a new generation of journals committed to open access, and to help existing journals that elect to make the transition to open access. Because journal articles should be disseminated as widely as possible, these new journals will no longer invoke copyright to restrict access to and use of the material they publish. Instead they will use copyright and other tools to ensure permanent open access to all the articles they publish. Because price is a barrier to access, these new journals will not charge subscription or access fees, and will turn to other methods for covering their expenses. There are many alternative sources of funds for this purpose, including the foundations and governments that fund research, the universities and laboratories that employ researchers, endowments set up by discipline or institution, friends of the cause of open access, profits from the sale of add-ons to the basic texts, funds freed up by the demise or cancellation of journals charging traditional subscription or access fees, or even contributions from the researchers themselves. There is no need to favor one of these solutions over the others for all disciplines or nations, and no need to stop looking for other, creative alternatives.

Open access to peer-reviewed journal literature is the goal. Self-archiving (I.) and a new generation of open-access journals (II.) are the ways to attain this goal.

Some supporters of self-archiving even suggest that [self-archiving] alone can bring about full Open Access to the world's scientific literature.

OA's focus is only on peer-reviewed journal articles -- 2.5 million annual articles in 24,000 peer-reviewed scholarly and scientific journals -- not on "the world's scientific literature" in its entirety [i.e., not books, magazines]).

(1) To self-archive one's own article is to provide Open Access (OA) to one's own article. Every author can do this, for every one of his articles. If/when every author does this, for each of the annual 2.5 million articles, we have, by definition, 'full Open Access' (Green).

(2) By the same token, if/when every publisher of each of the 24,000 journals converts to OA publishing, we have, by definition, 'full Open Access' (Gold).

The rest is simply a question of probability: Is it more probable that all or most journals will convert to OA, or that all or most of their authors will self-archive their articles? Which faces more obstacles, costs, delay, uncertainty, risk? Which requires more steps? Which can be facilitated by university and research-funder OA mandates? Which is already within immediate reach?


In this paper, it is argued that each route actually corresponds to a phase in the movement toward Open Access; that the mere fact of self-archiving is not enough; that providing some branding ability to the repositories is needed.

The mere fact of self-archiving is not enough for what? Would 100% self-archiving not correspond to 100% OA (just as 100% OA journals would)?

And as we are talking about the self-archiving of peer-reviewed, published journal articles, why is there a need for "branding"? Branding what? The journal articles? But those are already branded -- with the name of the journal that published them. What is missing and needed is not branding but Open Access to those journal articles! (J-CG's preoccupation with branding will turn out to be a consequence of the fact that he is not proposing a way to make current journal articles OA, but a way to replace current journals altogether.)

However, doing so will eventually bring about the creation of overlay (or database) journals. The two roads, therefore, will merge to create a mature OA landscape.

It is very easy to imagine how OA journals (and indeed non-OA journals) might one day evolve into mere "overlays" on their OA articles, which are all self-archived in OA Archives by their authors. The OA journal could provide the peer-review service, and certify its outcome with the 'brand', namely, its journal-name.

But right now, this is merely a speculation about what could possibly happen, some day. Today, only 5% of journals are OA journals, providing 5% OA, and 15% OA is provided by author-self-archiving  of articles published in non-OA journals. And 0.01% of journals (whether OA or non-OA) are 'overlay journals.'

What is accordingly needed today is 100% OA -- not 'branding', nor conjectures about how journals might somehow, some day, evolve into 'overlay' journals.

The notion that the self-archiving of published, "branded" journal articles to make them OA is somehow not "full OA" -- because it lacks "branding" and awaits "overlay journals" -- represents a rather profound misunderstanding of both self-archiving and OA.

And what is certain is that the notion that the self-archiving of published, 'branded' journal articles to make them OA is somehow not 'full OA' -- because it lacks 'branding' and awaits 'overlay journals' -- represents a profound misunderstanding of both self-archiving and OA.

Historically, Open Access (OA) emerged largely as a reaction to the fast increasing prices of scholarly and scientific journals.

Historically, the journal pricing/affordability  problem drew attention to the access/impact problem, but OA itself certainly was not a reaction to the journal pricing/affordability problem. The first ones to provide OA (long before 'OA' was defined, and long before OA journals existed) were researchers themselves, self-archiving their articles as a reaction to the new possibilities opened up by the Internet. Two prominent early cases of OA self-archiving are well known -- physics (300,000 papers to date) and computer science (500,000 papers to date) -- but in fact there is now evidence that a good deal of self-archiving has been going on for at least a decade now in just about all disciplines:

All this OA self-archiving has been going on as a natural reaction to the access/impact problem by researchers -- most of them not even aware of the pricing/affordability problem, although there is a causal connection, of course. (If the online version of all journals were affordable to all research institutions, then there would be no access/impact problem, and hence no need for OA self-archiving.) But it is not true that OA self-archiving emerged as a reaction to the fast increasing prices of scholarly and scientific journals. It emerged as a reaction to the obvious potential of the Web to maximize the access to and the impact of research findings.

researcher's work.

The concern, first expressed by librarians, was that the high prices of journals obviously limited access by economic means. Gradually, the question has evolved, and issues of access have been increasingly distinguished from issues of costs (or affordability).

Librarians were the first to draw attention to the pricing/affordability problem, but the access/impact problem was already felt by researchers, and they were already doing something about it, on their own initiative, thanks to the advent of the Net and Web.

(It was in fact the library community that implicitly mixed up the affordability and access problems, especially in the OA context, and these are lately now beginning to be unmixed, at last.)

In parallel, Open Access has been increasingly focusing on articles, beside journals. A number of reasons have contributed to this gradual shift: scientists as readers tend to pay more attention to articles;

Users have always focused on articles, not journals. The OA movement has been increasingly re-focusing on article self-archiving, having temporarily forgotten it. The research (author) community has not only not forgotten article self-archiving, but has been doing it, not only in parallel with the OA movement, but well before it, and with no explicit focus on journal affordability. It just has not been doing enough of it yet.

digital publishing maintains the journal titles mainly for branding reasons,

It is very difficult to put a comprehensible construal on the foregoing sentence:

Digital publishing'? What sort of entity is that? (Journal publishers? They publish both paper and online editions of their journals.)

Maintaining journal titles for branding reasons? What does that mean? Journals publish journals, and their journals have names, and their authors and users recognise those names and their associated track records (and impact factors), and use them in deciding which journal to publish in and which journal-articles to read. The service provided by the journal includes peer review, publishing (online and on-paper), dissemination, and (to an extent) archiving (of the online version).

What has this to do with the proposition that "digital publishing maintains the journal titles mainly for branding reasons"? (This is in fact the first sign of a speculation that J-CG will be making later in his paper, about a hypothetical day when journals will become mere "overlays" of some kind.)

What has this to do with the proposition that 'digital publishing maintains the journal titles mainly for branding reasons'? This sounds like another speculation about the hypothetical day when journals may become merely 'overlays' of some kind.

but the bundling strategies used by several major publishers tend to rest about equally on number of titles and number of articles; the very dynamics of the "Open Access" movement, as we shall see, have also contributed to give greater prominence to the article as a unit.

Researchers focus on access to articles because it is articles that they write, publish, read, use and cite. This has next to nothing to do with publishers' bundling strategies. Nor does OA.

"Open Access" became a movement after a meeting was convened in Budapest in December 2001 by the Information Program of the Open Society Institute. That meeting witnessed a vigorous debate about definitions, tactics, and strategies,1 and out of this discussion emerged two approaches which have become familiar to all observers, friends, or foes. First, existing journals find a way to transform themselves into Open Access publications, or new Open Access journals are created. Second, authors and/or institutions "self-archive" published peer review articles or a combination that then becomes the equivalent of published, peer-reviewed articles.

(There is a minor historical error here: OA publication (BOAI-2, "Gold") was not the first of the BOAI routes to OA but the second. OA self-archiving of articles published in non-OA journals (BOAI-1, "Green") was the first.)

The first strategy amounts to a reform of the existing publication system. It fundamentally relies on journals as its basic unit, and it simply aims at converting or creating the largest possible number of Open Access journals.

Both OA self-archiving and OA journal publishing (and indeed, OA itself, and the definition of OA) 'fundamentally rel[y] on journals as [their] basic unit' because it is the articles in peer-reviewed journals that are the target literature of the OA movement.

It is true, however, that only BOAI-2, OA journal publication (Gold), aims at a reform of the existing publication system. BOAI-1, OA self-archiving (Green), is neutral about that. It aims only at OA.

BioMed Central, a commercial operation, has played a crucial pioneer role in this context. More recently, it has been joined by the nonprofit Public Library of Science (PloS). This strategy obviously threatens the "reader-pays" business plan2 and therefore immediately faces the issue of financial viability, with the result that spirited debates have been generated, largely centered on the viability of the "author pays"3 model used by BioMed Central and the Public Library of Science.

Unfortunately, these spirited debates, centered on the viability of OA Publishing (BOAI-2, Gold), have been both perceived and portrayed as debates on the viability of OA itself, at a considerable cost in lost time and lost OA (for having all but forgotten about BOAI-1, OA self-archiving, Green).

There has been a plus side to this disproportionate focus on OA publishing: it has drawn a good deal of attention to OA, especially among those who are more interested in economic problems and iniquities. But I am not sure that this plus altogether compensates for the minus, which is that this disproportionate focus on OA publishing has not generated very much OA. Instead, it has drawn attention and energy away from OA self-archiving, which has the immediate potential to generate 100% OA virtually overnight, institutional OA archives being incomparably cheaper, faster and easier to create than OA journals. During all that 'spirited debate' about the viability of the 'author pays' model we could instead have been informing authors that they themselves can provide this OA they purport to want and need so much -- by simply self-archiving their own published articles.

But perhaps the spirited debate on the viability of BOAI-2 was needed for everyone to come to realize in the end that it is BOAI-1 that is in the immediate position to provide 100% OA, and hence needs to be mandated by research institutions and funders.

In other parts of the world, a number of research councils or academies supported by governmental, public, funds have also begun transforming their journals into Open Access publications.4 In such cases, the issue of financial viability simply rests on the will of governments to support scientific publishing-a point that varies very much with each country and circumstances.

All these new and converted OA journals are valuable and welcome, but their numbers and the rate of increase of their numbers has to be realistically noted : About 5% of journals are OA ('Gold') journals today (1400/24,000). In contrast, about 93% of journals are 'Green' -- i.e., they have given their authors the Green light to self-archive their articles if they wish. The rate of increase in the number of Green journals has been incomparably faster than the rate of increase in the number of Gold journals in the past few years. The amount of OA (15%) generated via self-archiving has also been three times as great as the amount of OA generated via OA publishing (5%); (although direct measures have not yet been made) it is likely that the rate of growth of OA via OA self-archiving is also considerably higher than the rate of growth of OA via OA publishing -- for obvious reasons that have already been mentioned: It is far easier and cheaper to create and fill an institutional OA Archive than to create and fill an OA journal. Moreover, there is a considerable financial risk for an established journal in converting to the OA cost-recovery  model, which has not yet been tested long enough to know whether it is sustainable and scaleable.

So whereas all new and converted OA journals are welcome, it makes no sense to keep waiting for or focusing on them as the main source of OA. The real under-utilized resource is OA self-archiving -- underutilized even though it already provides three times as much OA as OA journals and is probably growing faster too: because OA self-archiving is already in a position to provide immediate 100% OA, if only it is given more of our time, attention and energy.

It is unrealistic to imagine that the reason the number of new and converted gold journals is not growing faster is that governments are not willing to subsidize them! It is not clear whether governments should even want to subsidize them, at this point, when OA is already reachable without any need for subsidy, via self-archiving, and researchers are simply not yet ready to perform the few keystrokes required to reach for it (even for the 93% of their articles published in green journals) despite being alleged to want and need OA, despite being willing (in their tens of thousands!) to perform the keystrokes to demand it from their publishers -- -- and despite the fact that the benefits of OA itself are intended mainly for researchers and research.

If government intervention is needed on behalf of OA, surely it is needed in order to induce their researchers to provide the OA that is already within their reach to provide, rather than to subsidize journals to do it for them.

While in the United States, such governmental intervention may sometimes seem problematic, especially from the perspective of the publishing business, in other parts of the world, this is accepted and practiced as a matter of course. However, what is at stake in all countries is how to integrate the publication costs within the research costs, given that the latter are largely supported by public money (including the United States, this time).

If OA is a desirable enough thing, and reachable, government should certainly intervene to see that it is reached, if it can. Making government funding available to pay the costs of publishing in OA journals is fine, but that cannot generate much immediate OA (5%). In contrast, mandating self-archiving can generate 93% immediate OA at the very least! Hence it is not clear why we keep indulging in this 'spirited debate' on governments subsidizing OA publishing costs when governments could be generating at least 93% immediate OA by simply mandating self-archiving (for government-funded research).

And that is exactly what the US and UK self-archiving mandates have proposed to do:

Deceptively simple to describe - hence its rhetorical seduction - the "self-archiving" strategy appears much more complicated and subtle when approached conceptually.

I will try to show that self-archiving is exactly as simple as it purports to be, and that what confuses the picture is merely the unnecessary complications introduced by speculating (gratuitously) about the need for reforming the publishing system (instead of concentrating on the non-speculative need for providing OA).

[OA self-archiving] both relies on, and forgets about, journals.

As will now be demonstrated, it is not OA self-archiving that forgets that OA self-archiving is the self-archiving of peer-reviewed articles published in non-OA journals: Rather, it is those who speculate about the ultimate need for a conversion to OA publishing who keep forgetting that OA self-archiving is the self-archiving of peer-reviewed articles published in non-OA journals.

Generally speaking, [self-archiving]  rests on the preeminence of the article as fundamental unit. From this perspective, journals matter only to differentiate between peer-reviewed articles and non-peer-reviewed publications and to provide symbolic value:

Symbolic value? Consider how much simpler and more straight-forward it is to state this theory-independently:  Today, most of the 2.5 million articles published in the world's 24,000 peer reviewed journals are inaccessible to many of their potential users because they cannot afford access. If the articles are made accessible free online (by self-archiving them), this problem is solved.

I need not theorize about why users want to use peer-reviewed journal articles. I can take that as a rather obvious premise. Yes, users want the peer-reviewed articles (and the journal's name tells them which ones those are); and peer review itself provides the 'value' they seek in an expert-vetted literature rather than an unfiltered free-for-all. There is no need to debate the value of peer review in an OA context: One of the premises of the OA movement is that OA is about access to the peer-reviewed journal literature, not access to something else. So peer-review and the journal-names come with the territory. The only problem to solve is access. And Open Access solves that. And self-archiving is by far the fastest and surest way to provide immediate OA.

No further theorizing, or complicating, is needed: We have peer-reviewed journal articles, but we don't have Open Access to them. Self-archiving them provides that access. End of story. The rest is merely speculation (needless speculation, needless complication) needlessly delaying OA.

If I archive an article published in Cell, it still benefits from the Cell branding effect. Therefore, journals contribute to the impact of individual articles by their prestige - a dimension generally associated with the notion of "impact factor."

But what is the point being made here? Of course my purpose in self-archiving my Cell article (Cell is a Green journal, by the way) is to add to (1) the impact I already get from having successfully published it in Cell and thereby successfully reached those potential users who can afford access to Cell, (2) the further impact that I would otherwise have lost, from all those would-be users who cannot read, use and cite my Cell article because they (their institutions, actually) cannot afford to access it

Why all this theorizing about 'branding' effects? Cell is the name ('brand') of the journal. Cell has built up, across the years, a track-record for selectively publishing articles of a certain quality level (by applying, across the years, peer-review standards of a certain quality level). So the reason authors prefer to publish their articles in Cell (rather than a lower-quality journal, or no journal at all) is to meet, and show they meet, Cell's established quality-standards. And the reason users prefer to use articles published in Cell (rather than a lower-quality journal, or no journal at all) is because they prefer to devote their limited reading time to reading -- and to risk their limited research time in using and trying to build upon -- articles published in Cell (rather than a lower-quality journal, or no journal at all).

Nothing changes with self-archiving, except that access, and hence impact, are maximized -- for the same articles, in the same journals.

As becomes obvious from these remarks, journals are useful mainly to the researcher-as-author; the author-as-reader, on the other hand, cares mainly about articles and pays attention to journals only to the extent that they may help guide his/her reading choices. "Self-archiving" consequently proceeds in parallel to, and largely independently from, journals. It acts "as a supplement to toll access" and not as a substitute.5

I cannot follow this argument, and I suspect that one must be in the grip of some theory in order to see any point here: The journal, which provides the peer-review and certifies its outcome as having met its established quality standards, performs exactly the same kind of function for both the author and the user! It tags the work as having met a known quality standard. The author chooses in which journal to (try to) publish his article on the basis of the journal's quality track-record, and the user chooses which article to read and use on the basis of the journal's quality track-record.

Hence the self-archived version of the article is precisely as described above: a supplement to the toll-access version, for those who cannot afford to access it, not a substitute for it (for those who can). This is only unsettling for someone who is in thrall to a theory to the effect that what researchers really want and need is a substitute! This is just one step away from declaring that OA itself is not in fact enough: What we really want and need is OA publishing. And that would come rather close to undermining the entire case for OA, making it a mere accessory to a hypothesis about the optimal publishing system, rather than an end in itself.

('Not enough for what?' one is inclined to ask? Was Open Access meant to provide Open Access or something else -- like a solution to the pricing/affordability problem, perhaps? and/or a reform of journal publishing? The right reply is: Hypotheses non Fingo! Open Access was meant to provide Open Access!)

Finally, and seen from the perspective of "self-archiving," journals might become (negatively) relevant again only when and if they implement policies that make "self-archiving" difficult or even impossible.

I am not sure what is meant here, but I suspect it is something like: "If Green journals had not become Green, or if they changed their minds"? This is again a counterfactual speculation. One can of course counter-speculate that if publishers had not given self-archiving the Green light, authors could have, and would have, self-archived anyway. Fifteen percent had been doing it already, some since the early '90s. But I think it is far more sensible (and more productive of OA) to leave off speculating and counter-speculating and instead get to work actually generating the OA that is within reach.

Speculation just invites counter-speculation, and one can counter-speculate as well as the speculators can speculate, if one must: If publishers had not given self-archiving the Green light, authors could have, and would have, self-archived anyway. (Fifteen percent had been doing it already, some since the early '90s.)

As to publishers changing their minds about giving self-archiving the Green light: It was difficult enough, in the light of the demonstrated benefits that self-archiving confers on both authors and users (i.e., on researchers and research), for publishers not to give it the Green light today (and 93% have done so already) . As OA grows (and is mandated) it will only become more difficult not to give it the Green light, let alone try to withdraw the Green light :

In summary, "self-archiving" is a strategy that has been designed by researchers and for researchers, with little interest for any other player involved in scientific publishing.

But really, isn't the content of the 24,000 peer-reviewed research journals -- the annual 2.5 million articles -- mostly research conducted and reported by researchers for users (mostly again researchers) who wish to use, apply and cite it? Other 'players' do play a role in this too. (Publishers add value; librarians provide valuable service.) But doesn't the purpose of Open Access to this research output concern mainly its providers and users (including their institutions and funders), rather than other 'players'?  

[Self-archiving] simply aims at improving the research impact of established scientists and little else.

This is dead wrong (and startlingly so!). The purpose of self-archiving is to maximize every user's access and every author's impact! Why on earth would one imagine that the benefits of OA would be reserved for 'established scientists' alone? If anything, maximizing impact and access stands to benefit less-established researchers even more than more-established ones!

If [self-archiving] should help (or hurt) other categories or people, so be it, but it is neither its concern nor its worry. It is a tough-minded vision, narrowly focused on scientific communication. Supporters of this vision are essentially interested in only one thing: extracting every ounce of impact a published article may hope to claim.

And the above is a rather tough verdict -- but without giving even a clue of a clue as to who would be hurt by maximizing access and impact through OA self-archiving!

Having pointed out that all authors and users benefit, who are the 'players' who lose? Publishers? There is no evidence of that, just speculation (for which there is equally plausible counter-speculation that the system can adapt naturally if the need should ever arise). Librarians? How? In not providing them with a solution to the pricing/affordability problem? But we cannot solve all problems at once. World hunger continues too, and is more pressing. Moreover, one would think that library budgetary problems could only become less pressing, not more-so, in a 100% self-archiving world, where the supplementary OA version is available to all as a safety-net.

"Green" and "Gold" Open Access: Are They in Competition?

Various Internet lists (e.g., Liblicense-L discussion list or American Scientist Open Access Forum) have been the site of vigorous discussions about the two strategies identified in the original Budapest meeting and now regularly labeled as the "Green" and the "Gold" roads to Open Access. This colorful vocabulary emerged in a study led in the United Kingdom under the name of Rights Metadata for Open Archiving (RoMEO) and now located within another project called Securing a Hybrid Environment for Research Preservation and Access (SHERPA).6 Essentially, "Gold" refers to Open Access journals; Green refers to publishers that allow some form of article "self-archiving."

Sometimes shades of Green have been carefully distinguished: Pale Green limits "self-archiving" to preprints only, dotted, or some form of mitigated; Green limits "self-archiving" to postprints; and solid Green is reserved for publishers allowing both preprint and postprint "self-archiving." Publishers that allow no form of "self-archiving" are often described as Gray publishers (personally, I would have expected red but perhaps I am too influenced by traffic lights to the point of confusing "Gold" with orange).

Whatever the perspective adopted, the "Gold" and "Green" strategies are generally treated as parallel approaches by both sides, and little attention has been paid to the ways in which they might relate to one another.

The reason little attention is paid to how Green and Gold might relate and interact is that this calls for speculation, and the non-speculative facts are in far more urgent need of action. We need to promote both OA self-archiving and OA journal publishing (but in proportion to their capacities to deliver immediate OA, which are currently about 95 to 5, respectively).

One can speculate on the possible, eventual interaction between Green and Gold (and I confess I too have in the past done so), but speculating is not an optimal use of time when OA has been within reach for a decade that we have instead spent mostly speculating!

When perchance their relationship is addressed, it is tangentially and mainly in the suggestion that they might be in some form of competition. This has been particularly true of the "Green" side.

This requires some corrective context: What has actually happened is that the Gold side has for several years been receiving most of the attention, even though Gold can only deliver 5% immediate OA, and even though proponents of Gold tend to completely ignore the Green option -- to the point of speaking about and arguing for OA as if OA were synonymous with OA publishing! It is in this context -- but particularly because Green has the (unexploited and overlooked) power to generate immediate 95% OA that it has become necessary for the advocates of Green to compete for attention with the proponents of Gold! Competing for attention has required pointing out quite explicitly that devoting more attention and energy to Gold than to Green, instead of in proportion to their respective power to deliver immediate OA, is in fact disserving the interests of OA (because it is!).


Treating the "Green" and "Gold" approaches as separate and in competition, explicitly or implicitly, is not useful; worse, it is potentially divisive and could ultimately weaken the Open Access movement. 

The two approaches are in competition for whatever time, resources and energy we have to devote to OA. So far, that time, resources and energy have not been invested in Green and Gold in proportion to their respective capacity for providing a return on the investments -- i.e., their capacity for delivering immediate OA. That is not useful (for OA); and efforts to redistribute the available time, resources and energy stand to benefit OA. What could weaken the OA movement is failure to make progress toward OA, or needlessly heading in an inertial direction that can deliver far less OA than the alternative direction. I think the evidence and arguments for the respective probabilities and powers of Green and Gold need to be pointed out, rather than suppressed in an effort to preserve an ecumenism (and one-sidedness) that is far from optimal for OA.

Far from being essentially separate and in a potential state of competition for resources, I shall argue in this paper that the "Gold" and "Green" approaches can actually support each other, and ought to. Rather than favoring one approach exclusively at the expense of the other, Open Access promoters should design better strategies by making use of both approaches simultaneously. Only in this way can Open Access become a reality within a not too distant future. This is the challenge for this paper.

That sounds constructive, but it does seem to imply that the status quo is that Green is being favored at the expense of Gold, whereas the reality is quite the reverse: that Gold has been vastly favored at the expense of Green, for several years now! And Green is currently working to readjust the overall energy investment so it is more in proportion with each approach's immediate capacity to deliver OA.

Two very recent events help understand this issue and they have also provided an interesting backdrop to this whole question. For one, the appearance of "Scientific Publications Free for All," the Science and Technology Committee of the UK House of Commons7 immediately gave rise to a number of important comments and reactions. Less openly, but quite significantly nonetheless, the House Appropriation Committee in the United States, in its recommendations for the 2005 National Institutes of Health (NIH) budget, included language about the need for Open Access. That very language is reverberating within supporters as well as within publisher associations even as I write these lines.8 Together, they give fascinating insights into the ways in which Open Access is actually progressing.

The "Green" side has claimed to be particularly pleased with the UK Commons Select Committee Report, and it has certainly taken advantage of its publication to clamor its preference for "self-archiving."

Its summary of the UK Commons Select Committee has been presented in a way that Stevan Harnad, its most representative spokesperson, calls an "order of concreteness." This "order" really corresponds to the hierarchic scale of objectives favored by the "self-archiving" side. He summarizes the report's recommendations as follows:

1. Mandate author-institution self-archiving of all UK-funded research output (and fund and support the practice, as needed)

2. Fund author-institution costs of publishing in OA journals.

3. Encourage the transition to OA publishing and study it further.9

This reaction to the British Report also praises the members of the Select Committee for "getting it," as many would say colloquially. "Getting it" in this case really means that, according to the "Green" side, the Select Committee put the accent exactly where it should, namely, at the first point cited above; it also claims that the report has placed other possible characteristics of Open Access in a hierarchically inferior position. For the "self-archiving" side ˆ la Harnad, point one is all that Open Access really needs-a thesis he has constantly supported for about a decade now. He particularly praises the fact that this report's specific recommendation to Parliament is the only "mandatory" recommendation (hopefully this is not an oxymoron) while points two and three are presented as recommending recommendations (hopefully this is not a tautology). In short, the Report vindicates his own position, or so he claims.

So far, despite the needless irony, this is a correct summary of what I said.

(What is omitted is only the fact that the actual outcome of the report is very different from the language with which the inquiry was launched: The original Call for submitted evidence was 100% biased toward Gold, making no mention of Green whatsoever:  It took 'Open Access' to mean 'Open Access Publishing,' and called for evidence to be submitted on the question of the need to reform the publishing system. It must be noted [with no irony] that this original position would be far more congenial to J-CG -- and his own theory about what the problem is and what the solution needs to be -- than was the actual outcome, which was to recommend mandating Green and to merely encourage and fund further experimentation with Gold. This was indeed a surprising turn of events; it was clearly a result of the committee's evaluation of the actual evidence submitted in response to its Call, rather than just a re-assertion of its original terms; and it may represent a historical turning point in the fortunes of both Green and OA!)

Why is the "mandating" part so crucial? To answer this, a fairly long detour is needed. Let us begin by a precise outline of the "Green" argument:

1. Librarians initially blew the whistle on the fact that something was amiss in scientific communication when they began observing steep increases in the prices of journals.

The following summary is largely correct. I note here only that whereas librarians blew the whistle, researchers were already doing something about it, as of the advent of the Internet: they were already self-archiving.

2. They should be thanked for that, but alas, this particular angle of analysis also favored a certain degree of confusion between access and affordability.

3. While affordability has been the traditional, library-based, route to access, access can be analytically distinguished from affordability.10

4. In other words, access can be treated entirely separately from science publishing and its economic characteristics. This is the road that researchers (as distinguished from librarians) ought to follow.

5. If researchers carefully train their sights on the issue of access and nothing else, they can issue themselves the following challenges: How can one provide free access to refereed articles that are locked up behind a price barrier in the published, refereed, journals and that are owned by the publishers of these journals? How can this goal be achieved without relying on having them bought up by a proxy organization such as a library?

6. The solution to this problem begins with technology: Without digital versions of the articles and the Internet, the problem would have no practical solution. However, technology is only a necessary condition for the existence of a solution. Beyond the technology, human agency is also needed: Authors (mainly) are asked to "self-archive" their articles.

7. The point for authors is not to engage in some sort of civil disobedience, for example, by breaching intellectual property laws. Instead, one must either obtain permission to "self-archive" or find loopholes in intellectual property laws (and possibly journal policies). If explicit permission to "self-archive" is not available, one can still "self-archive" the article as submitted first to the publishing journal and then in a separate file "self-archive" the corrigenda that transform the submitted version into the actually published version.

8. More recently, a number of publishers have simply decided to allow authors to "self-archive" either preprints or postprints or both. The willing or "Green" publishers (all shades of Green conflated) control around 85% of the (surveyed11) scientific titles published in the world.12 The somewhat complicated maneuvers associated with point seven above are mentioned as little as possible. They remain necessary, however, for the Pale "Green" publishers who accept only the archiving of preprints.13

So far, so good! However, the issue becomes more contentious when the "self-archiving" side extends the argument to include the "Gold" road. It does so as follows:

9. The other possible approach to Open Access is through the publishing of Open Access journals (the "Gold" road).

10. A survey of the present situation reveals that Open Access journals cover around 5% of the titles (or number of articles) at best. It also shows a slower growth than the number of articles accessible in open repositories.

11. The reason for this is that the "Gold" road is costly, risky, and inefficient.14

12. Consequently, anyone genuinely interested in Open Access should recognize that supporting the "Gold" road is a somewhat ineffective effort at best.15 At worst, it delays success by diverting resources to an inferior strategy,16 thus intimating that the two roads, in actuality, compete for rare resources and that money should be diverted to the "Gold" road only in proportion to its (very limited) usefulness.

J-CG's summary 1-12 above is largely accurate. I would add only that it is not so much the funding that needs to be more rationally distributed to the two roads to OA but rather our time, attention and action. It is the Golden road that needs a lot of money (to create and support new OA journals, to fund author-institution OA publication charges, to encourage non-OA journals to convert to OA). The Green road hardly requires any money at all: Creating an institutional archive is extremely cheap (about a $1000 linux server, a couple of days of sysad start-up-time, and a couple of hours a month maintenance time -- plus the few dozen additional keystrokes per paper it takes to self-archive the paper (over and above the keystrokes it takes to write it and submit it for publication): ). Hence the resource the Green road competes for is not money, but action: authors need to perform those keystrokes, and their institutions and funders need to adopt policies that mandate that they do so (for their own good -- if OA is indeed the desideratum it is purported to be!).

So money is a red herring. What Green (and OA) needs is less rumination on Gold and its financing, and the long-term future of publishing, and more action on Green and the immediate future of OA (and access, and impact).

Although this argument appears watertight, it is pragmatically flawed. The problem with the self-archiving argument is that, until now at least, its results are unimpressive.

OA self-archiving's results to date are indeed unimpressive. I don't disagree at all -- but compared to what? Certainly not compared to the results of OA publishing, since OA self-archiving has generated 3 times as much OA as has OA publishing and is probably growing much faster too. Green is only unimpressive relative to its own immediate potential for generating OA, which is at least 93%, compared to Gold's 5%. In that respect, it can be said that Gold, at 5%, is much closer to its full immediate growth potential, whereas Green, at 15%, is not. But surely the remedy for that is to devote more time, attention and energy to exploiting Green's full immediate potential! That is what the impending self-archiving mandates will do. In the meanwhile, however, it would help if (1) less time and attention were devoted exclusively to Gold, as if OA publishing and OA were the same thing, and if (2) the Gold option were always balanced by pointing out the Green option too.

(Green has for several years now adopted the unified OA provision strategy: 'If there is a suitable Gold journal for your paper, publish it there; if not, publish it in a Green journal and self-archive it.' Just taking that step, of fairly presenting the two options at all times, would go a long way toward redressing the imbalance between Gold and Green.)

The reason is relatively simple to identify: The "self-archiving" side describes its own strategy as a smooth, yet anarchic, way to Open Access. Beyond the fact that smoothness and anarchy do not couple easily, we are going to see that it creates documentary lacunae that are fatal to the whole project.

(1) The fact that OA self-archiving grows anarchically, article by article, means that it is uncertain whether and when 100% of a particular journal is 100% OA. If journal OA instead grew in an all-or-none way, journal by journal, it would be easier to decide when and where to cancel.

(2) Self-archiving creates 'documentary lacunae'? A more theory-neutral way to describe it is as filling lacunae (with OA)!

As a result, librarians looking for credible alternatives, understandably, have not been convinced.

Librarians are looking for credible alternatives to what? and for the sake of what? Journal affordability? But researchers do not and will not provide OA to their articles for the sake of journal affordability -- though they just might possibly do it for the sake of maximizing the usage and impact of their articles. And researchers are the ones who need to be convinced, not librarians, as researchers are the only ones who can provide OA to their articles (whether by publishing them in a Gold journal or by publishing them in a Green journal and self-archiving them.)

Yet [librarians] often are the ones left with the duty of organizing institutional repositories.

The duty of organizing institutional repositories? All that needs to be done with institutional OA archives is to set them up (and sysads do that -- see above). Then the only remaining "duty" is to fill them -- and only researchers can do that (though librarians can certainly help!):

More important still, a majority of scientists have not been swayed either.

Not just more important: most important. Indeed the problem of 'swaying' researchers to provide the OA that they are purported to want and need so much is the only real challenge for OA. And it is already clear what will meet that challenge: (1) Empirical evidence of the OA impact advantage
plus (2) an OA self-archiving mandate on the part of researchers? institutions and research funders to ensure that advantage is taken of that advantage -- by naturally extending their existing "publish or perish" mandate to "publish and self-archive" (so as to maximize the access to, and the usage and impact of, your articles):

And we already know from a recent survey that just as they currently comply with their "publish or perish" mandate, most researchers report they will not self-archive if it is not mandated, but they will self-archive -- and self-archive willingly -- if ever it is mandated by their institutions or funders:

Before examining in more details why this is the case, let us revisit the issue of the relative importance of the two roads. This is important because, it seems to me, the situation is often portrayed in somewhat disingenuous terms. For example, the number of articles published in "Gold" journals (5%) - and these are actual numbers of Open Access articles - is often contrasted with the total number of articles published under "Green" titles (85% or more), without any mention of the fact that a majority of those are not actually and presently available in Open Access repositories.

On the contrary, it is always stated very explicitly (including in an article co-appearing in the very same issue as J-CG's article!) that whereas 93% of journals are Green, only 20% of articles are OA, 15% of them OA via self-archiving! This fact is not being concealed, it is being
trumpeted, in order to point out that if researchers really want and need OA and its benefits in terms of access and impact as much as they are described (by OA advocates of both the Green and Gold hue) as wanting and needing it, then it is up to researchers to provide it -- particularly where they have even been given their publisher's Green light to go ahead and do so!

But it is clear that just as far fewer researchers would publish anything at all (despite the advantages of publishing -- advantages that researchers presumably want and need) if it were not for their institutions' and research funders' "publish or perish" mandate to do so, so researchers will likewise not self-archive until their institutions and research funders make their employment, salary and research funding conditional on their doing so. (Institutions and funders already do this implicitly, in making researchers' employment, salary and research funding conditional not only on publication, but on the impact of publication. Since OA maximizes impact, this implicit causal connection and contingency now simply needs to be formalized explicitly.)

The reality is more modest. Harnad himself is more careful and generally speaks in terms of percentage of articles available to self-archiving; however, the direct quantitative comparison between "Gold" and "Green" is often implied, intimated, suggested, connoted, or whatever, in many of the discussions on Open Access. Harnad himself, when he faces this issue squarely, estimates the ratio between the "Green" archived articles and the "Gold" articles to be roughly three to one in favor of the former17-a result that, if real, is far from insignificant, but quite different from the 5:85% ratio.

This apparent inconsistency is very easily resolved: The Green/Gold ratio for actual OA is 3/1. The Green/Gold ratio for potential (immediate) OA is 95/5.

This said, a more fundamental problem remains: Why are repositories not growing at the rapid pace one could hope for.18 This is the topic of the next section.

The answer is exactly the same as if the question had been 'Why are publications not growing at the rapid pace one could hope for?' -- asked before the era of 'publish or perish': Because it needs to be mandated, for researchers' own good (and for the good of research itself).

But, to put this in context, there is no special question for OA self-archiving: Although 5% of journals are OA, many of them are still short on submissions. (Some BioMed Central journals publish only 5 articles per year.) So it is true of both Green and Gold that researchers are not yet taking full advantage of their potential. A critical difference, however, is that one can mandate OA self-archiving but one cannot mandate OA publishing -- for that would be to abrogate the author's right to choose which journal is most suitable for his paper (and that would most definitely meet with stout resistance from researchers!). Nor can one mandate that non-OA publishers become OA publishers. Researchers' institutions and funders can only mandate OA self-archiving -- or, as I have proposed, more ecumenically: they can mandate OA provision, where OA can be provided either by publishing in an OA journal if a suitable one exists (Gold), or otherwise by (OAA) publishing in a suitable non-OA journal and self-archiving the article (Green).

Open Access vs. Accessibility: A Potential Source of Confusion

Intuitively, the advantages of Open Access appear obvious: Better access should enhance more reading, and more reading should enhance more citations so that any right-thinking scientist ought to respond positively to such strong incitations. Spontaneously, he should rush and "self-archive." No mandating should even be needed. The reality, however, is a little different. Even defenders of "self-archiving" have had to admit this:

Institutional archives are being created, but need to be filled more quickly, by authors, with research journal papers. Attracting authors and their papers requires evidence of services that will improve the visibility and impact of their works.19

Correct, and we are gathering and disseminating the requisite evidence:

However, as noted, this evidence, and the probability of enhanced usage and impact to which they attest, are still not enough to induce a high or fast enough rate of OA-provision, just as the probability of the usage and impact that will result from publishing at all are not enough to induce publishing: The incentive to provide OA, like the incentive to publish, must be made explicit, as being among the formal conditions for employment, promotion and research funding, much the way both publishing and research impact are already among the conditions for employment, promotion and research funding.

By "evidence of services," the authors of this declaration presumably mean that the increased visibility and impact brought about by Open Access need to be made ... well ... more visible. Is it just a question of advocacy, or are there other factors come into play that make most scientists neglect the impact advantages linked with "self-archiving"?


It is a question of information and advocacy, but also of the need for measures to overcome researcher inertia.

Let us begin with the question of impact. Impact, let us remember, is generally measured by the total number of times a given article is cited from the moment of its publication.20 Discipline-based studies have now confirmed what common sense suggests. Open Access does create more opportunities for more downloads and more "reads"; these parameters, in turn, correlate positively with more citations. The first notable study in this regard was Steve Lawrence's article, which appeared in 2001 and which, thanks to the number of times it has been quoted, has itself enjoyed quite an impact.21 Lawrence concludes his study in the following manner:

Free online availability facilitates access in multiple ways ... To maximize impact, minimize redundancy, and speed scientific progress, authors and publishers should aim to make research easy to access.

Note in passing that Lawrence quietly moves from "free online availability" to "research easy to access." The two are not quite equivalent. The difference, as I am going to argue, amounts to a crucial distinction that must be drawn between Open Access and accessibility.

This distinction -- which sounds here like a distinction between accessibility and accessibility -- will turn out to be a distinction between accessibility and ease-of-access. Accessibility is a necessary precondition for ease-of-access, and inaccessibility is what OA is concerned to remedy; no amount of increase in the ease-of-access to the accessible will remedy the inaccessibility of the inaccessible.

Since the appearance of Lawrence's article, several other studies dealing with astrophysics, mathematics, or computer science have also underscored the impact advantage of articles placed in Open Access repositories.22 What emerges from these subject-based studies is that, all things being equal, Open Access articles do present a significant impact advantage over toll-gated articles. Impact coefficients of two to five have been mentioned, which is indeed impressive.

These results, I believe, should be broadly accepted and I strongly suspect that more studies will continue to bolster this important claim. However, we must also remember the "all things being equal" clause and once again carefully distinguish access from accessibility. The task now is to define "accessibility" as precisely as possible.

We generally oppose toll-gated access, i.e., access conditional upon sufficient financial resources, to Open Access situations; however, in practice a research scientist enjoys what amounts to "Open Access" to everything in his/her library

I don't think it is at all useful or instructive to speak of licensed institutional online access as "Open Access." OA means online access free for all and not only for those whose institutions have paid for the access. No institution can afford licensed online access to all 24,000 journals; hence OA is always for the sake of what is not accessible to any given institution because it cannot afford paid access. It does not help, in this regard, to speak of what is accessible to an institution online via licensed access as being OA. That simply muddies the waters.

- hopefully this [licensed access] is a significant fraction of the scientific literature. That is, after all, why libraries exist in the first place. How significant a fraction? This varies with each library and its financial resources, but Open Access it is, and thanks to the library.

OA is accordingly needed for that equally significant fraction of the journal literature that any institution cannot afford.

As a result, and from the users' perspective, genuinely "Open Access articles" actually compete with other documents that, although very costly, appear nevertheless to be in Open Access as well.

I cannot follow this at all. Where is the competition? If a given article is accessible to a user via licensed access and is also accessible free via OA (a self-archived version), what is competing with what, for what? The article benefits from all the usage it gets, in both versions. What is the problem here? (I think J-CG is implicitly thinking of OA journals competing with non-OA journals here, but what we are speaking of is self-archived versions of non-OA journal articles, and the notion of 'competition' simply makes no sense in that case.)

In effect, the end user, the scientist-as-reader, is being subsidized and thus benefits from a situation of artificial (and partial) Open Access.

I don't understand why this is being put in this rather tortuous way: The article is accessible to some of its potential users for a fee (the institutional license toll), and to the rest of its potential users for free (OA); that's all there is to it. From the user's standpoint, I can access some articles for (institutional) fee (toll), others (sometimes the same article) for free (OA).

What is the fuss about here? What is clarified by referring to ordinary toll-access (whether institutional subscription, site-license, or pay-per-view) as 'subsidized' access, and by referring to institutional toll-access as 'artificial' or 'partial' OA? That licensed access is being "subsidized: by institutional tolls. And what OA is about is what the institution cannot afford to subsidize through institutional tolls.

Obviously, this greatly distorts the market conditions and it artificially allows toll-gated articles better to compete with Open Access articles

None of this makes any sense! What has the market to do with this? And what is competing with what? Articles compete with each other for usage and impact, and the articles that can only be accessed via tolls lose to the articles that can also be accessed toll-free (i.e., are OA). The OA advantage is between articles, not between non-OA and OA versions of the same article. The comparison is always non-OA versus OA within the same journal and year, where OA includes the impact of both the non-OA version and the OA (self-archived) version of each OA article.

If one treats this straightforward access/impact metric as a pseudo-economic variable, the picture is simply confused, not clarified. (I suspect that here too J-CG is implicitly thinking about the competition between OA (Gold) and non-OA journals, not noticing that this does not make sense for the Green case of either (1) competition between OA and non-OA articles in the same non-OA journal, or (2) "competition" between the OA (self-archived) and non-OA versions of the same article!)   

- an ironic point that was obviously misunderstood by the drafters of the recent open letter to Dr. E. Zerhouni when they complained about undue governmental intrusion in the private sector.23 Without governmental intrusion (in the form of support for libraries which produce the conditions for subsidized readers), the whole business plan of most scientific publishers would simply collapse. In the present, distorted, market conditions, the competition between Open Access articles and toll-gated articles simply cannot be played out on the plane of price comparison; if it is to be played out at all, it will be on the plane of accessibility and value.

I have no idea why all this economic theorizing is obtruded into what -- without it -- was a rather simple, straightforward phenomenon: Toll access alone allows less access, usage and impact than free online access (OA). That's all there is to it; the rest is just gratuitous hermeneutics.



Let us start with accessibility. It is more complex than a mere opposition between open and toll-gated access. For example, it can involve the ease, including psychological ease, with which a reader both retrieves information and navigates in it. If Reed-Elsevier prefers flat rate, bundling approaches to pay-per-view tactics, it is to enhance the accessibility of its products, not their access. If price is an issue, a concern, each time an article is accessed, use is inherently deterred because the reader is inhibited by constantly thinking about costs. As a result, accessibility may actually decrease while access remains constant.

Compare: "accessibility can decrease while access remains constant" with (my gloss): "ease-of-access may decrease while accessibility [i.e., possibility-of-access] remains constant."

'ease-of-access may decrease while accessibility [i.e., possibility-of-access] remains constant.'

Articles, whether published by Elsevier or anyone else, are more accessible, and have higher usage and impact, if there is a free online version of them, in addition to the toll-access version. That is all there is to it, and that is all that is meant by accessibility: Toll access versions are accessible only to those users whose institutions can pay the tolls. OA versions are accessible to everyone.

'Bundling' concerns ease-of-access, not accessibility; and if/when there is enough OA content (and not the mere 20% there is now), then that can be bundled too. And Firefox can make a back-end that -- like  but automatically and silently -- performs an OA version of any search being done on a bundled toll-access database, as well as seeking an OA version of any hits yielded by the toll-access search.

J-CG is simply underestimating the power of this medium on account of its not yet having been more fully mobilized (simply because the OA content has so far been too thin to warrant or reward the effort). The priority now is obviously to increase the OA content (the prospect that J-CG is here both minimizing and misunderstanding), and then the functionality will quite naturally follow.

From another perspective, a significant part of Andrew Odlyzko's paper (cited in note22) actually deals more with the ways in which accessibility can be improved than with Open Access per se. These ways include factors apparently as trivial as the amount of time needed to reach a document-and differences measured in minutes have been shown to be quite significant. Delays in access drastically reduce use even though access per se is not modified.

Agreed, but irrelevant to the issue at hand, which (I take it) is non-OA vs OA.

Yet another way to approach the question of accessibility is to ask: what is more accessible? A large collection of articles licensed by a library (or a consortium of libraries) that is readily exploitable through an easy-to-use, easy-to-reach, portal, or scattered collections of open access articles, more or less systematically (but how systematically?) harvested, and perhaps drowned in collections of very uneven value. Is this not the situation that presently emerges with OAIster, for example?

This question is too vague: What percentage of the literature are we imagining to be OA, in comparing OAIster to a licensed collection? Harvesting and indexing can and will be improved once there is more OA content. (Right now it would be ridiculous to invest in improving OAIster as a precondition for providing more OA content! In principle, OAIster can be made exactly as convenient and functional as a licensed collection -- with the added benefit that it can cover 100% of the 24,000 journals (and not just the fraction for which we can afford licensed access) and that it is accessible to everyone (not just those who can afford license fees).

But it does not stop there: Once OAIster has enough OA content to make it worthwhile, OAIster search can easily be integrated with searches of licensed content, via software (as described earlier).

So these are all just spurious and arbitrary comparisons, and they neither contradict nor cast further light on the simple fact that articles that are accessible only via toll-access are for that reason less accessible, and hence have less impact, than articles for which toll-access is supplemented by a self-archived OA version.

Clifford Lynch writes something very important in this regard:

'I think we are very shortly going to cross a sort of critical mass boundary where those publications that are not instantly available in full-text will become kind of second-rate in a sense, not because their quality is low, but just because people will prefer the accessibility.'24

It is not clear whether Clifford Lynch is speaking here about instant online licensed access (to those who can afford it), or Open Access. Either way, this has nothing to do with the fact that OA maximizes access and impact.

In the same pragmatic spirit of ease of use, Andrew Odlyzko reveals a trade secret that I have also given to my own students (and I thought I was so smart): if you search information about recent books, use Amazon; it is far better and much more user-friendly than any library system. This tactic is based on a feel for accessibility rather than a concern for access.25 Both Odlyzko and Lynch are talking about accessibility, not Open Access per se.

What is this distinction between access and accessibility? If I can access it online, it is more accessible than if I cannot access it online, whether I access it by non-OA or OA. But if it can be accessed by OA it is more accessible, to more users, than if it can only be accessed by non-OA.

(Amazon only lets you access the metadata and a few pages of the book anyway, so what is the point here? We are concerned with full-text online access to journal articles. And what does this have to do with the user-friendliness of Amazon vs. the user-friendliness of library-based systems? or their respective scope of coverage? This is all very hirsute, and one senses that it must be driven by a theory, for otherwise it is just needlessly complicating and obscuring very simple phenomena.)

The point of all this is that accessibility is wider than Open Access and encompasses it; Open Access derives its real value from its ability to improve accessibility.

The accessibility/access distinction is so far completely without content. The only distinction that makes sense is limited access (non-OA; i.e., <100%) versus unlimited access (OA: 100%). At the single article level, this means accessible-to-some (non-OA) versus accessible-to-all (OA). There is no more to be said here; the points about ease-of-access and design of search engines or interfaces are irrelevant.

However, other approaches can also improve accessibility. Yet, if all other things are equal, Open Access will come ahead of toll-gated publishing.

But we knew all this already, before this needless detour through the accessibility/access non-distinction!

However, if toll-gated access is artificially subsidized as it is presently,

Why are we talking about toll-access (institutional subscriptions, site licenses, pay-per-view) as being 'artificially subsidized'? The tolls are being paid by institutions. That's all there is to it. And OA is needed for all the users at institutions that cannot afford the toll-access.

and if commercial publishers design good retrieval and navigational tools, then Open Access documents may actually look less attractive to scientists than their commercial counterparts.

Toll-based access will continue to look attractive to those who can afford it. But no matter how attractive or easy-to-use its retrieval and navigational tools, they are useless to those who cannot afford it. And that is what OA is for.

"All other things being equal": there is the rub! Open Access has to contend with more than toll-gated articles; it must also compete with various enhancements to accessibility.

OA is not competing! (OA journals may be competing with non-OA journals, but we are speaking here about OA, not OA journals. Is J-CG so much in the grip of an economic/sociological theory that he cannot think of OA as anything but OA publishing?)

OA is competing with neither non-OA nor with enhanced non-OA accessibility tools.

OA is competing with neither non-OA nor with enhanced non-OA accessibility tools. OA is not competing. It is complementing: It is providing access for those who can afford neither the non-OA tolls, nor (a-fortiori) their accompanying enhanced ease-of-access tools.

And let us remember the bitter irony of the situation: The very librarians who profess pro-Open Access positions are presently working very hard to ensure that toll-gated articles may enjoy an even playing field with Open Access articles by artificially removing all economic barriers to the reading scientist.

This becomes more and more baroque: Librarians who are pro-OA are presumably helping to promote Gold or Green OA or both. They are also continuing to pay for whatever non-OA they can afford. That is clear and quite natural. But what is this about an 'even playing field' for non-OA and OA articles? What are they playing at or competing about or for? Non-OA articles get only the would-be users whose institutions can afford access; OA articles get all would-be users, and vice-versa. OA journal articles may be competing with non-OA journal articles; but we were speaking here about OA, were we not, rather than collapsing everything again into just OA publishing alone?

Yes, librarians have to keep on buying in non-OA journals, even if they would love to see all journals become OA journals, because only 5% of journals are OA journals, and 95% are not. That is not irony; that is reality. And the remedy is to try to think of another way to reach 100% OA than to wait for 95% of journals to convert to it!

No wonder if the library profession sometimes appears caught in a prisoner's dilemma, as Ken Frazier put it so aptly a while back while discussing some of the downsides of the "big deals."26

The library profession is not caught in a prisoner's dilemma. They are trying to buy in the access they can afford, and 95% of journals are non-OA, so those journals (or rather the fraction of them that any given library can afford) need to continue to be bought in. And librarians are not themselves in a position to provide OA by any means: They are not the authors, nor are they the publishers; nor do they have a lot of spare cash to cover OA journal publishing costs. So they are not in a prisoner's dilemma; they are merely trying to keep on making ends meet from year to year. If they have any time to spare, that time is best spent trying to promote institutional self-archiving, for if that practice spreads, librarians will not only have helped in facilitating the provision of their own institutional OA output, but they and their users will become the beneficiaries of other institutions' OA output. Time much better spent than just trying to promote OA journals.

Commercial and some association publishers have been quick to seize and capitalize upon new possibilities offered by the digital world, in particular the capacity to move seamlessly from a bibliographic tool to a full-text article or from a citation to the cited text. In short, commercial publishers seem to have read Clifford Lynch closely and taken his advice very seriously. Ventures such as CrossRef, Ex Libris SFX, and others27 aim at creating smooth navigational spaces that enhance accessibility; meanwhile, the user remains largely and comfortably blind to the costs of this process thanks to those (oh so discrete!) librarians.

This is all completely irrelevant: It applies only to the non-OA literature that an institution can afford to buy in, whereas OA is about the literature it cannot afford to buy in.

On the Open Access side, similar efforts are being pursued to improve the accessibility of peer-reviewed research papers. It must be remembered, however, that any significant advance on the accessibility front on the Open Access side will quickly be taken up by the toll-gated side as well because, to be effective, such an advance must be openly available to all. The reverse, however, is not true. Tools useful for the stitching together of disconnected archives may be proprietary, putting Open Access endeavors at an economic disadvantage.28

It is again completely irrelevant that OA resources (both articles and search tools) are accessible to all, whereas non-OA resources (both articles and search tools) are accessible only to those who can afford them. The only relevant thing is that OA complements non-OA for all who cannot afford the non-OA. J-CG here seems to be spiraling higher and higher in theory-driven epicycles that have nothing whatsoever to do with OA or what OA is needed for (which is not to reform journals or remedy affordability problems but to provide access to all who cannot afford it, so as to maximize impact).

[I suspect that what J-CG once again has in mind here in referring to the two 'sides' is the competition between OA journals and non-OA journals, not between OA versions of articles (self-archived from non-OA journals) and non-OA versions of those same articles, or with non-OA articles that have no OA version (because they have not been self-archived). There is simply no relevant competition to speak of in the case of articles!]

Rival toll-based OAIsters locking up OA content? They're welcome to try, but I'd bet on the ingenuity of the free OAI-service providers beating that of the toll-based ones every time... (But we are counting our chickens -- toll-based and free services -- before the OA eggs are laid!)

A case in point is the Open Archive Initiative-Protocol for Metadata Harvesting... (OAI-PMH). OAI-PMH is absolutely essential for the Open Access depositories because they are evolving in a completely decentralized fashion, but OAI-PMH is equally applicable to open and closed collections29 as Carl Lagoze sometimes explains in his conferences on OAI-PMH.30 This is because OAI-PMH really deals with accessibility issues, not with Open Access per se.

Not quite: OAI-PMH does not deal with 'accessibility' issues, it deals with interoperability (including ease-of-access) issues, and interoperability (of metadata) is completely indifferent as to the accessibility of the full-texts -- i.e., as to whether the full-texts are OA or non-OA.

(Amongst the possibilities for OAI metadata, there is of course also that of tagging the rights associated with a digital object:
That measure is -- like the Creative Commons License itself
-- welcome, but not necessary, in the special case of OA self-archived versions of published journal articles, which continue to be protected by the publisher's copyright agreement.)

Accessibility tools such as ParaCite exist and they have appeared on the Open Access side, but ParaCite is still experimental. Therefore, commercial offerings to improve accessibility appear more developed than the tools available for Open Access repositories.

Correct. But what OA needs now is more OA, not more tools for the limited OA there is to date. The enhanced tools will come with the territory, as the percentage OA increases. In contrast, there are plenty of incentives for perfecting tools on the non-OA literature, which is far bigger (indeed, by definition, complete, 100%).

As a result, thanks to the subsidized reading context of scientific publications, commercial publishers can credibly defend the argument that their literature, although toll-gated, is more accessible to researchers today than are articles placed in Open Access, at least in the richest institutions of the richest countries.

This argument can only be defended if we use this rather arbitrary ostensive definition of "accessibility," which seems roughly equivalent to "ease of access". But, to repeat: No degree of ease of access is of any use to those users who cannot afford it! So enhanced non-OA tools are simply irrelevant to OA.

(It is also no doubt the case that in the online age, more articles are accessible to more users, via non-OA, than were ever accessible previously. This too is true, but irrelevant, because OA is about the remaining articles that still cannot be accessed, and about the users that still cannot access them.)

How do scientists, research institutions, or granting agencies react to the issue of accessibility?

How they react to the issue of ease-of-paid-access  or to the issue of access, simpliciter?

How they react as users (of what their institutions can and cannot afford to access)?

Or how they react as authors (of what would-be users at other institutions can and cannot afford to access)?

All of them obviously want to maximize impact, but they may not want to do it in exactly the same way or for the same reasons. A granting agency, especially if it is publicly financed, likes to demonstrate a degree of public service beyond the support of research scientists. For such an institution, impact will mean more than getting citations; it will also mean demonstrating that Open Access increases the number of readers and attracts individuals from wider walks of life (various levels of education well below the research level, for example, or patients in the case of medical research). As many of these different kinds of readers will not have the benefit of a research library, their access to the scientific literature is not artificially subsidized. This means that without Open Access, it is simply inaccessible.

It is equally true of all potential users -- whether researcher or general public -- who cannot afford the non-OA versions, that what is not OA they cannot access. So what is the point here? The funder wants to maximize usage and impact, and OA will ensure that. And self-archiving can provide immediate 100% OA. ('Impact,' by the way, does not just refer to citation impact: There is now also measurable download-impact, and still richer impact indicators will emerge as the full-text OA corpus grows.)

A private foundation will not react very differently: it too wants to enhance its social function (and public image). Much of the discussions around the granting agencies [Wellcome Trust, Hughes Foundation, Max Planck Gesellschaft in Germany, INSERM (The French Institute of Health and Medical Research) in France, and more recently, NIH) confirm this favorable attitude. A research center, especially if it is publicly supported, or a university will also listen to these kinds of arguments. All will tend to view Open Access in a positive light. Indeed, a simple examination of the list of signatories to the Budapest Open Access Initiative shows widespread acceptance of Open Access ideas within the universities and the research centers.31

(It is not apparent what the point is here. This analysis was meant to be about the relation of Green to Gold. Instead we are being reminded here of why we need OA at all, and that it is to maximize access and impact. But are not all agreed on that point?)


Scientific associations often display ambivalence to Open Access. Most of the time it is for economic reasons: Demonstrate to us, they say, that a good business plan exists for Open Access and we will consider it.

Scientific associations are not (and cannot be) against OA itself. Many are against OA publishing (Gold), because it puts their revenues at risk. Some (fewer and fewer) are also against self-archiving (Green), because they think that even that might threaten their revenues (but the number of publishers who think this way is shrinking fast).

But other, more surprising, objections are sometimes raised as when the Royal Society of Chemistry claims that scientists often favor a limited number of "quality" readers and laboratories over maximum dissemination.32

The Royal Society of Chemistry is now Green

So the above demurral can only be about not wanting to convert to Gold, not about OA in general.

However surprisingly the issue of accessibility is recast, it recurs nonetheless.

I cannot for the life of me figure out how and why J-CG construes non-OA publishers' disinclination to convert to Gold as having to do with 'the issue of accessibility' -- particularly in J-CG's sense of that word ('ease-of-access'), and particularly when the non-OA publisher is already Green!

Where the issue appears more complicated is at the level of the researchers. The lack of enthusiasm for institutional repositories displayed by scientists and scholars is an interesting symptom.

Researcher sluggishness in providing OA to their articles is in fact the only interesting symptom (of why 100% immediate OA -- which has already been reachable for over a decade now -- has been so slow in coming); and it is also the only real obstacle to 100% OA. The symptom is probably closely related to whatever made it necessary to create publish-or-perish  carrots-and-sticks in order to get scientists and scholars to publish at all. But there are also at least 32 groundless Worries that have been holding back OA provision despite the fact that each Worry has been easily (and repeatedly, and decisively) rebutted:

The justifications that scientists sometimes use to express skepticism can be a little surprising, as when authors advance the spurious fear of "information overload" argument.

That's Worry #4 of the 32 Worries, (and it pertains to OA, not just to OA self-archiving)...

But "information overload" is not really the issue: Open Access can accommodate filters, hierarchies, and branding just as well as toll-gated contexts.

And that's the rebuttal to Worry #4:

BioMed Central and the Public Library of Science (PloS) offer good examples in this regard. In fact, the whole point of the Public Library of Science is to demonstrate that extremely high ratings can be achieved with an Open Access journal, and Open Access journals therefore can help focus reading as well as toll-gated publications.33 Value in scientific publishing is measured by content, not by price. These fears appear even more pronounced on the "self-archiving" side, but in reality, they are just as imaginary: The traditional value signals are still operational since "self-archived" articles are peer reviewed and therefore can exhibit a title with some branding ability.

Self-archived journal articles are not only peer-reviewed: they continue to bear the name of the non-OA journal that published them! (The 'branding' issue will hence turn out to be a complete red herring for self-archiving! Non-OA journals already have their names, i.e., their 'brands. OA self-archiving hence does not have any branding problem. It is only new journal start-ups [whether OA or non-OA] that have an [initial] branding problem, while they are still new and trying to establish their quality standards and impact factors.)

The factors that inhibit the progress of Open Access obviously lie elsewhere.

A partial explanation to this puzzle may be found in a remark recently expressed by Michael Kurtz. In a note devoted to the positive (and important) correlation between "reads" and number of citations, he concludes by a little remark that has not attracted much attention so far:

'The fact that many of the inaccessible papers are in the ArXiv probably does not change this much, as the additional effort involved [from leaving ADS's34 unified resource to go to another system] is a great deterrent.'

What Kurtz is alluding to is that tools providing the standard, accepted, research pathways and corresponding to the accepted research tactics also provide a level of accessibility related habits and these are not changed easily or lightly.

Astrophysicists essentially use one single source to do their research.

It also needs to be pointed out that (as Kurtz notes) astrophysics is unique in that it has virtually 100% access already -- not through Gold and not through Green either (although a great deal of astrophysics is self-archived too) but because astrophysics has a small, closed circle of journals and all active astrophysicists worldwide are at institutions that can afford licensed access to all of them.

Sometimes some of the articles found in this manner cannot be accessed within the familiar research environment. Could these articles have been accessed nonetheless? In some cases, yes, states Kurtz, and simply by going to ArXiv; however, doing so would require changing the search context (and habits) of scientists and would force them to move beyond their favored research aid, in this case ADS. Kurtz' comments on this point is quite simple and direct: "the additional effort involved is a great deterrent."35 In other words, even when articles can be accessed, a significant difference in accessibility is sufficient to reduce usage.

My interpretation of this would be that in an anomalous field -- unrepresentative because all researchers have almost 100% access to everything they need already -- users will often not make the extra effort of consulting another accessible source even though that means something might be missed. This luxury (or laziness) does not translate to most other fields, which are far from 100% OA, and where the OA version is a mainstay rather than just a minor supplement with small marginal utility.

It must also be added, however, that even in astrophysics, with its near-100% access, articles that are also self-archived in ArXiv still show an impact advantage: 

So some astrophysicists still make the extra effort after all.

Librarians know Michael Kurtz' point well: being formally able to access a document is simply not enough; the availability of attention (and therefore time) must also be taken into consideration. In particular, if, through some familiar method, a scientist finds what appears to be "enough" information and does so within a limited amount of time, chances are that the search will stop there.

The evidence that this is not true in general, especially where access is nowhere near 100%, is the consistent and sizable OA impact advantage observed across all fields. At the very least, the higher number of downloads and citations for articles that have been self-archived over those that have not is evidence that users who cannot access the non-OA versions do make whatever added effort may be involved in order to access the OA versions:  

The point of all this is that a typical scientist seeking information is a prime subject for what has been called the "attention economy."36 In a world with enormous amounts of information, the limiting factor is not information itself; it is the processing capacity of the brain multiplied by the time that can be devoted to a particular task, reading for example. Typically, a scientist or scholar will begin interrogating some bibliographic tool, e.g., the Web of Science. In the ideal situation, the scientist would immediately access the search results by simply clicking on the references. Often, this is not possible simply because the local library does not have a subscription to the relevant journal. At that point, rather than trying systematically to gather the whole collection of relevant articles, the scientist reads what is most readily and rapidly available. After all, the information found in that incomplete collection of articles may be enough. The rest is then neglected unless really glaring gaps subsist. The eyes of the beholder are crucial here.

Again, the empirical fact that the usage and impact of self-archived articles is consistently higher than that of non-self-archived ones, across all fields, is evidence against this (although laziness might explain why the OA advantage is not higher still). And once there is more OA, the ease-of-access will no doubt be improved, increasing the OA advantage still further -- until 100% OA is reached, when the OA/non-OA advantage will of course vanish, although usage will remain uniformly higher for all articles, and the decision about what to cite will be based more equitably on merit rather than mere affordability (i.e., accessibility).

Andrew Odlyzko does not say anything fundamentally different when he concludes:

Also, the reactions to even slight barriers to usage suggest that even high-quality scholarly papers are not irreplaceable. Readers are faced with a 'river of knowledge' that allows them to select among a multitude of sources, and to find near substitutes when necessary.37

But Odlyzko is referring to online vs. on-paper here, not to non-OA (licensed toll-access) vs. OA. And when he does refer to non-OA vs. OA, it is to neglecting non-OA in favor of OA (even if incomplete), not the reverse.

Odlyzko's remark suggests that a scientist or a scholar will typically find enough useful information to justify writing what he/she wants to write rather than first researching a field very carefully in order to survey what he/she can usefully add to the field. This may well be one of the unexpected (and not entirely welcome) consequences of the "publish or perish" institutional atmosphere. It may also derive from Bradford's law of scattering: Exhaustively gathering the literature on any question would require almost infinite time and resources.38

We are far from the possibility of exhaustive access and search (except perhaps in astrophysics), so there is no point talking about it now. What we have now is something varying from about 30% - 70% paid institutional access to the non-OA corpus (95%) depending on the budget of the institution, plus about 1% - 25% OA (depending on the field). So we are talking about sufficient search, not exhaustive search. J-CG seems to be interpreting Odlyzko's remarks as implying that sufficient search means relying on the available non-OA corpus and neglecting what is available OA. But the evidence (from the OA usage/impact studies) contradicts this (nor is this what Odlyzko is arguing).

For the sake of the argument, let us assume that the information found in a first hunting expedition is not sufficient. At this stage of his/her work, the author will probably adopt one among the following tactics. The traditional approach (at least from the librarian's standpoint) would be to make use of interlibrary loans. Since this is a relatively time-consuming operation, a scientist may decide to proceed more directly. If there is an Open Access repository in a relevant discipline, he or she may take a look at it. But please note two details: First, the scientist's second move is at best a second order recourse; furthermore, in most disciplines, such repositories simply do not exist and institutional repositories cannot substitute for them.

First, this is not how users search for OA content -- repository by repository.

Second, there is no need for disciplinary repositories, any more than there is a need for disciplinary indexes: ISI's Web of Knowledge, for example, serves all disciplines. It is via subject and journal-name tags (and full-text indexing) that journal contents are partitioned in the online age, not by putting them in a different geographic locus.

Third, institutional OA archives, for each institution's own published article output, in all its disciplines, are the natural 'feeders' of the OA full-texts whose metadata can then be harvested by the OAI search and indexing services like OAIster. 

Fourth, for anyone lacking access, the OA version is not a second-order recourse, it is the only recourse.

Fifth, the data on the enhanced usage and impact of self-archived vs. non-self-archived articles (in the same journals and years) are a-posteriori evidence that users actually do make use of that recourse, despite the a-priori theorizing above that would imply they may not.

Finally, search engines for Open Access collections are not widely known among researchers. And when they are known, they are often considered with some degree of skepticism.39

What we should perhaps be considering with a degree of skepticism are the pronouncements of librarians on what researchers are doing at their desktops! The objective data on OA usage and impact suggest that the OA search engines and 'collections' are not as unknown as all that! And increasing OA content from its current 20% level just might make them more widely known -- and might inspire them to provide enhanced functionality, in keeping with that enhanced OA content.

In general, and this is the fundamental obstacle, Open Access articles are not yet sufficiently part of existing research strategies.

Why should OA articles form a bigger part of existing research strategies when OA content is still so small: a pathetic 20%! Instead of interpreting this current low percentage of OA content as evidence for OA's dysfunctionality, we should work to increase its functionality -- increasing the percentage of OA content: by providing it (though self-archiving)!

And the consequence is direct: If Open Access repositories do not appear very visible and/or credible within a given arsenal of research strategies, why should a scientist spend time to "self-archive" his/her works in what can only look like a dump - OAI-PMH notwithstanding?

J-CG has both the priorities and the causality exactly reversed: Researchers should self-archive to maximize the impact of their own work; this will increase OA content and generate more OA services and functionality.

[J-CG is making these backwards arguments as if they were arguments against Green OA, because (for theoretical reasons) he prefers Gold OA (in order to reform the publishing system, solve the affordability problem, etc.). But the logic of these arguments (such as it is) applies equally against OA itself, whether Green or Gold: "Since there is so little OA visible, it will not be used, so why bother with OA at all?" Again, this has all the symptoms of being in the thrall of a theory, and having to tie oneself into knots in order to support the theory. Isn't it simpler to say: OA is desirable and beneficial, so we should endeavor to increase OA from 20% to 100% by all means available -- Gold and Green -- each in proportion to its immediate potential to provide OA?]

Of course, there is still the recourse to brute force and this is what the use of Google is all about. I can imagine many librarians' eyes rolling at this point-how can one trust the results acquired through Google, goes the mantra-but not only is this approach common, it actually works rather well.40 What does Google yield? It may lead to some Open Access site, such as a personal page or an institutional repository.

Google (or Google Scholar) can only yield the OA that authors have provided. That's just one more reason authors should self-archive (and their institutions and funders should adopt the carrot/stick policies that will induce them to do it)!

Google's discussions with the DSpace network also demonstrate that the owners of the famous search engine are aware of the possibilities. It may also lead to an e-mail address: from there, requesting the interesting article directly is easy. From personal experience, I know this approach works rather well; yet, it has little to do with Open Access but has much more to do with the extraordinary facilitation of communication-accessibility all over again-the Internet provides. E-mail is still the "killer app" of the Internet and improves the accessibility of articles, toll-gated or not.

(This sounds rather vieux-jeu: Is it not more sensible to self-archive, once and for all, rather than to answer countless email eprint requests?)

As already mentioned, there are steps I do not see being taken very naturally by anyone at this stage of Open Access development: going to OAIster, for example.41 Why are researchers not using OAIster as a matter of course? The reason is quite simple. Although they would find a collection of texts - a large collection of some 3,420,891 records from 327 institutions - they would also discover that the value of these documents is difficult to ascertain.42

Here's a much simpler answer as to why OAIster is not used more: because OAIster still has so little OA (as opposed to just OAI) content! That means peer-reviewed journal articles. When you 'hit' one of those, you know it's OA, because it has the journal name. 'Journal-name' is a potential OAI metadata tag. So is 'peer-reviewed.' And OAIster could easily display it and make it searchable -- but there's hardly any point now, when the real problem is the pathetically small amount of OA content so far!

(Why keep focusing on the suboptimal searchability of next-to-nothing, instead of on transforming that next-to-nothing into something worth designing more optimal search tools for? And why keep focusing on the users of the little OA that there is so far, instead of on the authors of the much more OA that there still isn't?)

Even though OAIster limits itself to academic institutions, the value of what can be found in such repositories remains unclear. On this point, Harnad is completely right when he recommends building archives explicitly limited to peer-reviewed articles.43

My recommendation was that institutional OA Archives should focus on making their own peer-reviewed article output OA, now, by self-archiving it in their own institutional OAI-compliant OA Archives. As far as searchability is concerned, it is metadata tags (like 'journal-name' and 'peer-reviewed'), not archive locus or focus, that will sort things out optimally.

A scientist's search is already complex and uncertain enough as it is. There is no need to burden it further with the noise from other academic activities such as teaching, lab reports, and Gray literature. The case of preprints is interesting here, however Green they may be, because they too fall in the same murky category and should be stored separately.

Why "stored separately" rather than OAI-tagged appropriately ("not peer-reviewed" vs. "peer-reviewed" plus "journal-name")?

I must say that the view that email is still the "killer-app" does sound a bit out of tune with both the possibilities and the optimalities of the online medium! And as we shall soon see, just as J-CG keeps being driven by his own theorizing to think in terms of a "competition" between OA and non-OA articles (when this only makes sense for OA journal articles vs. non-OA journal articles, but makes no sense for OA versions of non-OA journal articles), so he keeps being driven by his own theorizing to thinking in terms of a need to add "value and branding" to self-archived articles (when this only makes sense for the self-archived versions of unrefereed preprints [which are virtually all destined for peer-reviewed journals], but makes no sense at all for OA's primary target: the self-archived versions peer-reviewed articles, already published in journals [whether OA or non-OA]).

One of the basic difficulties of "self-archiving" is that, given its necessarily distributed nature - consequence of the anarchic nature of the process - it becomes very difficult to mandate the form in which self-archiving will actually take place. For institutional repositories, the urge to fill it rapidly may translate in motley collections of documents that will serve no one.

Distributedness is a virtue, not a vice, on the Web. It is part of the intrinsic nature and power of the Internet itself. There is no need to mandate the 'form' of self-archiving beyond the stipulation that it needs to be done in an OAI-compliant institutional OA archive, and that it is peer-reviewed journal articles that must be self-archived. Self-archiving (like the Web itself) is anarchic globally, but an institutional policy can make it systematic locally, at the institutional level:

The success of repositories will be much more probable when scientists know better what to expect. Then scientists may decide to spend some of their valuable time hunting through these collections. In the case of institutional repositories, I do not remember ever seeing a study discussing their effect on impact. This lack of evidence creates a climate of uncertainty and may also account for the hesitations marked by scholars and scientists. If they are not intimately convinced that there is a clear and present advantage to "self-archiving," they will simply go by the constraints of an "attention economy" and forfeit going that extra step.

The evidence for the OA impact advantage is there, clear and present and empirical! Those findings will be widely publicized to researchers, but they will still not be enough to make enough researchers self-archive. Institutional and research-funder OA self-archiving mandates are needed.

As to the strategic advantages of local institutional self-archiving over central self-archiving, see:

Alma Swan, Paul Needham, Steve Probets, Adrienne Muir, Anne O'Brien, Charles

Oppenheim, Rachel Hardy, Fytton Rowland and Sheridan Brown (2005).

Developing a model for e-prints and open access journal content for UK

higher and further education. Learned Publishing, 18 (1), 25-40.

And, again, the problem not the users of what (little) OA there is so far, but the authors of the (much) OA there isn't, yet!

This efficient use of time, sometime labeled as "inertia," will be even more tempting in the case of a Pale green publisher where the procedure to self-archive is so much more complex as to become totally unrealistic.

Only 20% of articles are OA, even though 80% of journals are Full-Green! Why is J-CG focusing on the (putative) complexities of the 13% that are Pale-Green (or the 7% that are Gray) rather than on the 80% of journals for which the self-archiving procedure is straightforward?

We shall soon find out: For J-CG has a theory about how to replace the journal publishing system with an alternative system, and that theory depends on being able to build the alternative system on top of unrefereed preprints, not peer-reviewed, published postprints. Hence it is necessary to construe Green as being the self-archiving of preprints rather than postprints. The fact that 80% of journals have already given their Green light to postprint self-archiving -- and that BOAI-1 has always primarily targeted the published, peer-reviewed versions of journal articles, rather than just their preprints -- must accordingly be minimized (in the service of the theory).

Indeed, let us ask some crucial questions in this regard:

1. How many authors will go through the tedious exercise of creating the corrigenda allowing moving from the submitted paper to the published paper? If people do not appear ready simply to "self-archive" their postprints when they have a Green light to do so from the publishers, they will be even less ready to "self-archive" their preprints plus the corrigenda, especially if they harbor real or imagined fears about possible negative reactions from publishers and editors (i.e., powerful colleagues).44

To repeat, the vast majority (80%) of journal articles can already be self-archived without needing to give such complications a thought: so why this needless preoccupation with the more complicated 20% minority (when all of self-archiving has still barely reached 15%!)?

(Moreover, physicists have been uncomplicatedly self-archiving successive versions, adding corrections, for over a decade now: So self-archiving preprints plus corrigenda is eminently feasible. The preprint+corrigenda strategy was in any case formulated mostly as a reductio-ad-absurdum for Worry #10 of the 32 prima-facie Worries to which non-self-archivers were attributing their non-self-archiving!)

It was

2. How many readers will go through the tedium of making sure they have the right statement to use and cite in their own work when they have to deal with a main file and a list of corrigenda?

Does J-CG mean: Yow many would-be users who cannot afford access to the non-OA version at all will be scholarly about the OA version that they do access, in the minority of cases when it is only the corrected preprint to which they have access? Let's leave that to historians of scholarship to investigate and report to us when this is all over. In the meanwhile, can we get on with the far more urgent and important task of reaching 100% OA?

Given all the issues discussed here, in particular the question of accessibility (as distinguished from access),

The access/accessibility distinction is incoherent. The access/ease-of-access issue is irrelevant.

it becomes pretty obvious why "self-archiving" in simple institutional repositories will not be enough to create a really Open Access science communication system, even with OAI-PMH present.

What is obvious is that only 20% of articles are OA, 15% of them via self-archiving -- and that that is far from enough, and still growing too slowly. But that we knew already. What we have here are a few post-hoc conjectures as to why this might be the case. I find most of the conjectures off the mark, and based on misunderstandings about web use and functionality (and perhaps also about scholarly/scientific practice and needs).

Far more likely than these conjectures is the direct testimony of researchers themselves, who state that they will not self-archive till they are required by their institutions and research funders to do so -- but then the vast majority say they will do so, and do so willingly. In other words, it is the same as with requiring them to publish (or perish): Incentives are needed; the prospect of one's findings being read, used and cited is not enough, unless accompanied by carrots and sticks!

No wonder, therefore, that scientists are not rushing to self-archive; no wonder either that the "self-archiving" side has welcomed mandating "self-archive" so enthusiastically, even though it has nothing to do with the impact advantage argument. If research institutions, for example, through their promotion and tenure procedures, and the granting agencies, through their evaluation procedures, favor documents in Open Access in some ways, then Open Access will indeed progress. But one must understand that it is argument totally independent from the impact advantage argument.

Institutions and funders mandating OA for their research output has nothing to do with the 'impact advantage argument' (and evidence)? One might as well say that the existing weight that institutions and funders place on journal impact factors has nothing to do with impact either!

The very reason for OA itself, and the institutional and funder rationale for mandating that OA should be provided (by self-archiving) has everything to do with impact. It is in order to maximize the visibility, usage and impact of their research output -- instead of continuing to limit it to only those users whose institutions can afford to pay for access -- that universities and granting agencies are now planning to mandate self-archiving:

(Does J-CG imagine, instead, that they are mandating OA in order to reform the publishing system or to solve the journal pricing/affordability problem?)

The two [OA impact advantage and the OA self-archiving mandate] simply work in the same direction even though the presence of the latter argument (mandating) makes the limits of the former (impact advantage) more visible. It must be added that the mandating argument is a political argument, working therefore at exactly at the same level as the political arguments needed to convince various institutions to support Open Access. "Self-archiving," despite appearances, needs politics as much as the "Gold" road.

It is at this point that the reader must remember an important detail: For the "self-archiving" side, the goal is maximum impact and little more. Open Access is really nothing more than one instrument among several others capable of moving closer to this aim. Present toll-gated journals with subsidized reading on the part of the libraries also contribute to improving impact and this is the reason why some "self-archiving" advocates seem to live quite comfortably with the present publishing system, however unjust it may be for scientists who have not yet managed to establish themselves, for example, for economic reasons.45

It is unclear why a simple, straightforward point is being put so convolutedly: Publishing in a non-OA journal generates some usage and impact, but does not maximize it, because some potential users cannot afford paid access. They are denied access and their potential impact is therefore lost. Free online access (OA), through self-archiving, ends all access-denial and impact-loss by adding the usage and impact of those further OA users to the usage and impact of the non-OA users, to yield a maximized total impact. That's all there is to it!

And whatever does any of this have to do with injustice to "scientists who have not yet managed to establish themselves... for economic reasons"? We are not talking about OA journals here, and about how poor authors will find the money to pay OA-journal costs! We are talking about self-archiving -- doing the few extra keystrokes it takes to put the digital text of one's published journal article (whether published in an OA journal or a non-OA journal) online, free for all. .

Every author, rich or poor, established or unestablished, can do that!

Or is J-CG's point here about authors whose institution has no OA archive yet, or cannot afford one? There are now more and more OA archives that will let such authors self-archive in their archives, so this too is a non-problem. The real problem is the OA archives that exist, but are not being filled fast enough -- not the authors who have no archive to self-archive in!

Harnad explains in the following way:

Even if the growth of the open-access versions is destined eventually to reduce the demand for the toll-access versions, that is a long way off, because self-archiving proceeds gradually and anarchically, and journals cannot be cancelled while only random parts of their contents are openly accessible.46

In short, self-archiving, being anarchic in nature and incomplete in essence, works as a sort of impact bonus for those scientists willing to do it.

Self-archiving is anarchic in nature, but certainly not 'incomplete in essence'! Just incomplete in practice, today. Which is why systematic institutional and funder self-archiving mandates are needed: to remedy that incompleteness. That still leaves OA growing anarchically, article by article, rather than systematically, journal by journal. But that is fine, and a buffer against catastrophic cancellations.

And of course OA self-archiving promises an 'impact bonus for those scientists willing to do it' (and for their institutions and funders too, as well as for research productivity and progress itself). And even for those scientists who are presently unwilling to do it, but report that they will do it (willingly!) if/when it is mandated!

The anarchic nature of the process almost certainly guarantees its incompleteness.

I cannot follow this at all: Why/how does the fact that OA is growing anarchically, article by article, instead of systematically, journal by journal, guarantee that it will not reach 100%?

Worse, it becomes difficult to know where the incompleteness will appear.

It is not incompleteness that appears, but completeness! We know where the OA incompleteness  (80%) is today: just about everywhere. And we know that self-archiving will reduce it, and 100% self-archiving will reduce it to 0% incompleteness. One does not know today which articles are not OA (though it's a safe guess that at least 80% are not), but one does know that self-archiving can make that shrink, and that 100% mandated self-archiving will make it shrink to 0.

There is no way to cancel journals now, or in the near future, on that basis -- but OA is not about, or for, canceling journals! It is about maximizing access and impact.

As a result, for most researchers, Open Access repositories will probably not figure prominently in their literature search strategy.

On what evidence does J-CG base this rather gloomy prediction? Current author self-archiving rates? Current OA usage rates? The probability of self-archiving mandates? The probable efficacy of self-archiving mandates?

Moreover, in many disciplines, scientists cannot put too much faith in the capacity of Open Access to enhance their impact.

Which disciplines are those? I can find no difference in the uniform OA advantage across disciplines, just some variability in its size and in the extent and speed of OA growth:

Nor is there likely to be a discipline difference in the logical principle that access is a necessary (though not a sufficient) condition for impact...

Alas, the evidence that supports the impact advantage thesis is still too new and limited to be part of the scientists' common knowledge.

Then instead of denying the existence or scope of the growing evidence of the OA impact advantage across all disciplines, it might be more useful to spread the knowledge!

In the end, who can be interested by institutional repositories and "self-archiving" in its present form? Some are because they really and deeply believe in the sharing values that are at the foundation of scientific exchange, and that is wonderful. That argument is rarely, if ever mentioned by the supporters of the "self-archiving" strategy, but then altruism does not appear to be their forte.

I must say that I find this call to appeal to altruism rather puzzling: Of course authors self-archive in order to maximize access, not just to maximize impact (in fact, maximizing impact depends totally on maximizing access!). And they are encouraged to self-archive to maximize access, if that is what impels them to do it! But the fact is that far too few authors are self-archiving for either reason (access or impact) today. And an appeal to self-interest seems rather more promising than an appeal to altruism at this sluggish point -- even though they are two sides of the same access/impact coin, and whatever maximizes the one, maximizes the other too.

Nor will self-archiving mandates be made more palatable to authors if they are based on an appeal to their altruism, rather than on the benefits that OA will confer on them -- even though, again, whatever maximizes the one, also maximizes the other.

On the other hand, others may do so because of a strong obsession with their scientific status. They do not want any impact-loss and therefore "seek to eke," so to speak. In this case, the "eking" is aimed at the last ounce of prestige that can be extracted from their writing. From this perspective, Open Access "self-archiving," even though the results are "iffy" at best, cannot hurt indeed. However, my impression is that very good people do not really need this step [to maximize impact by self-archiving]; very mediocre people will not benefit from it anyway.

All research benefits from greater usage and impact. How the OA impact advantage distributes itself across the quality/seniority/impact hierarchy among articles and authors is an interesting empirical question that we will soon address. (J-CG seems to have some a-priori hypotheses that it will be interesting to text.)

Only a few average scientists might benefit a little from this strategy-hardly an earthshaking result. This is enough, in any case, to understand why self-archiving seems to generate so little enthusiasm at present.

Although J-CG does not appear very interested in gathering the pertinent empirical data himself, nor in examining the data we already have, he seems quite confident in making predictions and generalizations about their outcome

How Should We Build Open Access?

Does all this mean that Open Access will not work? Of course not! It does not even mean that "self-archiving" is fundamentally a bad idea. It only means that claiming that the only or, more modestly, the best road to Open Access is "self-archiving" is excessive, not to say wrong.

So far, all J-CG has actually said is that there is as yet very little OA self-archiving today: But that datum (15%)is contested by no one! I am concerned with finding ways to make that percentage grow (such as by designing self-archiving software, citation-linking, citation analyses and search engines, gathering and disseminating the evidence on the OA advantage, designing and promoting self-archiving policies, etc.), whereas J-CG seems to prefer to theorize about why Green doesn't grow, and won't.

It is unclear what "building Open Access collections" means: the outputs resulting from institutional OA self-archiving policies? Those are best thought of not as institutional OA collections but as institutional OA offerings!

Finally, it means that we had better think about ways to mix and match the "Green" and the "Gold" roads to Open Access if we want to ensure success and accelerate the growth of Open Access.

It is not clear how this exclusive focus on the slow growth of Green (and the speculations as to why it has not grown faster) gives a hint of what mixing and matching with Gold might have to do with it: J-CG has, after all, not focused on the far slower rate of growth of Gold

The one recurring theme that emerges from the previous discussion is value, and accessibility is but a tool to enhance it. When they play the part of an author, scientists obviously seek value (i.e., impact). Value is perhaps the single most important element for the Open Access movement but in dealing with value the Open Access movement must not forget that scientists are also readers. At that precise moment, the "attention economy" kicks in. Scientists look for value there too, but it is search and retrieval value that is of the essence in the scientist-as-reader context. Value [impact], in short, is a little more complex than what the supporters of pure "self-archiving" imagine.

In the case of the "Gold" road, thinking about value can be quite simple since it amounts to transposing the familiar practices and strategies of the traditional publishing sphere. This is exactly what PloS and BioMed Central journals already strive to achieve, and they are already reaching interesting results. Likewise, the more national or regional approaches of Brazil and other Latin American countries (plus Spain)47 are also bearing fruit. We may expect more of this latter kind of Open Access journals in the near future, particularly in countries like India48 and China.49 Similar trends may appear in richer countries with a centralist political tradition. France,50 Italy,51 and Spain52 are prime candidates in this regard. In the case of France, various national research centers are already studying the issue, for example, CNRS (The French National Center for Scientific Research), INSERM, and INRA (The French National Institute for Agricultural Research).

The "Gold" road is not always an easy road to follow. Stevan Harnad is right to underscore this point. But as the previous pages demonstrate, the "self-archiving" side is not easy to follow either.

The issue is not so much ease but speed, and probability. Both Gold and Green have been slow, but Gold has been three times as slow. (J-CG completely overlooks this fact, stressing endlessly the small number of articles that are made OA by self-archiving, ignoring that a three times smaller number is made OA by OA publishing!) And the probabilities are stacked much more heavily against Gold than Green: Only 5% of journals are Gold; the other 95% are highly resistant to converting to Gold, and with good reason (financial risk). On the other hand, 93% of journals are Green. That means the only obstacle to 93% OA via Green is the slowness of authors to self-archive -- for which a self-archiving mandate is the obvious remedy. There is no corresponding remedy for the slowness of publishers to convert to Gold.

Where governments decide to move in and press for Open Access publications, a great deal of time-consuming political groundwork must be done and requires countless interventions from people who need support. But, as we have seen, the need to rely on mandating shows that the "self-archiving" side cannot avoid political maneuvers either.

Governments cannot and do not "move in and press for Open Access publications": Whom can they press, how, to do what? All that governments can do is to cover OA journal authors' publication fees. What they can press for is Open Access itself. Research funders can require their grantees to provide OA to their funded researchfindings -- by either self-archiving all the resulting non-OA journals articles (Gold) (Green) or (if a suitable journal exists) publishingthem instead in an OA journal (Gold) -- as a precondition for receiving research funding at all. And research funders now seem well on the way to pressingfor exactly that. If/when they do, we will all be well on the way to 100% OA.


In the cases of associations or society journals, the issue of a business plan quickly surfaces and that is what the "subsidized author" model allows to explore. Various other schemes have been suggested to help the transition from a toll-gated to an open access journal, for example, offering open choice to authors in order to demonstrate that within a particular journal (the "all things being equal" issue again) Open Access articles enjoy a better impact on the average than their toll-gated counterparts.53

Recently, we have seen at least one commercial publisher (Springer) offering open choice to its authors.54 The move is certainly bold, especially with the steep up-front payment required, but it is also a clever move. Potentially, the following results can be demonstrated or achieved through such a move:

1. It may reveal a new business plan where money can be siphoned-off from granting agencies on top of what libraries already provide. In a world of plateauing library budgets, the perspective of extracting some money from the granting agencies to increase the revenue stream may look quite intriguing to a business leader who responds to his stock holders only.

2. [Springer's 'Open Choice': author can choose to pay for OA on a per-article basis] may also help set a value scale between impact factors and article costs.

There is already a way to translate the OA impact advantage into dollars. For example, the UK Research Assessment Exercise (RAE) rankings -- in which rank and funding are highly correlated with citation counts -- can be re-calculated to show how much a rank would be raised, and how much more income would result, if a department or a university mandated OA self-archiving:

3. Finally, it may also try to demonstrate that the whole "author-pays" (so-called) business plan simply does not and cannot work.55 Running an experiment within a large profitable company does lead to a great deal of leeway when time comes publicly to report financial results and interpret them.

Ultimately, the point of the "Gold" road is to create intellectual value in new and better ways.

'Create intellectual value'? Is it not simpler and more theory-neutral to say that OA journals make their articles OA, and charge the author-institution, whereas non-OA journals charge the user-institution? Do we really need the theorizing about 'creating intellectual value'? After all, we all know what journals do: they implement peer review and tag the result with the journal's name, and its associated track-record for quality. OA and non-OA journals both do that; they just charge different parties for it.

To achieve this goal, the "Gold" road must pay attention to more than the impact advantage that addresses only the author side of the scientist. A scientist is also a hurried reader and value can be built out of better searching, retrieving, and navigating tools. As a result, all the "Gold" projects should strive to collaborate to create citation links and indices.

Gold journals should certainly collaborate and citation-link, and they do. But so do all the non-Gold journals (Green and Gray) do the same. And so do OA Archives and harvesters, like . So what is the point here?

Meanwhile, the role of the "Green" road must be carefully and precisely defined. The "pale green" case should be treated apart from the two other shades of Green, and this specific status should be clearly indicated in the metadata, particularly in OAI-PMH.

I would suggest quite the opposite. 'Pale-Green' (endorsing preprint self-archiving) is a journal policy (of 13% of journals, currently; 80% are Full-Green, endorsing postprint self-archiving). Endorsing preprint self-archiving is not a property of an OA article. What an OA article needs to indicate clearly is whether it is an unrefereed preprint or a peer-reviewed postprint (along with the journal name in the latter case).

In parallel, a very deep shade of Green should be set aside for publishers that give an irrevocable right to "self-archive" to their authors, or alternatively they could leave the copyright in the authors' hands. The reason is that we simply do not want to see publishers suddenly rescinding the permission to "self-archive" and thus bringing down the whole OA edifice as if it were a house of cards. Stevan Harnad generally chooses to ignore this issue or treats it as useless speculation, but the danger is much more concrete and real than he is willing to admit or concede.

It is indeed completely useless speculation to worry (Worry #32) about Green reverting to Gray at a time when 93% of journals are Green (80% Full-Green) whereas only 20% of articles are OA! It is an utter waste of time to be contemplating the need for further shades of Green, rather than self-archiving while the going's Green:

More fundamentally, we must find a way to move from institutional to disciplinary and even specialty repositories. This is important because it is easier to create the former than the latter and they are presently multiplying.

It is not only unimportant to move from local institutional to central archives of any kind, but unnecessary and counterproductive. In the distributed online age, contents are sorted by metadata tagging, not by physical locus or focus. And the author's institution is the natural locus for mandating institutional self-archiving, monitoring compliance, and measuring and rewarding impact (for all of its disciplines).

However, as we have seen, the effects of such repositories are problematic at best and a failure now will set the Open Access movement back for many years. It becomes important therefore to move beyond simple, isolated, institutional repositories. This means aggregating and repackaging the information that is contained in these institutional repositories along subject lines. In practice, this also means inter-institutional collaboration and coordination.

(All these confident plans about which way institutional OA self-archiving archives need to go, having argued up to this point mostly against their utility a-priori!)

What OA archives need is only one thing: OA articles. And what is needed to induce authors to self-archive their articles is a self-archiving mandate. This has nothing whatsoever to do with aggregating and repackaging or with inter-institutional collaboration and coordination. Each institution need merely ensure that all of its own annual journal article output is self-archived in its own OAI-compliant OA archive. The rest will take care of itself, quite naturally, via OAI-interoperability (and the reciprocity inherent in institutions self-archiving their own research output).

In an institutional repository, the metadata should be organized in a sufficiently clear and standardized fashion to allow a quick disciplinary representation of what is available there. This would allow the efficient concatenation of disciplinary articles across a number of depositories. The point, indeed, is that harvesting across all repositories in one simple, single sweep is not enough. While this task must be maintained and even enhanced, disciplinary harvesting must be available and be as user-friendly and efficient as possible-a daunting problem in itself. In parallel, competitive forms of subject packages should also be allowed to emerge. This would lead to new value hierarchies and new ways to create value. As a consequence, the value creation capacity of toll-gated journals would tend to be somewhat diluted.

Is J-CG proposing to re-invent the OAI protocol? He is breaking down open doors! And all of this has to do with interoperability, harvesting, and search. Nothing to do with diluting 'the value creation capacity of toll-gated journals'! That is all just theory-spinning. One day 100% Green OA might engender a transition to Gold -- or it might not. Today, that is utterly irrelevant. What is relevant and pressing today is the need to reach 100% OA as soon as possible. That is already well within reach and long overdue. What we need is not "new ways to create value", but new ways to accessthe value we have created already -- for all those potential users who do not now have access to that value.

These goals raise a new question relating to ways and means: Who should take charge of this new form of presentation?

Who should take charge of what new form of presentation? And what does 'take charge' mean? The OAI-PMH was created so OAI service-providers (like OAIster and Citebase) can 'take charge' of services like harvesting, citation-linking and search across the distributed, interoperable OAI-compliant OA Archives.


Institutions make their own journal article output Open Access by self-archiving it in their institutional OA Archives for all potential users who cannot afford access to the non-OA version: What's to be taken charge of? And why is this a 'new form of presentation' rather than a complementary means of access?

Actually, the answer is not very difficult to sketch, but we all know the devil will be in the details. Various consortial forms already exist among sets of libraries: licensing consortia, interlibrary loan consortia, new kinds of consortia based on institutional repositories, such as the DSpace network.56 Consortia of any type among institutions that view each other as peers can become the bearers of these new kinds of disciplinary projects. In particular, the present deployment of DSpace might be a good place to start exploring and implementing such a strategy. In parallel, licensing consortia might consider extending their objectives to providing support for the creation of strong sets of disciplinary repositories across their members. Prestige hierarchies, based on the reputation of the institutions involved, will emerge from such efforts.

In this complicated congeries of consortia, could I ask a simple question? What is this conglomeration all about? We are busy conceptually repackaging content that we do not have yet, and content [i.e., articles published in journals with names] that will not need any repackaging. Why are we not focusing on the practical task of actually providing the (absent) content, now, instead of theorizing about hypothetical future consortia to re-label and redistribute it?

I would suggest starting not with peer-reviewed articles, but rather with doctoral dissertations [emphasis added].

Here we have arrived at the heart of J-CG's proposal. OA's target is peer-reviewed articles. Only 20% of them are OA to date. And J-CG suggests that we not 'start' with them, but with self-archiving dissertations (some of which are being self-archived today already).

Why are we being asked to start our mixing match with dissertations?

(The answer will shortly become apparent: We will never in fact be moving on to self-archiving peer-reviewed journal articles, in J-CG's recommended mix/match scenario: We will instead somehow -- it is never specified quite how (but one guesses that it is by going on to self-archive unrefereed preprints rather than peer-reviewed journal articles) -- segue into the construction, bottom-up, of a brand new alternative peer review and publication system, somehow, on top of these dissertations, to replace the existing peer-reviewed journals. This is in fact J-CG's version of the Golden road, of creating OA journals; it is not a mix-and-match of Green and Gold at all! It is an independent reinvention of the entire wheel. The first step is new forms of "quality evaluation" -- arising, somehow, out of new ways of 'promoting the intellectual value' of doctoral dissertations)

These documents are totally controlled by university professors and students, except in the case of patentable results stemming from doctoral research supported by private money. But these cases form a minority at best and can be left for special treatment. Meanwhile, an interinstitutional strategy to promote the intellectual value, authority, and prestige of doctoral theses could easily provide the testing ground for the emergence of interinstitutional disciplinary archives.

I am at a loss: At a time when out of 2.5 million annual peer-reviewed articles in 24,000 journals -- all of them already having whatever intellectual value they already have  -- only 20% are OA, we are being bidden to self-archive dissertations and to promote their intellectual value! But isn't it the other 80% of those already-evaluated but still inaccessible articles that we really need? And is this, then, the promised alterative Green-and-Gold mix-and-match proposal?

Evaluation Levels

The metadata should also be extended to provide some indication of quality.

Quality metadata? For dissertations? Whatever for? We are not trying to create a dissertation peer-review service. We are trying to provide OA for the other 80% of already peer-reviewed articles; and their 'indication of quality' is the name of the journal that already performed the peer review on them!

It could be designed to help identify the identity and the nature of the evaluating body that passes judgment over the documents in the repository.

For journal articles, the 'evaluating body' is the journal, tagged by its name. For dissertations, who knows? (And who cares, insofar as OA is concerned? Quality-tagging dissertations has nothing to do with OA.)

In other words, the metadata should help identify the quality, nature, and procedures of groups that begin to work as editorial boards would. The metadata could also help design evaluations scales - imagine a one brain, two brain, ... n-brain scale, similar to a Michelin guide for restaurants.

We don't need a Michelin guide to journal articles! We just need access to the articles, and their journal names! Here we are instead busy reinventing peer review, in hitherto untested and (as we shall see) rather extravagant new forms:

Users should have a clear idea of who the reviewers are and how much they can be relied upon.

Is J-CG suggesting that the journal name (contrary to what has been the case until today, irrespective of whether the journal is online or on-paper, OA or non-OA) is not enough? That articles should be tagged with the names of their (often anonymous) peer-reviewers too? (That sounds rather radical, for an untested proposal.)

Or is this just about tagging dissertations with their university and committee names? (Fine, but how does that generalize to journal articles? And what has it to do with gaining Open Access to those articles?)

This leads to a new project: If various universities create consortia of disciplinary repositories, then nothing prevents them from designing procedures to create various levels of peer review evaluation, e.g., institutional, consortial, regional, national, international.

This is a new project indeed! For the old project was to provide Open Access to the remaining 80% of the  existing 2.5 million annual articles in the existing 24,000 peer-reviewed journals. Now we are talking about what? Reviewing the peer review? Re-doing the peer review? Replacing the peer review?

I suspect that -- driven by his publication-reform theory -- what J-CG is really contemplating here is entirely replacing the non-OA journals in which 95% of these self-archived articles are currently being published! But OA itself is just about providing free, online access to those peer-reviewed journal articles, not about wresting them from the journals that published them, and from their peer-review -- at least not OA via the Green road of self-archiving.

Nor is that what a journal means in giving its Green light to self-archiving: that you may treat my published articles as if they had never been peer-reviewed and published, and simply start the whole process anew! Nor is there any point in starting the whole process anew, if it was done properly the first time. It is hard enough to get qualified referees to review papers once, let alone to redo it (many times? at many "levels"?) all over again.

Nor is that what the authors or the would-be users of those already peer-reviewed, published articles need or want. What authors and users want is Open Access to those articles.

How did we get into this bizarre situation? It was by accepting the invitation to populate the OA Archives with dissertations instead of peer-reviewed journal articles, as a 'testing ground.' Apparently, we are never to graduate to self-archiving journal articles (Green) at all: Rather, we are to rebuild the whole publication system from the bottom up, starting with dissertations, and then generalizing and applying it to the unrefereed preprints of articles-to-be. (J-CG has wrongly inferred or imputed that "Green" means mainly the self-archiving of unrefereed preprints, rather than the peer-reviewed, published postprints that are OA's main target!)

In other words, we are to reform the publication system after all, just as J-CG recommended. Never mind about providing Open Access to that other 80% of existing journal articles: We will instead create new evaluation bodies (Gold). We have time, and surely all parties will eventually go along with this project (in particular, those sluggish authors who had not been willing even to self-archive). They are to be weaned from their current journals and redirected instead to -- it is not yet altogether clear what, but apparently something along the lines of: "various levels of peer review evaluation, e.g., institutional, consortial, regional, national, international"

At that point, a recognized hierarchy of evaluation levels can begin to emerge [emphasis added]; as such, it should also be clearly identifiable through the metadata. Not only could the user know what level of peer review and evaluation is being used, but also which group is backing it. In effect, this is what a journal does and this is how it acquires some branding ability.

In effect, the current journal system is what J-CG is proposing (because 95% of journals have obstinately declined to go Gold) to replace (bottom-up, starting with dissertations) with the above "emergent" Gold system, consisting of "a recognized hierarchy of evaluation levels" ("clearly identifiable through the metadata"). And this alternate "branding" system, starting with dissertations, is to emerge as a result of mixing-and-matching Green-and-Gold.

What follows is speculation piled upon speculation, all grounded in this initial premise (that if you can't convert them, replace them):

An international registry of such evaluation procedures and of the teams of scholars involved could then be developed, perhaps in parallel to the Directory of Open Access Journals (DOAJ) at Lund University. This obviously would lend transparency and credibility to these value-building procedures. In this fashion, a relatively orderly framework for expanded peer review and evaluation can emerge.

J-CG has here managed to resurrect here, almost exactly, thevery same incoherence that beset Harold Varmus's original 1999 E-biomedproposal, which could never quite decide whether what was being proposed was: (1) free access to journal articles, (2) an archive in which to self-archive articles to make them freely accessible, (3) a rival publisher or publishers to lure away authors from existing journals (4) an alternative kind of journal or journals, with alternative kinds of peer review, or (5) all of the foregoing:  

That somewhat mixed-up and ill-matched 1999 vision is what has since become gradually more sorted out and focused in the ensuing years, as follows: First, PubMed Central   was created (February, 2000) as a central OA archive in which publishers were invited to deposit their contents six months after publication. When few publishers took up that invitation, the Public Library of Science (PloS) was founded (October 2000) and circulated an Open Letter, signed by 34,000 biomedical researchers the world over, demanding that existing journals should go Gold  .

When that too failed, PLoS became an Open Access Publisher (2001) and has since launched two Gold journals

(forgetting altogether about Green self-archiving).  Most recently (2004), perhaps having noticed that the Golden road to OA is a rather long and slow one, PLoS again took up the Green road of self-archiving by helping to promote the NIH public-access policy (which requests that all articles resulting from NIH-funded research should be self-archived in PubMed Central within 6 months of publication):

J-CG now proposes to resurrect something very much like the original 1999 matchless E-biomed mix once again! I would like to make the counter-proposal that once was enough: that we should forget about trying to rebuild the publishing system bottom up (with a vague, untested, speculative, and probably incoherent mix-and-match model of archiving and publishing) and instead reinforce the road that Harold Varmus, PLoS, the Wellcome Trust, the UK Government Select Committee, and many others have lately rejoined (and the one they would have been better off taking in the first place) -- the Green road to OA -- by promoting the OA self-archiving of the remaining 80% of the existing journal literature as a condition of research funding (and employment).

In parallel, these consortia should also work toward enhancing the accessibility tools that will tend to make these Open Access resources as valuable and easy to use as the best commercial products. In particular, this could be the expected (but not necessarily exclusive) province of the librarians.

These remarks lead to intriguing possibilities. For example, any paper could be evaluated more than once, and in any case peer review is certainly no longer limited to the prepublication stage.

Now J-CG is resurrecting JWT Smith's (likewise untested and unrealistic) 1999 'Deconstructed Journal' proposal, and its easy profligacy with a rare and already overworked resource -- the pool of qualified referees -- contemplating not just one but many peer-reviews for the same article, at multiple stages and levels.

Yet recall (faintly, if you still can) how and why this all started out: The need to provide Open Access to the 2.5 million annual articles published in the 24,000 peer-reviewed journals, because of the many potential users who could not afford the toll-access to them. Here we are now, offering them, in place of actual OA to the actual 2.5 million annual articles, hypothetical OA to hypothetical future articles, in hypothetical future journals, refereed -- not just hypothetically but miraculously -- by multiple hypothetical systems of peer review, implemented by hypothetical 'Deconstructed Journals.'

Today's would-be user, denied access to those of the 2.5 million actual articles he wants and needs, today, can be forgiven for not feeling much better off with what he is being offered here instead

All this demonstrates that new forms of evaluation can (and probably will) develop. For example, while the number of formal citations obviously defines the impact of an article, the number of informal citations-e.g., within the Web, the number of links that refer to a particular paper-can also provide further evaluation information.57

There are further advantages accruing from this approach. By involving researchers in the design of these new modes of evaluation, the debate about Open Access begins to take on a tangible, credible, even vibrant form. While thinking about evaluation, scientists should also begin to understand how these new tools can improve both the process of scientific research and the management of a scientific career. By moving in the direction of Open Access, granting agencies can do their part and help clarify the evaluation levels and processes. In this fashion, they would constructively participate in the general reworking of value creation that is so lacking in pure "self-archiving" at the present time.

Consider how much more OA granting agencies would get in exchange for their pains if instead of "clarifying evaluation levels and processes to constructively participate in the general reworking of value creation" they simply mandated that "pure self archiving that is so lacking at the present time": That would ensure Open Access to the portion that they fund of the 2.5 million annual articles in the world's 24,000 peer reviewed journals -- an extant value that wants only access-creation, not value re-creation.

Finally, scientific associations and societies that have decided to go further than Pale Green can take advantage of this situation to create specialty juries and develop new metrics of science that will ensure pushing some articles up the value chain that would have simply been ignored otherwise. Many articles are regularly undervalued not because of their content but because they appeared in journals with a modest status. Others are overvalued for symmetrical reasons. This part of the value creation project can take the form of the "Faculty of Thousand" invented by BioMed Central, or it can take the form of prizes or any other form susceptible of attracting more attention to these articles. In short, "marketing" could be totally redesigned along lines that have more to with the quality of content than with the ability to bundle huge amounts of articles and titles. Let us remember that the latter method-the "Big Deal"-amounts to promoting mediocrity rather than excellence: It does so by making inferior products far more accessible and thereby artificially stimulates usage through a clever exploitation of the attention economy principles. And it justifies all this by using the fallacious pretext that the cost per title is decreased!58

New Journal Models

Transparency, prestige, and rigor are needed to create credible value. In effect, something like "overlay journals"59 begin to emerge, and as such they can gradually acquire visibility and respect. At that point, the institutional repositories will have effectively morphed and matured into a consortium-based network of repositories with a rich set of value-creation tools and increasingly recognized names or labels.

The trouble is that all the "morphing" so far is happening only in the mind of the passive speculator; and meanwhile 80% of articles continue to be inaccessible to those would-be users who cannot afford access -- yet that was supposed to be the problem OA was remedying.

Just like E-biomed and "deconstructed journals," "overlay journals" are at the moment figments of the armchair theorist's imagination. Moreover, it is not even clear what 'overlay journals' means. If it just means conventional journals (whether hybrid or online-only) implementing online peer review by having submissions deposited on a website and then directing referees and revised drafts to the site, then most journals are already overlay journals in this banal sense.

If 'overlay journals' means journals that are online only, then that is nothing new or interesting either. If it means that the archive to which the referees go to find the paper and where revised drafts are put is not the journal's website but an OA Archive (whether institutional or central), then that too is uninteresting -- just a trivial (and quite natural) implementational variant of a standard feature of extant journals and conventional (online) peer review:  

If the journal itself performs only peer review and certification -- and the archives do all access-provision and archiving, then this may have some potential interest, some day -- but there exist at most a handful of journals that resemble that description today, and between them and the remaining 99.99% of journals is the still unsettled future of OA (Gold) journals and their cost-recovery model.

So overlay journals are still just an armchair speculation: But 100% OA need not be -- and it certainly need not wait for the 'morphing' of the current 24,000 journals into overlay journals.

As a result of this evolution, overlay journals can hope to become part of the search strategies of the scientists. Eventually, original submissions will be addressed to these new channels of scientific communication. They will own a reputation, a profile, editorial orientations, and this despite, or in parallel with, the fact that most of these articles (or even all of them), at first, will have been already "published" in traditional journals. This is where the importance of "self-archiving" really finds its anchoring point.

It is hard to see how an article that has already been published in a traditional journal can become an "original submission" to an "overlay journal" -- harder still to see this as self-archiving's real "anchoring point." Perhaps self-archiving should just stick to the more mundane task of providing immediate OA to the remaining 80% of the current journal literature, rather than waiting for this hypothetical new multilevel, multivalent system to evolve?

All kinds of consortia should begin to engage in this kind of activity, sometimes building on the experience gained from doctoral theses online, sometimes not, to enhance their offerings to their constituencies. When "republishing overlay journals" turn into original publishing channels, they will be equivalent to "Gold" journals. The difference is that they will result from a succession of small, incremental, transformations rather than from a dramatic one-step creation or conversion of scientific journals. All this can be accomplished with a good alliance between librarians and scientists. All this can be achieved by treating the "self-archiving" strategy as a transition phase on the way to the "Gold" objective.

The only "transition phase" that is worth talking about (and is tested, and visible, and reachable) is the transition from today's 20% OA to 100% OA via self-archiving. After that, nolo contendere -- and hypotheses non fingo!

Such a strategy does not exclude "self-archiving"; neither does it compete with the direct creation or transition to Open Access journals. It simply welds these various parts of Open Access into a coherent vision of what Open Access ought to look like.

The "Green" and "Gold" roads toward Open Access will thus merge while helping each other. The haphazard, anarchic process of "self-archiving" will be made more orderly by the disciplinary classifications and the evaluation hierarchies. Researchers will know that the collections are incomplete, but they will also know that they are rich and therefore quite valuable and useful. Moreover, various filters corresponding to requirements about quality, beside those organized around topics (keywords, etc.), will greatly enhance the accessibility of these repositories. Discovering this will incite ever more people to "self-archive." The fear of information overload will vanish. Granting agencies will not have to fear the resistance of scientific communities.

Finally, because an even playing field will be established between toll-gated publications and open access articles, be they "Gold" or "Green," the impact advantage of genuine Open Access will have a much better chance of asserting itself unambiguously.

I still have no idea what this means! What is meant by an 'even playing field' between OA and non-OA articles (as opposed to articles in OA journals versus non-OA journals) when articles in the same non-OA journal have both non-OA and OA (self-archived) versions? How can an article, or an OA advantage, compete with itself? And why?

Probably at this stage, the tipping point toward Open Access will truly be in the offing.


The vision presented here is nondogmatic. It leaves plenty of room for revisions, critiques, and reevaluations. It tries to present a constructive evolutionary scenario where the "Green" and "Gold" roads can find their proper place without feeling in competition with one another. It also rests on the two following premises that some advocates of the "Green" road do not seem ready to accept:

1. The finality of the scientific exchange is not just for scientists-as-authors; it must also take into consideration the scientist-as-reader, and it is in this context that the issue of Open Access must be complemented by that of "accessibility."

2. Even if we accept reducing science to maximizing impact-a dubious, simplistic claim, at best-scientists, now limited to being scientists-as-authors, appear incapable of implementing a complete form of Open Access simply through "self-archiving," be it mandated or spontaneous. In fact, the need to rely on institutional policies and parliamentary committees demonstrates the incomplete nature of the "self-archiving" strategy taken in isolation.

The assertion is clear. What is not at all clear is (1) how and why is asserted to be incapable of providing complete, 100% OA and -- especially (2) once it is mandated. Nor is it clear (3) why relying on a mandate means self-archiving is asserted to be incomplete (rather than just too slow!).

If we now refer back to the British Report from the Commons Select Committee, we can now suggest a more interesting reading of this document. Far from putting all of its eggs in the first "self-archiving" strategy and mentioning other actions only as inferior and secondary (the "mandating" versus the "recommending"), perhaps the Select Committee meant to lay out a phased strategy: Right now, they seem to say, we can begin by doing all the "self-archiving" we can.In parallel, other strategies should be studied and implemented later to complete the "self-archiving" strategy and make it viable.

That is precisely what the UK Select Committee did recommend: immediately mandating OA self-archiving (Green), and in parallel, continuing to support and study OA publishing (Gold). And that was indeed the right decision, and the right priorities and weight.

What might be interpreted by some "self-archiving" supporters as the results of a blurred vision may be better interpreted as symptoms of deep wisdom indeed. It is not that the committee members "got it" on one point and wasted time on some others; rather, they appear to have "gotten it" from beginning to end, and this paper has done little more than try and argue why this is the case. In the end, Open Access will not be reached by a narrow focus on impact optimization alone-although this argument is a very useful one-it will be reached only if the full complexity of the scientific communication process is taken into consideration.

Open Access can be helped by tough-minded forms of arguments. Taking the relentless quest for impact optimization as a basis for modeling scientific behavior does yield interesting working hypotheses; however, it is false to assert that this is enough to encompass and fully apprehend the rich behavior patterns of those engaged in knowledge creation and validation. Fundamentally, science deeply agrees with Open Access not only because it is the best way to achieve the greatest impact for a particular individual, but also because it provides the most favorable environment to foster the widest form of distributed intelligence on this planet. And deploying distributed intelligence should not be fostered for the sake of intelligence alone: This activity has meaning and use and all of humanity is concerned by it.

Presently, scientific communication is limited roughly to one fourth or one fifth of the individuals that, through their native abilities, can contribute to scientific progress. Even in rich countries, many good brains are incapable of reaching their full potential and therefore are wasted because of the ways in which access to scientific literature is limited. Think of a promising young professor who must take a first job in a small college with limited library facilities (and probably limited lab facilities as well) simply because he happened to graduate in a period of PhD glut. While Open Access does hold the promise of enhancing the career of many established scientists, more fundamentally, it promises to create a much more open field for a widened circle of researchers. It can also reach into communities of concerned users. Think of patients seeking useful information to assist in their own treatment.

Open Access should not be the tactical tool of a few, elite, established, scientists that want to enhance their careers and little else;

No one has suggested OA is, or should be the tactical tool of a few, elite, established, scientists. It is J-CG, however, who suggested (without saying how, or why) that impact enhancement through OA self-archiving would only benefit the elite, established scientists. The analysis by author/article seniority and quality-level is yet to be done, but there is no particular reason to expect that the OA-impact advantage will be only, or even mostly, at the top.

neither should it be approached only from a kind of Hobbesian attitude where the worst scenario is used to demonstrate that even seen in this dire way, things turn out right in the end. OA mainly aims at improving the knowledge creation system of science and better insert its results within our societies.

Open Access does not need to draw an absolute knowledge divide between scientists and the rest of the population, between elites and the "masses";

Who is drawing such a divide? Most of the annual 2.5 million articles published in the world's peer-reviewed journals are specialized scholarly/scientific articles that are only of interest to fellow-specialists; but OA articles are accessible to any interested user. So what is the problem?

while it does not eschew vigorous competition, Open Access insists that the playing field should remain reasonably even and fair.

Vigorous competition between what and what? (J-CG again seems to be thinking only in terms of OA journals vs. non-OA journals here.)

In the end, Open Access is the sine qua non condition for the optimal deployment of scientific research worldwide, as well as for its widest applicability in the general population.

And if, finally, some people should object to the last argument as being irrelevant to scientific research, they should also remember that the public pays for much of it.60


Several people have had a very direct and most precious input into this paper and I would like to thank them for having taken the time to read this little study. I want also to thank them for having occasionally saved me from my own foolishness. In this group, I would like to include Fred Friend, the well named, David Prosser, Colin Steele, and Ray Siemens. I would also like to thank the editors and referees for their useful and important comments. Last, but not least, I want to mention my very dear Frances, better known to most librarians as Frances K. Groen; please accept this modest expression of my fullest gratitude.


1 The Open Access movement has been characterized by a common objective-namely Open Access to peer-reviewed, scholarly articles-and a dual strategy to attain this objective. See the Budapest Open Access Initiative (BOAI) published on the Web on February 14, 2002, To  qualify as Open Access, a document must follow two different sets of conditions that were clearly outlined in the Bethesda declaration, peters/fos/bethesda.htm#note1. (1) The user is granted a number of rights (e.g., "a free, irrevocable, worldwide, perpetual right of access to, and a license to copy, use, distribute, transmit, and display the work publicly and to make and distribute derivative works"); (2) the document must be archived "in at least one online repository that is supported by an academic institution, scholarly society, government agency, or other well-established organization that seeks to enable open access"; these are the exact words of the Bethesda Statement on Open Access. They refine and elaborate upon the definition that emerged with BOAI. The Public Library of Science endorses the Bethesda definition of Open Access (see

The term 'Open Access' was coined and defined by the BOAI, and, as noted, two roads to OA were specified : the Green road (BOAI-1, OA self-archiving of articles published in non-OA journals) and the Gold road(BOAI-2, publishing articles in OA journals). The Bethesda statement was the first of several subsequent statements and declarations that re-focussed both the definition of OA and the strategy for achieving it on Gold alone. Note how in the Bethesda statement, self-archiving (BOAI-1) has become merely 'archiving' (which is of course a necessary condition for both Gold and Green, for even Gold must provide free online access somehow!). That, together with the (needless) stipulation that OA requires rights re-negotiation (which it does not, for what is already at least 93% of journals!), effectively meant that the only articles that would be considered OA would be those that were either published in OA journals or in journals that were prepared to become formal OA journals-on-demand on an individual-article basis. Self-archived articles -- even those from most of the 93% of journals that are Green -- would not count as OA according to this new Bethesda criterion, because they had not renegotiated rights with the publisher to make them exactly equivalent to articles published in an OA journal! This unnecessary, arbitrary and counterproductive criterion was subsequently propagated to the Berlin Declaration and most further formal statements on OA for several years, with the result that OA was taken to be equivalent to OA publishing (BOAI-2) and self-archiving was relegated to its archiving dimension, rather than the independent -- and far more powerful and direct -- Green road to OA provision that it is, and continues to be, according to the original BOAI definition!

2 This "reader pays" phraseology is as inaccurate as the "author pays" expression. Later in this text, we shall speak about a "subsidized author.".

This is correct. The descriptor should be 'author-institution payment' vs. 'user-institution payment' (the former being the OA publishing cost-recovery model and the latter being the non-OA publishing cost-recovery model: subscriptions, site-licenses, pay-per-view). Just as non-OA online access is no longer based on individual payment, but institutional payment, neither OA nor non-OA authors are meant to pay anything out of their individual pockets.

3 This is, at best, shorthand for journals deriving their income at the point of production and not at the point of sale. Effectively, the point of sale disappears with Open Access. Someone, perhaps a granting agency, a foundation, a research institution, or even in some rare cases, an author, pays the publishing fee set up by the publisher. A better expression would be "paid on behalf of the author," which is accurate but a little unwieldy. Perhaps a "subsidized author" would foot the bill and provide a nice parallel for the "subsidized reader" expression used later on.

4 In India, Brazil, Chile, Cuba, etc. See notes 47-52.

5 Harnad/Temp/self-archiving.ppt, slide 47. Specifically, Harnad writes: "Open access through author/institution self-archiving is a parallel self-help measure for researchers, to prevent further impact-loss now. Open access is a supplement to toll-access, but not necessarily a substitute for it." Note the reference to "impact-loss." This is really a "manque-ˆ-gagner" (loss of possible gains) rather than a direct loss. What Harnad means to say is not that impact already gained is going to be lost; it is that impact that might be added to already gained impact is not being added. What he really meant to write is that self-archiving is a self-help measure to open up the possibility of further impact gains.

Correct. Self-archiving is author-institution self-help to maximize impact (or minimize the loss of potential impact) by providing online access to all would-be users, web-wide, and not just those whose institutions can afford paid access.

6 See The SHERPA version of RoMEO, which is to be preferred as it is current, can be found at SHERPA is funded by JISC and CURL. It is hosted by the University of Nottingham. The "Green" and "Gold" terminology itself seems to have been invented by Stevan Harnad while discussing the results stemming from the RoMEO study.

The information on individual journal self-archiving policy (rather than just publisher self-archiving policy) is available at -- a site which also dispenses with the excessive, unnecessary and uninformative color codes at the SHERPA site (yellow, blue, red, white) and provides the relevant information: Green light to self-archive peer-reviewed postprint, FULL-GREEN [70%]; pre-refereeing preprint, PALE-GREEN [13%], neither yet, GRAY [7%].


8 A summary of the House Committee recommendations (July 15, 2004) can be found at the following URL: For the publishers' reactions, see their open letter to Dr. Elias Zerhouni, dated August 28, 2004, available at It is important to read the full letter rather than the excerpts published by Ann Okerson on Liblicense-l on August 30th ( llicense/ListArchives/0408/msg00137.html).

9 Harnad/Hypermail?Amsci/3875.html.

10 The tradition of exchanging offprints among scholars and researchers is a clear example of a situation where affordability and access are sharply kept distinct.

More accurate description: Researchers generally provide access to their own work merely in order to provide access (and maximize impact), with no thought one way or another regarding journal pricing or affordability! If forced to think about it explicitly, researchers would no doubt agree that (1) if their work were already available online to all of its potential users (2) then they would not need to bother self-archiving and that (3) this would be the case if every journal in which they published were affordable to and licensed by every institution of every potential user -- but (4) unfortunately it is not. The reasoning from 1-4 is rather trivially obvious, however, and need not really be explicitly formulated in order to arrive at this much simpler decision: I will self-archive in order to maximize the access to and the impact of my work!

11 This is an allusive reference to a very recent discussion (August 6, 2004),

12 For an interesting discussion on the number of refereed journals and articles, see Harnad/Hypermail/Amsci/2983.html. The figures quoted in this discussion range from 15,000 titles (Eugene Garfield) to 24,000 titles (Stevan Harnad) with a corresponding spread in the number of articles published annually: from 1.5 to 2.5 million-the ratio of 100 articles/journal/year is commonly used in the scientific, technical, and medical disciplines (STM). The figure of 85% dates back to the early part of August 2004. On August 25th, Stevan Harnad advanced the 93% figure along with the conversion of the Royal Society of Chemistry ( Harnad/Hypermail/Amsci/3938.html).

This is all correct. It should only be added that it makes no difference whether the true number of peer-reviewed journals is closer to 12,000 (Garfield), 24,000 (Ulrichs) or 48,000 (as others have suggested). The percentage Green is an estimate based on a sample of publishers (107 publishers, publishing 9000 journals, and including the top journals in most fields), and the relative amount of OA via Green and Gold is in the ratio of 3/1 regardless of what the true percentage of the total is.

13 It must be noted that until publishers gave their various forms of Green light to self-archiving, its very possibility was very problematic at best. From the standpoint of intellectual property laws, no one has tested Harnad's tactic of archiving two files (submitted file plus corrigenda)-a point which would worry any university manager in charge of an institutional repository. This approach has to be tested in at least two ways: with regard to the notion of derivative work, and also plagiarism. It may sound strange to say that an author could be accused of plagiarizing himself or herself, but copyright law, let us remember, deals with property, intellectual property in this case, and signing away copy rights is signing intellectual property away. Copyright laws emerged in part to prevent an author from selling a manuscript to several publishers. Without the publishers' agreement, self-archiving is also problematic from a practical standpoint. I shall return to this point later.

All of this is irrelevant and unnecessary, and sounds like a call for (retrospective!) Zeno's Paralysis: . Authors who have been self-archiving since the early '90's have sensibly just gone ahead and self-archived instead of waiting for a green light from anyone (let alone worrying that once given, it might be taken back again)! For example, out of 300,000 papers self-archived by physicists since 1991, only 4 have since been withdrawn citing copyright considerations

Had the authors of those 300,000 papers instead thought along the lines J-CG seems to be thinking (and recommending?), over a decade of impact would have been needlessly lost for those authors -- as it was lost for that vast majority of authors who did not self-archive during that decade!


15 See, for example, Stevan Harnad's reaction to an article in an Indian publication at Harnad/Hypermail/Amsci/3156.html (accessed November 8, 2003).

16 See, for example, Harnad/Hypermail/Amsci/3161.html (accessed November 12, 2003) where Stevan Harnad writes: "I'm afraid that all this eminently accessible open access will continue to be needlessly delayed as long as our attention and enthusiasm continue to be directed solely or primarily at the slower road. We should really be promoting both roads, and each in proportion to its immediate capacity to deliver Open Access. What is happening now is instead rather like trying to increase the population by promoting in vitro fertilization alone, neglecting the faster, surer path..." Note, in passing, the rhetoric: Gold is to Green as in vitro fertilization is to natural fertilization! The metaphor is funny because it caricatures the situation. But it is only a caricature, not an analysis.

17 See harnad/Hypermail/Amsci/3162.html. Harnad estimates that 10% of all articles are in Open Access. Of these, one fourth or 2.5% of all articles published appear in Gold publications while about three fourths or 7.5% of all articles published appear in Green titles. While 85% of all articles could potentially be placed in Open Access, about a tenth of that quantity actually is.

18 Harnad, Harnad/Temp/self-archiving.ppt, slides 42-43.

19 Steve Hitchcock, Tim Brody, Christopher Gutteridge, Les Carr, and Stevan Harnad, "The Impact of OAI-based Search on Access to Research Journal Papers" (September 2003),

20 Some authors have defined impact differently. For example, Sidney Redner suggests to multiply the total number of citations by their average age. This suggests that measuring impact is not as simple and transparent as a simple citation count suggests, but I shall not address this question here and will act as if citation counts suffice. See S. Redner, "Citation Statistics from more than a Century of Physical Review" (July 27, 2004),

The latest findings on the self-archiving OA advantage show that it is present across years, for recent and 10-year-old articles alike. (What is not yet taken into account yet, however, is when the self-archiving was actually done.)

21 Steve Lawrence, "Online or Invisible?" lawrence/papers/online-nature01/. Edited version appears in Nature 411, no.6837 (2001): 521.

22 See, for example, Stevan Harnad, Tim Brody, Francois Vallires, Les Carr, Steve Hitchcock, Yves Gingras, Charles Oppenheim, Heinrich Stamerjoanns, and E.R. Hilf, "The Green and the Gold Roads to Open Access," Nature (Web focus) (2004), See also Michael J. Kurtz, "Restrictive Access Policies Cut Readership of Electronic Research Journal Articles by a Factor of Two" (2004),, and Andrew M. Odlyzko, "The Rapid Evolution of Scholarly Communication." Learned Publishing 15 (January 2002): 7-19,

23 See note 9 above.

24 Educom Review Staff, "Networked Information: Finding What's Out There-Clifford A. Lynch Interview," Educom Review 32-36 (1997),

25 Incidentally, why has no librarian, so far as I know, ever tried to implement a similar system on any campus? I have not systematically investigated this question and I would be delighted to stand corrected.

26 Kenneth Frazier, "The Librarians' Dilemma. Contemplating the costs of the 'Big Deal'", D-Lib Magazine 7 (March 2001) (3)

27, Elsevier's Scopus proceeds from the same argument. It also introduces fascinating implications about who will eventually control the search engines of science: Google, ISI's Web of Science, or Elsevier's Scopus?.

28 This is the case with the "Digital Object Identifier" (DOI). As stated in the DOI Handbook, "specifically, DOI relies on copyright and trademark law to protect the DOI brand and reputation. DOI is not a patented system; the IDF has not developed any patent claims on the DOI system and does not rely on patent law for remedy,"

29 See,

30 Carl Lagoze (Cornell University) and Herbert Van de Sompel (Los Alamos National Laboratories) are two of the leaders of the OAI-PMH protocol.


32 At the hearings of the UK Commons Select Committee, the Royal Society of Chemistry advanced this kind of argument in the following terms: "Currently most authors care where their work is seen and who it is seen by far more than they care about how many people have seen it," "Scientific Publications Free for All," the Science and Technology Committee of the UK House of Commons, vol. II, Oral and Written Evidence, p. EV-209 (p. 217, section 4.5 within a PDF reader). This statement is quoted in the main report (p. 9, item 8) and commented as follows: "This dispute goes to the core of the question of who should pay for the costs of scientific publications: those who argue in favor of the widest possible dissemination tend to be more receptive to the author-pays model of publishing; those who prefer targeting publications at a small, selected audience tend to be more content to maintain the status quo." 

Although it thus argues against Gold, the Royal Society of Chemistry has nevertheless gone Green:

Odlyzko, on the other hand, suggests that Open Access brings the literature to new categories of readers (and appears to enjoy it): "Much of the online usage appears to come from new readers (...) and often from places that do not have access to print journals." Odlyzko, "The Rapid Evolution," 8. As Odlyzko puts it, "... scholars ... are engaged in a 'war for the eyeballs'." Ibid., p. 9.

33 We are talking about impact factors here, as we are dealing with "Gold" journals. For good or bad reasons-probably bad ones in fact-most scientists are more familiar with impact factors than with impact (and their tenure and promotion committees also).

34 ADS = Astrophysics Data System. The part in brackets that clarifies Michael Kurtz' statement presumably comes from Stevan Harnad as moderator of the American-Scientist-Open-Access-Forum,

35 Ibid.

36 On this concept, see Michael H. Goldhaber, "The Attention Economy and the Net," First Mondgy, (accessed April 1997).

37 Odlyzko, "The Rapid Evolution.".

38 The original article is Samuel C. Bradford, "Sources of Information on Specific Subjects," Engineering 137 (January 26, 1934): 85-86. The law of concentration appears in Eugene Garfield, "The mystery of the transposed journal lists-wherein Bradford's law of scattering is generalized according to Garfield's law of concentration," Essays of an Information Scientist (Philadelphia, ISI Press, 1977): 222-223. The original article appeared in August 1971. Conversely, Garfield's law of concentration could be (ironically?) read as a way to justify a more pragmatic and relaxed attitude to the documentation search problem.

39 See for example the recent remarks by Heather Morrison, For a related argument, see Eugenio Pelizarri, "Harvesting for Disseminating. Open Archives and Role of Academic Libraries" to be published in January 2005 in the Acquisitions Librarian. Available online at

40 It works rather well, but it is not perfect, far from it. David Goodman, whom I thank, has attracted my attention on a study done by PŽter Jacs— ("PŽter's Picks and Pans CiteBaseSearch, Institute of Physics Archive, and Google's Index to Scholarly Archive," Online 28, no.5 (September 5, 2004): 57-58, showing that Google did not perform all that well on deep searches within Open Access databases. A summary of the results is found on Peter Suber's precious Weblog on Open Access: peters/fos/2004_08_29_fosblogarchive.html#a109406153195893347.


42 See the description of the article base at In a recent intervention, Stevan Harnad writes, "But Pubmed and PMC are not only better because of their better search features (which can all, of course, be fully duplicated by OAIster and by any other OAI search engine, whenever we wish to implement them!):..." However, if this search engine is so simple to duplicate, why is it not already done?.

Because OAIster does not yet have remotely enough OA content to make the effort worth its while! Most effort now needs to be invested in generating more OA content, not it in gussying up the little OA content we have!

43 Stevan Harnad, Harnad/Temp/self-archiving.ppt, slide 45: "Don't conflate the different forms of institutional archiving.".

44 It is important to recall that the varieties of Green involve a shade of Pale Green limiting "self-archiving" to preprints. In Stevan Harnad's powerpoint presentation ( Harnad/Temp/self-archiving.ppt, slide 41), the Pale Green publishers account for 30% of all publications but they are not treated separately, presumably on the basis that the preprint plus corrigenda strategy is realistic. Personally, I have always questioned the viability (and even legality) of the "self-archiving" strategy to the point that I had given very little credence to "self-archiving" before "real" Green publishers began to be identified in the RoMEO project.

Fortunately, the authors of, for example, the 300,000 physics papers and the 500,000 computer science papers that have been successfully and uncontestedly self-archived since the early '90's did not think the same way J-CG does in this matter! (Moreover, the percentage of Pale-Green journals has further shrunk from 30% to 13% since J-CG wrote these words?)

45 This remark applies particularly well to scientists in poor or in "transition" countries.

46 Harnad/Temp/self-archiving.ppt, slide 47.

47 See

48 See, for example, The Indian Academy of Science has also placed its journals in Open Access. See Subbiah Arunachalam, "India's march towards open access," (March 5, 2004),

49 Some ideas about China's evolution with regard to Open Access can be found in Liu Chuang, "Recent Development in Environmental Data Access Policies in the Peoples' Republic of China,"

50 For France, the best sites to find information on Open Access are HŽlne Bosc's site ( and the INIST site (

51 See, for example, Susanna Mornati, "Progetto AEPIC: gli archivi aperti italiani su une piattaforma nazionale," and Valentina Comba, AEPIC Academic E-Publishing Infrastructures-CILEA: Progetto di editoria elettronica per la ricerca e la didattica (2002),

52 Spain appears a little behind in the Open Access movement. However, the efforts of Crist—bal Pasadas Ure–a (University of Granada) must be noted (he pushes for Open Access within IFLA, for example). Likewise, Catalonia appears to be moving ahead, at least with theses and dissertations (

53 On both fronts, the Information Program of the Open Society Institute has been extremely active and useful.

54 The new Springer is the result of the merging of the old Springer plus Kluwer. The CEO for this new publishing behemoth is Derk Haank, formerly Reed-Elsevier's CEO. A better understanding of what is happening at the new Springer can be derived from the fascinating interview of Derk Haank, "Put up or Shut up," recently published by Richard Poynder (

55 Springer places a US$3,000 fee on its articles, i.e., twice as much as PloS. This is how Derk Haank explains this decision in his interview with Richard Poynder: "As always, I am very serious-$3000 is a very competitive price. Even Open-Access advocates would have to acknowledge that. The Wellcome Trust report, for instance, estimated the true cost of publishing a paper at more like $3500.".

56 On DSpace, see McKenzie Smith, "An Open Source Dynamic Digital Repository," It was originally published in D-Lib Magazine 9, no.1 (2003).

57 See Odlyzko, "The Rapid Evolution," p. 9.

58 This is the argument that David Kohl, for example, regularly gives in his talks. See, for example, "Better value from bigger deals: issues and experience" available from as a PowerPoint presentation.

59 The expression "overlay journal" may not satisfy all and other terms have been suggested, such as "Article Database" or "deconstructed journal." Debates and usage will eventually stabilize these terms. On the notion of "deconstructed journal, see John W. T. Smith, "The Deconstructed Journal-A New Model for Academic Publishing," Learned Publishing 12, no.2 (April 1999),

All these expressions are unsatisfactory, because the concepts behind them are premature and have not been carefully thought-through (let alone tested and shown to be viable!). If/when journals ever decide to become (1) online-only and to offload all (2) text-generation, (3) archiving and (4) access-provision on the author's institutional OAI archive network, (5) leaving the journal with only the service of peer review to perform, and its outcome to certify with its journal-name, then the journal-name will be the certification tag (no 'overlay journals' to speak of) and the journal will have become a peer-review service provider. (Still nothing like a 'deconstructed' journal, which is, and will probably remain, an untested, unrealistic, and probably incoherent hypothesis.) ).

60 A very recent (September 3, 2004) statement released by NIH completely supports this view. See: "Notice: Enhanced Public Access to NIH Research Information," Thanks to the guest editor for this issue of Serials Review for having attracted my attention to this document.


What is needed today is already quite clear: 100% OA by the fastest and surest means possible. It is also clear what that means is: self-archiving (Green), which now needs to be mandated by researchers? institutions and funders. There is also scope both for the growth of OA journals (Gold) and for experimentation with hypothetical new systems in parallel with the self-archiving of all peer-reviewed, published journal articles (Green) -- but not in place of it. Let there be no mix-up about that!

Acknowledgement: Many thanks to Alma Swan for her astute and helpful remarks on this critique.