Genevieve Fioraso, French Minister of Higher Education and Scientific Research, had declared just 18 months ago:
"Au regard de l'importance des enjeux, sur les plans scientifique, économique et sociétal, le gouvernement français réaffirme son soutien au principe du libre accès à l'information scientifique."
Or
"Given the high stakes - scientific, economic and societal - the French government reaffirms its support for the principle of open access to scientific information."
I just don't get it. These editors offer no real added value (paper formatting and a website is not worth 172 Million a year), and we do nothing about it?
Of course they see a dime: they receive public money. Elsevier are such geniuses: public money paid papers need to be paid a second time, and still, they remain unavailable for the public who has paid for it (twice). Rich geniuses.
Al-Khwarizmi was not talking about the publisher but about the scientists actually doing the peer-reviewing work: they don't see a dime from the publisher for this work (which is normal because it is part of their job as scientist to peer-review, what is abnormal is that the result of this work is given for free to private companies who put it behind paywalls).
I wasn't implying that the publishers but the reviewers were getting public money for that. Having spent around 12 hours between Sunday and Monday to write a single review for a journal paper, I must say that I am pretty surprised that this system has been working until now… The scientists are united in making their reviews, while someone on top is getting so much money for it (this catches up with your argument). I can't help myself considering us as naive people.
I see. Naive I don't know, not interested by money, certainly. Otherwise we would be following much more lucrative career paths, which are well, basically anything else than public research.
That said, I've encountered many scientists who simply never thought about it before we discussed it, and who now think the situation is absolutely crazy, as anyone would. The questions are: What can they do about it? Do they have the time to do it? How can they do it without hurting themselves?
Indeed, it's hard enough to get a position as it is, refusing to publish in prestigious journal owned by asshole publishers won't help young scientists. Older scientists often work with young ones, and they can't tell them "no we won't publish in the asshole-owned prestigious journal because it's the right thing to do", it wouldn't be fair.
The reality is that the solution can't really come only from individuals, it has to be a legal battle, not only for future publications (which is the easiest part), but also for all the past ones. The entire Elsevier "Freedom Collection" for example should be seized by governments (UN?) and made available for free to anyone on the internet and in public libraries worldwide. But in this "free-market" economy, this is not something people seems to be ready to ear, let alone understand.
> Indeed, it's hard enough to get a position as it is, refusing to publish in prestigious journal owned by asshole publishers won't help young scientists
It seems to vary strongly by field though, some fields seem much more bullish on "open access" than others.
I think the solution has to come from public funding agencies (i.e. from governments).
If my country's government says that from 2015 onwards, only publications that can be freely accessed count towards public grant requests, tenure and positions in public institutions, etc.; in a year no one would be published in the asshole-owned journals.
Instead, my government has policies that actively incentivize publishing in said journals (basically ISI JCR journal papers are what counts for everything).
>What can they do about it? Do they have the time to do it? How can they do it without hurting themselves?
Get the top people in a field to collectively decide to create/back a new journal. Because they are the top, what they publish in will "automatically" become prestigious, assuming the journal is selective. Once the journal is well established, the up and coming will start publishing there too.
> Of course they see a dime: they receive public money.
Scientists receive money for the work they do, but that money is generally in the form of grants for their own sciencing, I've never heard of "review grants".
The "peer" part of "peer review" means it's other scientists in the field doing the reviewing. The editor only acts as matchmaker/anonymisation filter, and maybe does the most trivial part of reviews (typos, layout, that sort of thing).
Why, then you simply get an incredibly negative review, or one with impossible requirements, or some other nebulous reason why your work should not be published in that journal.
Source: currently working in academia in a full-time research position
No, I mean, that there is some probability that someone intelligent within the system is manipulating the infrastructure of the peer review system for an advantage.
I don't understand why I was downvoted for asking a question. I just meant to imply that there must exist some middle infrastructure, and I do not believe that it is a trivial problem to solve.
Source: I studied computer security systems that attempted to automate the flow of information to maintain security level thresholds on information types. When code 'touches' other code, meaning that it uses code with information at one security level to reason about code with information at another security level, unless you are particularly careful in the structure and organization of the code, information can leak out, structures can be manipulated, rules can be broken.
I am not saying scientists are unethical, I am saying they are human and prone to emotional reaction occasionally before objectivity (unless they are perfect). Combining information inference, intelligence, and an understanding of the flaws of an automated system; they are capable of 'accidentally' tweaking those things in their favor. It's not in the interest of science when this happens, and it requires a substantial amount of real intelligence to moderate. It's basically 'accidental bias'.
About the possible flaws in the reviewing process, the open-access publisher Frontiers (recently acquired by the Nature Publishing Group) proposes to reveal the names of the reviewers once the paper has been accepted, since they take part in the writing process:
"Frontiers is striving to remove any bias from the review process and acknowledge the reviewers for the significant contributions in improving the paper. To guarantee the most transparent and objective reviews, the identities of review editors remain anonymous during the review period. Only in case an article is accepted do their names appear on the published manuscript, without exceptions. However, if for any reasons a review editor withdraws during any stage of the review process, his/her name will not be disclosed." [Source : http://www.frontiersin.org/Design/pdf/ReviewGuidelines.pdf]
> I don't understand why I was downvoted for asking a question.
Because your question, in the context of the thread, appeared to imply you think Elsevier and co. are doing something essential, in return for their 40% profit margins resulting in billions (!) of profit per year. Taken by itself it's an interesting question.
I don't really know, honestly. When groups are arguing between opposing world views, all of which can appear equally likely depending on data selection, I find that absolutely frightening that our current scientific process of peer review can not help us tell our heads from our asses. Or it does so often enough, that it's doing a very good job at keeping the bulk of knowledge stable. Science can be frightening.
Personally, I think it might take a lot more time than I've been alive to come close to a solution. I don't feel comfortable having an opinion on this, but I'm fine having questions. I don't intend to make implications with my questions, they are not consciously intentionally leading.
It's only been about 15 generations since the time of Galileo. I think science'll muddle through even if a couple of generations are wasted on dead-ends along the way.
There's stuff you expect (or have read, heard, thought about, observed) science to be about, and there's stuff you don't expect science to be about. The stuff you don't expect science to be about is the stuff that concerns me.
I don't know whether I can convey the change in understandings I've had about science since leaving academia. I don't want it to be typical, nor do I want it to be atypical in that the definition of atypical is dependent on the opposite of what is typical.
I don't have firsthand experience publishing in a peer reviewed journal, but from all my experience in academia, some of which involved very close observances, it doesn't seem to be a perfect solution.
As other pointed out, the peer reviewing is not done by the publishers. It is done by other scientists and coordinated by the editorial board of the journal, which most of the time consists in more scientists who are not paid by the publishers either.
However, I don't see any point in downvoting you as it seem people did, what you ask is a very common misconception that benefits to publishers and is thus amplified by them.
As an author, the added value is the name of the journal. The pedigree of the journals you have published in is one of the primary qualities that grant and hiring committees look at. The whole system is obviously a scam, but you have to play the game if you want a career in science.
This is why we have to fight for the right Open Access model.
Most publishers offer "Gold Open Access", which is "authors pay" instead of "readers pay", and of course the price is decided by the publisher…
This is not what Open Access means, and it doesn't solve the main problems.
Diamond Open Access focuses on the real problems.
1- It should be free to publish, free to read, free to reuse. The Diamond Open Access model implies free licenses such as Creative Commons BY for articles and data.
2- Rather than being owned by publishers, the journals have to be owned by their editorial board, which itself has to be a legal entity of some sort. This would avoid problems such as prestige inertia which forces researcher to publish with asshole publishers: if the scientists actually running the journal own it, they could have switched to open access and/or to a different publishers when needed.
Apart from that, we of course have to fight against bibliometrics…
Money comes from the same place it always did: funders (usually the public purse). The difference is in the OA model the cost of publishing is included in grants, rather than through libraries and other routes.
Instead of having each university pay millions each year to subscribe to privately owned journals, we use (a part of) this money to develop and maintain the necessary infrastructures (servers, software developers, librarians, …). The rest of the money could be used for additional research.
Open access can be mutually independent of "selective" for a journal. How readers access a journal has nothing to do with the pedigree and the quality of the articles it publishes.
It's not true that we do nothing about it. What is true is that it is hard to be heard on that subject, and even harder to act practically on it. Since you are French and interested by this subject, I suggest that you read my (rather long) introduction to open access, here: http://pablo.rauzy.name/openaccess/introduction.html.
The added value of publishers is more than just the editing. Why do you think a published paper is "worth" more than a pdf on a random website?
- Almost nobody in academia is willing to publish in substandard journals only because they are open access. The publishing process is a necessary filter for people to select what is important enough to read.
- You _need_ publishers that make a profit and stay healthy over centuries. If Elsevier went bancrupt, they could no longer take care for their archives. And in the long run, we would loose access to several hundred years of research.
It's a problem in the first place that these archives are held by a private company.
Why do they keep it centuries of it by the way? Because they hold the copyrights of the articles, but isn't that supposed to dissolve at some point in time after the death of the authors?
The enormous amount of public money thrown away at publishers by each individual country is more than enough to conserve all the archives of all the published papers, past, present, and future.
If Elsevier went bankrupt, and actually even if it does not, we should have them hand over the archives so that they are taken care of by public institutions who will not put barriers around their contents.
Also, you imply that "open access publication = substandard publication", which is entirely wrong. There are countless counterexamples.
I choose my words carefully. Please read them carefully.
1. My argument was never about the quality of open-access journals. Given the choice between a closed journal and a "lesser" open-access journal, most (but not all) authors would still try to publish in the closed one. There are highly regarded open access journals. But they are comparatively few.
2. The publishers already retain the rights of hundreds of years of published papers. They will not give that away to some (governmental or otherwise) science-preservation organisation. So the only way to "safe" these results is to keep the publishers alive.
I would like the world to be different. And I would have loved to see aaron swartz "liberating" more paper archives. But that is a different story.
But I do not agree with your second point, I don't see what is the problem. Of course they won't agree! Who cares? The law could force them to hand the archive over. And if they don't, public authorities could help to enforce such a law, even if a physical intervention is necessary to take the archives of humanity's knowledge back from them.
I know this sounds "too much" for most people, but for me the current situation is not acceptable.
> If Elsevier went bancrupt, they could no longer take care for their archives.
I could think of more than a few entities that would love to take over their archives. archive.org would likely volunteer, and it's in the interests of every single university and corporate entity that uses these papers that they continue to exist. I could very easily see universities agreeing to share the storage of the archives.
If the end goal is to have guaranteed storage of these papers... why on earth is a for-profit company which could theoretically go bust despite all best efforts in charge of looking after them?
You cannot just take intellectual property away from some company. And if a single country tries, they just hand the rights over to the next-best subsidiary on the other side of the world.
> You cannot just take intellectual property away from some company.
If the company's about to die, why not? Your argument only works when the company's still going strong. If it's never going to actually make money with its IP, it's entirely illogical for it to spend huge amounts of effort trying to prevent that IP from being made public.
JSTOR is the default entity that would take care of archives. They provide access to articles older than 5 years (in most cases) for many journals from both commercial and academic publishers.
The “editors” of an academic journal are usually academics, professors who are employed by universities, and they don’t sit at the offices of Elsevier or another publisher. They assign referees and make final decisions on whether to publish an article. They are sometimes paid, but not always, and usually do their work as a service to the profession.
The publisher does copy-editing, typesetting, distribution, and other tasks. They sometimes pay for administrative support for the editors, but that is often paid by the editors’ own universities.
For journals in the fields that I’m familiar, what often ends up happening is the authors do all the typesetting perfectly with LaTeX. Then they submit a PDF, along with a Word file with all the text. Some contractor of the publisher does the typesetting all over again, using LaTeX!
> These editors offer no real added value (paper formatting and a website is not worth 172 Million a year),
I believe some of the Elsevier journals have formatting which is mostly done by the authors (via existing LaTeX templates). So they are not even adding that type of value. Copy-editing—which the journals generally provide—does improve things, but probably not 172 M/yr worth.
Genevieve Fioraso, French Minister of Higher Education and Scientific Research, had declared just 18 months ago:
"Au regard de l'importance des enjeux, sur les plans scientifique, économique et sociétal, le gouvernement français réaffirme son soutien au principe du libre accès à l'information scientifique."
Or
"Given the high stakes - scientific, economic and societal - the French government reaffirms its support for the principle of open access to scientific information."
I wonder if it'd be possible to short-circuit this process? It'd be difficult, but many states have public document laws. If the manuscript is written physically using state-university equipment, I'd hope there's a nexus for the public to get that document. To try this I'm thinking about submitting a request to my alma mater for one such document. The request might need to be massaged and resent a couple of times. At that point, what's preventing me from asking for say the top 100 papers that are cited on GScholar? Hacking the system in this way could be pretty interesting.
Actually, it should be checked by a lawyer, but it is entirely possible that the copyright transfer agreements that scientists sign when publishing in asshole-publisher owned journals have no legal value in France. The reason for that would be that: 1- the form is only signed by one party (the authors), 2- there is no counterparts to the copyright giveaway, and 3- the signature of a single authors is sufficient to give the consent of all the others, who may be from a different institution or even country.
Normally, scientists retain the copyright to their work, but sign an exclusive license to publish the work.
Often this license is based on the additional work done by the journal (minor proof-reading and formatting), that's why preprint servers always serve the unformatted version (or in the format of the preprint server).
For Science, for example:
>Authors can immediately post the accepted work to their personal or institutional archive.
>The accepted version is the paper that was accepted for publication by AAAS, including changes resulting from peer review, but prior to AAAS copyediting.)
Same goes for Nature:
>NPG does not require authors of original (primary) research papers to assign copyright of their published contributions. Authors grant NPG an exclusive licence to publish, in return for which they can reuse their papers in their future printed work without first requiring permission from the publisher of the journal.
I don't know that it is normal yet for authors to retain their copyrights. Certainly, Elsevier's policy is to require a copyright transfer unless the authors pay an open-access fee.
There is the arXiv[0]. In some fields—particularly physics and astronomy—the vast majority of papers are posted there, in addition to being available through the traditional journals. Not all papers have been peer-reviewed before being posted, but many have, an others are often updated to match their peer-reviewed versions. The formatting and copy-editing may be slightly different, but the scientific content generally matches quite well.
There should be a law that prohibits publishers from claiming the copyrights of articles that describe work which is publicly funded. Its a no brainer and would allow authors to maintain their preprints in an open archive.
Because it has survived for more than two centuries, publishes several hundred thousand scientific articles a year, and has huge profit margins? There are problems with the Elsevier-model of scientific publications, but longevity doesn't appear to be one of them.
I do not disagree with the gist of your argument, but Elsevier isn't centuries old. It was founded in 1880. Also, it only entered the scientific publishing business in 1933.
I thought you were thinking of the Elsevier family, who started printing in 1583 (the modern Elsevier company recycled the name; see http://www.elsevier.com/about/history)
Come on, the profit margins are basically rent-seeking. Elsevier is not offering anything that can't be replaced overnight, they just have a tight grip on the egos of scientists and a funding system that feeds them, thats all.
A solution would be for universities to host the journals. They can hire salaried staff for proof reading/marketing etc. Scientists can volunteer for peer review. The journal can be as selective as it wants. Once an article is accepted for publication, it's hosted on the university website.
The indirect costs from grants can cover the salaries of the admin staff - and very few staff are needed because the authors themselves can give the staff a good head start by using latex to prepare the manuscript.
While France signs with Elsevier, many scientists around the world have been protesting against Elsevier in a movement called "The Cost of Knowledge".
The Open Journal (http://theoj.org) is an example of doing open science well and efficiently. All the papers are hosted in arXiv, which means they are publicly accessible. And the review is performed using their own open source platform.
78 comments
[ 4.9 ms ] story [ 152 ms ] thread"Au regard de l'importance des enjeux, sur les plans scientifique, économique et sociétal, le gouvernement français réaffirme son soutien au principe du libre accès à l'information scientifique."
Or
"Given the high stakes - scientific, economic and societal - the French government reaffirms its support for the principle of open access to scientific information."
Reference: http://www.enseignementsup-recherche.gouv.fr/cid66992/discou...
Shame that she didn't follow through!
I just don't get it. These editors offer no real added value (paper formatting and a website is not worth 172 Million a year), and we do nothing about it?
Peer reviewing maybe?
That said, I've encountered many scientists who simply never thought about it before we discussed it, and who now think the situation is absolutely crazy, as anyone would. The questions are: What can they do about it? Do they have the time to do it? How can they do it without hurting themselves?
Indeed, it's hard enough to get a position as it is, refusing to publish in prestigious journal owned by asshole publishers won't help young scientists. Older scientists often work with young ones, and they can't tell them "no we won't publish in the asshole-owned prestigious journal because it's the right thing to do", it wouldn't be fair.
The reality is that the solution can't really come only from individuals, it has to be a legal battle, not only for future publications (which is the easiest part), but also for all the past ones. The entire Elsevier "Freedom Collection" for example should be seized by governments (UN?) and made available for free to anyone on the internet and in public libraries worldwide. But in this "free-market" economy, this is not something people seems to be ready to ear, let alone understand.
It seems to vary strongly by field though, some fields seem much more bullish on "open access" than others.
If my country's government says that from 2015 onwards, only publications that can be freely accessed count towards public grant requests, tenure and positions in public institutions, etc.; in a year no one would be published in the asshole-owned journals.
Instead, my government has policies that actively incentivize publishing in said journals (basically ISI JCR journal papers are what counts for everything).
Get the top people in a field to collectively decide to create/back a new journal. Because they are the top, what they publish in will "automatically" become prestigious, assuming the journal is selective. Once the journal is well established, the up and coming will start publishing there too.
Scientists receive money for the work they do, but that money is generally in the form of grants for their own sciencing, I've never heard of "review grants".
http://en.wikipedia.org/wiki/The_Cost_of_Knowledge
Source: currently working in academia in a full-time research position
I don't understand why I was downvoted for asking a question. I just meant to imply that there must exist some middle infrastructure, and I do not believe that it is a trivial problem to solve.
Source: I studied computer security systems that attempted to automate the flow of information to maintain security level thresholds on information types. When code 'touches' other code, meaning that it uses code with information at one security level to reason about code with information at another security level, unless you are particularly careful in the structure and organization of the code, information can leak out, structures can be manipulated, rules can be broken.
I am not saying scientists are unethical, I am saying they are human and prone to emotional reaction occasionally before objectivity (unless they are perfect). Combining information inference, intelligence, and an understanding of the flaws of an automated system; they are capable of 'accidentally' tweaking those things in their favor. It's not in the interest of science when this happens, and it requires a substantial amount of real intelligence to moderate. It's basically 'accidental bias'.
"Frontiers is striving to remove any bias from the review process and acknowledge the reviewers for the significant contributions in improving the paper. To guarantee the most transparent and objective reviews, the identities of review editors remain anonymous during the review period. Only in case an article is accepted do their names appear on the published manuscript, without exceptions. However, if for any reasons a review editor withdraws during any stage of the review process, his/her name will not be disclosed." [Source : http://www.frontiersin.org/Design/pdf/ReviewGuidelines.pdf]
- Independent Peer-Review Initiative http://www.openscholar.org.uk/independent-peer-review-initia...
- Academic self-publishing: a not-so-distant-future http://www.openscholar.org.uk/academic-self-publishing-a-not...
Because your question, in the context of the thread, appeared to imply you think Elsevier and co. are doing something essential, in return for their 40% profit margins resulting in billions (!) of profit per year. Taken by itself it's an interesting question.
Personally, I think it might take a lot more time than I've been alive to come close to a solution. I don't feel comfortable having an opinion on this, but I'm fine having questions. I don't intend to make implications with my questions, they are not consciously intentionally leading.
I don't know whether I can convey the change in understandings I've had about science since leaving academia. I don't want it to be typical, nor do I want it to be atypical in that the definition of atypical is dependent on the opposite of what is typical.
They filter unfair peers by having multiple reviewers, and so can anyone else.
And papers generally get multiple reviewers (though that can be hard, especially while keeping things anonymous, in very small/tight fields)
My point is that it requires intelligence to moderate and create balance between intelligence(s).
However, I don't see any point in downvoting you as it seem people did, what you ask is a very common misconception that benefits to publishers and is thus amplified by them.
Most publishers offer "Gold Open Access", which is "authors pay" instead of "readers pay", and of course the price is decided by the publisher…
This is not what Open Access means, and it doesn't solve the main problems.
Diamond Open Access focuses on the real problems.
1- It should be free to publish, free to read, free to reuse. The Diamond Open Access model implies free licenses such as Creative Commons BY for articles and data.
2- Rather than being owned by publishers, the journals have to be owned by their editorial board, which itself has to be a legal entity of some sort. This would avoid problems such as prestige inertia which forces researcher to publish with asshole publishers: if the scientists actually running the journal own it, they could have switched to open access and/or to a different publishers when needed.
Apart from that, we of course have to fight against bibliometrics…
You may have a career, but i doubt that qualifies one as a scientist.
It's not true that we do nothing about it. What is true is that it is hard to be heard on that subject, and even harder to act practically on it. Since you are French and interested by this subject, I suggest that you read my (rather long) introduction to open access, here: http://pablo.rauzy.name/openaccess/introduction.html.
- Almost nobody in academia is willing to publish in substandard journals only because they are open access. The publishing process is a necessary filter for people to select what is important enough to read.
- You _need_ publishers that make a profit and stay healthy over centuries. If Elsevier went bancrupt, they could no longer take care for their archives. And in the long run, we would loose access to several hundred years of research.
It's a problem in the first place that these archives are held by a private company. Why do they keep it centuries of it by the way? Because they hold the copyrights of the articles, but isn't that supposed to dissolve at some point in time after the death of the authors?
The enormous amount of public money thrown away at publishers by each individual country is more than enough to conserve all the archives of all the published papers, past, present, and future.
If Elsevier went bankrupt, and actually even if it does not, we should have them hand over the archives so that they are taken care of by public institutions who will not put barriers around their contents.
Also, you imply that "open access publication = substandard publication", which is entirely wrong. There are countless counterexamples.
1. My argument was never about the quality of open-access journals. Given the choice between a closed journal and a "lesser" open-access journal, most (but not all) authors would still try to publish in the closed one. There are highly regarded open access journals. But they are comparatively few.
2. The publishers already retain the rights of hundreds of years of published papers. They will not give that away to some (governmental or otherwise) science-preservation organisation. So the only way to "safe" these results is to keep the publishers alive.
I would like the world to be different. And I would have loved to see aaron swartz "liberating" more paper archives. But that is a different story.
But I do not agree with your second point, I don't see what is the problem. Of course they won't agree! Who cares? The law could force them to hand the archive over. And if they don't, public authorities could help to enforce such a law, even if a physical intervention is necessary to take the archives of humanity's knowledge back from them.
I know this sounds "too much" for most people, but for me the current situation is not acceptable.
I could think of more than a few entities that would love to take over their archives. archive.org would likely volunteer, and it's in the interests of every single university and corporate entity that uses these papers that they continue to exist. I could very easily see universities agreeing to share the storage of the archives.
If the end goal is to have guaranteed storage of these papers... why on earth is a for-profit company which could theoretically go bust despite all best efforts in charge of looking after them?
You cannot just take intellectual property away from some company. And if a single country tries, they just hand the rights over to the next-best subsidiary on the other side of the world.
If the company's about to die, why not? Your argument only works when the company's still going strong. If it's never going to actually make money with its IP, it's entirely illogical for it to spend huge amounts of effort trying to prevent that IP from being made public.
The publisher does copy-editing, typesetting, distribution, and other tasks. They sometimes pay for administrative support for the editors, but that is often paid by the editors’ own universities.
I believe some of the Elsevier journals have formatting which is mostly done by the authors (via existing LaTeX templates). So they are not even adding that type of value. Copy-editing—which the journals generally provide—does improve things, but probably not 172 M/yr worth.
"Au regard de l'importance des enjeux, sur les plans scientifique, économique et sociétal, le gouvernement français réaffirme son soutien au principe du libre accès à l'information scientifique."
Or
"Given the high stakes - scientific, economic and societal - the French government reaffirms its support for the principle of open access to scientific information."
Reference: http://www.enseignementsup-recherche.gouv.fr/cid66992/discou...
Shame that she didn't follow through!
Often this license is based on the additional work done by the journal (minor proof-reading and formatting), that's why preprint servers always serve the unformatted version (or in the format of the preprint server).
For Science, for example: >Authors can immediately post the accepted work to their personal or institutional archive. >The accepted version is the paper that was accepted for publication by AAAS, including changes resulting from peer review, but prior to AAAS copyediting.)
Source for Science: http://www.sciencemag.org/site/feature/contribinfo/prep/lice...
Same goes for Nature: >NPG does not require authors of original (primary) research papers to assign copyright of their published contributions. Authors grant NPG an exclusive licence to publish, in return for which they can reuse their papers in their future printed work without first requiring permission from the publisher of the journal.
http://www.nature.com/authors/policies/license.html
This actually vary between journals.
For the ACM Digital Library for instance I was given the choice between the two options (and of course chose the license one).
http://www.elsevier.com/journal-authors/author-rights-and-re...
[0] http://arxiv.org
The indirect costs from grants can cover the salaries of the admin staff - and very few staff are needed because the authors themselves can give the staff a good head start by using latex to prepare the manuscript.
Why can't this be done?
The Open Journal (http://theoj.org) is an example of doing open science well and efficiently. All the papers are hosted in arXiv, which means they are publicly accessible. And the review is performed using their own open source platform.