Programme on publishing

[Sarah Bridle]

Not sure which section to put this in, but anyway.

I found this radio programme quite thought provoking: http://www.bbc.co.uk/radio4/science/publishorbedamned.shtml

Not sure I agree with all of it but it gets more interesting towards the end, talking about the cost of journals to libraries (~100 pounds per issue). Explains a bit about some new journals which charge the authors per paper (~800 pounds) and then the papers are free for anyone in the world to look at.

How can it cost this much to publish papers in journals? We don't need typesetting (what appears on the arxiv always looks fine to me). We don't get paid to referee them. We basically need someone to assign referees to papers and deal with any arguments before checking a box saying "yes this has been accepted". I suppose the editors do this - do editors get paid so very much?

The programme mentions the open archives (e.g. arxiv) and says that, surprisingly, journal subscriptions have not been affected by ~95 per cent of the articles being available on the arxivs.

I don't really understand this though. I have to admit, I almost always get papers from arxiv rather than getting the journal version through ADS and the university subscriptions. Many journals are free to publish in, we get the benefit of having the peer review stamp on our cvs, and then we download the papers for free from arxiv.

So why bother subscribing to journals?

[Syksy Rasanen]

I actually often get papers from ADS (or the PRD pages, as the case may be) rather than arXiv, since published versions have at times benefited from the referee's attention. (But I'm not so motivated as to go to the library to look at the paper version, if it not available on the web - especially as the physics libraries were recently shut down here at Oxford). If people would post the published versions on the arXiv, there would be no need for this, of course.

I have also wondered about the high cost. Refereeing is necessary, but can't really cost that much. Is the reason just the publishers' drive for profit?

[Antony Lewis]

Sometimes people update the Arxiv with corrections after the published version - as long as people do this (and I think they should) the Arxiv should be a more reliable resource than the published version. The only paper from the last decade I've had to dig out of an online journal in the last year was one with zillions of figures in which was probably too large for what Arxiv allows.

An example of a purely electronic journal is JCAP. This is charged people in rich western countries to 'cover costs' - software development and editorial time presumably. However this still seems to be a whopping 850 quid (~US$ 1500) a year. I suspect libraries subscribe because they don't realise how effectively useless a subscription is, and historically the number of subscribed journals has been quoted as a sign of quality of a particular institution and its library (if you spend more, must be better). Most other subjects are also miles behind astrophysics/hep in terms of online arxives - which seems a tragedy, esp. for the medical sciences.

I agree with Sarah - ultimately a `journal' should just be a quality mark that appears with a particular online version of an article in an online repository (i.e. Arxiv). There is no need at all for separate version on a journal website, and all journals could be run with effectively the same highly automated software. [with the exception of things like Science/Nature/BMJ/etc that are nice to have in paper form to flick through over coffee]. In my opinion the referees' reports should be published so dissenting views and ignored comments can be seen, and allow assesment of the rigour of the refereeing process (which currently seems to be very variable). The site should also allow third party comments as on CosmoCoffee and various maths journals. It might also be an idea to have two versions - one for online viewing and one for [q]B&W[/q] printing.

If a condition of `publication' is that you do your share of editing and refereeing, the costs should be restricted to computer issues and someone to manage and appoint editors (editors could appoint replacement editors, or there could be a voting system once the system is in place). I would have thought the entire academic publishing industry could be replaced for less than 1 million dollars a year (1/70 of what Nottingham currently spends).

One possible caveat is with papers from non-English authors that may benefit substantially from time consuming language editing. This cost would at least be optional, and could be borne by the submitters.

PS. Since people like me only ever use the arxiv, it would be very nice if papers always included arxiv references along with journal references. Using e.g. BibTex and RevTex4 this can be done automatically, and would save the unneccessary process of doing an arxiv search to find each referenced paper you want to look at --- which can be unobvious in the case of several papers by the same authors in the same year. I find proper LaTeXed equations significantly easier to read than what comes out of e.g. Phys. Rev. D's production process.

[Samuel Leach]

On the issue of typesetting: a large part of Don Knuth's motivation for implementing LaTeX was becuase of what he saw as the declining standards in typography in the 60's and 70's at a time when he wanted to publish his Art of Computer Programming series. I agree with him that if we're going do dedicate a lot of time and effort to our work then the end result deserves the highest standards of presentation. That means typesetting is a service that needs to be provided by a professional. Although we can go quite far with LaTeX, which has clearly contributed quite a bit to the continuing evolution of publishing in physics and mathematics, as far as I know it's practically unknown outside these disciplines.

On the arxiv: Clearly it's a major piece of infrastructure in our discipline, but I personally don't see it evolving into a complete replacement for journals in some form or other. The notion that refereeing is not important is one that should be resisted, but I agree that the process could be made more effective. What about a published commentary of important papers, like in many statistics journals.

Finally, maybe you know that the latest round of changes to scientific publishing are occurring very rapidly in the last few years especially in the Open Access movement. A very good quality news blog by Peter Suber can be found at

http://www.earlham.edu/~peters/fos/fosblog.html

[Syksy Rasanen]

I agree with Antony. In particular, I think that making referee reports openly available would be an important step. Both because goodl referee reports can add insight that others can also benefit from, and because careless reports would be easy to discount. This would also encourage people to be more careful with their reports, even if they're still anonymous. As it is, there is little incentive to look carefully at papers one is refereeing. Only the authors and editors will see the report, there is no merit to be gained from allocating time to more careful refereeing, and no demerit in writing a useless report, whether it's "OK, publish" or "rubbish, don't publish".

(BTW, I too find PRD's layout hard to read. It's perverse that a journal is being paid to make the layout worse!)

[Guest]

None of this is quite as simple as it appears.

Publishers do put considerable effort into the editorial process with the high quality journals having very high reject rates (80% - 90%). The cost of sales and marketing and providing access through online systems are also a factor. Also most journals remain in both print and electronic mainly because libraries do not trust publishers to provide them access over the long term.

This is not to say there isn't quite a lot of inefficiency in the current model and the challenge from online respositories to the old models is very real.

We at the British Library are looking to find new ways to support the research process through a range of information services that I think will look quite different from the existing approaches.

Richard Boulderstone
Director eStrategy
British Library

[Antony Lewis]

Let me make this into a concrete proposal for "Open Journals" (this is not a new idea; Sarah and I and others have discussed this before). The idea is to separate entirely the process of distribution and of journal stamping.

* The Arxiv currently has everything that is needed for the distribution process, including document submission, updates, distribution, mirroring and backup. In practice many astrophysicists already get all their papers from this site.

* The purpose of an Open Journal should then be to assign journal quality marks to papers submitted to the Arxiv. This could be integrated into Arxiv, or a separate website. One particular edition of a paper on the Arxiv would have a special "published in journal X" marking once accepted after the refereeing process. A paper could be given an extended arxiv-based reference, e.g.
[q]Open-Journal-Comology/astro-ph/xxxxx[/q] which simultaneously indicates the journal stamp, and the universal Arxiv reference.

* There is no separate "journal published" version, removing the redundancy, duplication and confusion of the current system where one paper version is on arxiv, and some other slightly different version is sitting on a publisher's web site.

* Since "published" papers are on the Arxiv, the arxiv becomes more up to date and reliable: there is no question of published papers not being updated in the arxiv version. (and if corrections need to be made later, these can be made by updating the Arxiv version as they can be now)

* The arxiv paper page is linked to the web web page to authenticate the journal quality mark, and to let people view the referee reports and comment on the paper if they wish. Prior to publication referees and editors should be encouraged to read third-party volunteered comments, as a way of improving the rigour of the refereeing process and making sure that dissenting views are known before acceptance.

* Users can use customized filtered version of the Arxiv, so if they wish they only see reliable published papers. But there is also the opportunity as now to see papers as soon as they are submitted, removing unneccessary delay in the dissemination of important papers. I think it is already almost accepted that Arxiv submission is what counts for priority, not journal publication date. Filtering is trival, as on the CosmoCoffee Arxiv New page.

Publishers do put considerable effort into the editorial process with the high quality journals having very high reject rates (80% - 90%).

Surely it is editors that put a lot of time into this not the unqualified publishers. In an Open Journal editorial services and refereeing would be supplied free by people instead of payment for submission. In fact most academic posts already stipulate in the job description that you are expected to do your bit for the refereeing process: there is no reason full-time employed academics should be paid separately for academic publishing services.

The cost of sales and marketing and providing access through online systems are also a factor.

An Open Journal does not have any of these costs. There is nothing to sell and online access is provided entirely via the Arxiv (which is funded via Cornell, NSF, etc). Computer costs are minimal, and the Open Journal website wouldn't be all that more complicated than CosmoCoffee.

Also most journals remain in both print and electronic mainly because libraries do not trust publishers to provide them access over the long term.

The point is that papers in electronic journals (at least in astrophysics) are also on the Arxiv, which is freely publically available. The online journals are merely duplicating content that is available free elsewhere (in a centralized and more convenient form). The separate electronic journals are redundant.

In the Open Journal era the only role of libraries would then be to mirror/backup the global publically available database of papers. If an Open Journal closes, the papers will remain on the Arxiv with the appropriate journal stamp. There is no risk of papers disappearing if your library has a backup of the entire database.

Unfortunately Arxiv currently only supports a limited range of subject areas, but I don't see any reason in principle why it shouldn't be extended to all areas. The Open Journal web site could handle an arbitrary number of different journals with the same software, submission would be free, and papers would be globally freely available to anyone from the Arxiv. Each journal could choose its own criteria for the number of referees, importance, length, style of papers etc, as now.

Any objections? Would people submit to and support an Open Journal?

[Syksy Rasanen]

Any objections? Would people submit to and support an Open Journal?

The idea is excellent; it's true that journals are already redundant as a way of distributing research results (in high energy/astrophysics, at least).

One big issue is the attitude of various funding agencies towards "quality-stamped" arXiv publications vs. publications in well-known traditional journals. After all, one of the main reasons of publishing in journals today is to demonstrate the quality of the work to outsiders not familiar with the field. If an arXiv related quality stamp system is not seen as prestigious as traditional publication, there will be a strong incentive to publish in traditional journals instead (regardless of whether their quality control is in fact better or not).

[Anze Slosar]

Any objections? Would people submit to and support an Open Journal?

I agree, that it is an excelled idea, in principle. However, many european countries still rely on "paper count", i.e. to advance your position you need x number of points where each publication gives you y points divided by number of authors, each citation gives you z points, etc, etc. Numbers y and z are calibrated off Web of Science or something like that.... So unless the institutional b*ds recognise you it is not going to work. But they don't have any interest in doing so as it brings them money... So yes, if I were a bigger cheese I would publish in an open journal exclusivelly... Until I have a job, though...

I would be quite happy to lend some of my coding time / resources to help it start up, though, if you decide to do it....

[Andrew Jaffe]

In an Open Journal editorial services and refereeing would be supplied free by people instead of payment for submission.

It's probably obvious but I think the main obstacle is indeed the need for editors: In addition to the scientific editors (who perhaps should be paid for their time and effort even in an Open Journal system), journals also need at least some other editorial staff, to do the administrative work (keeping the database of submissions, etc.) and the not-quite scientific aspects of editing (checking formats, proofreading, etc). Some of these would remain necessary -- and these people would definitely need to be paid, even if the scientific editors were willing to do their work pro bono.

But if PPARC in the UK is willing to fund our IDL licenses, they and similar funding agencies elsewhere could perhaps be convinced to pay these relatively small amounts.

[Sarah Bridle]

Replying to Andrew's comment, how about referees rate the paper:
(1) Needs negligible non-scientific editing
(2) Needs some non-scientific editing before acceptance
(3) Needs substantial non-scientific editing before acceptance
(In fact there is often something like this on the referee report anyway).
If (1) then this could be done by the small number of people running the Open Journal e.g. paid for by grants from funding agencies.
If (2) or (3) the authors could decide whether to do the editing themselves, or whether to pay for it to be done for them. A company (or several competing companies) could be set up offering these services.

I'd also actually quite like the refereeing system to not just say accepted or rejected, but to rate the paper, e.g.
* seminal (e.g. require >=2 out of 3 indep referees to agree on this)
* very important
* important
* useful
* an accepted contribution
* not useful
* probably incorrect
In fact we have to do this when refereeing for many journals anyway, so its just a matter of making this (as well as the referees reports!) public.
This would give Open Journal more potential to be of use at all levels, rather than just being another journal in the hierarchy of journals.

So how do we get this quality stamp to mean something? Could we get a load of big cheeses to be on an executive board? What determines the status of a journal? e.g. JCAP seems to be well respected, despite being quite new. How did it manage that?

[Vesa Muhonen]

In the Open Journal era the only role of libraries would then be to mirror/backup the global publically available database of papers. If an Open Journal closes, the papers will remain on the Arxiv with the appropriate journal stamp. There is no risk of papers disappearing if your library has a backup of the entire database.

This might be a bit of off-topic, but do you see this as a potential problem? Nowadays it really isn't a problem to store large quantities of data, but what about the contents of that data. Printed works do have the advantage that the storage media and the interface, i.e. ink, paper and the eyes, have been around for some time and this propably won't change in the near future. But what about ps/pdf/whatever? Can you expect that people N years from now can read the magnum opus you've just written and published only on-line?

[Chris Lintott]

Printed works do have the advantage that the storage media and the interface, i.e. ink, paper and the eyes, have been around for some time and this propably won't change in the near future. But what about ps/pdf/whatever? Can you expect that people N years from now can read the magnum opus you've just written and published only on-line?

I agree - I think a serious open journal project would need to bypass this objection somehow. Partly this is because I do think it's important that all the material is available long term but also because otherwise there is a simple argument which could be used by the established journals (which could claim to be 'journals of record'). The referee's comments (which if I understand correctly would be public) would be of enormous interest, for starters.

As Vesa says, the easiest way to acheive long-term storage is to produce paper copies - properly printed (text from an office laser would fade over a few years). However, this is cheaper than continually renewing a digital archive (transfer to new disks every ~5 years or so) and so, reluctantly, I think we need to add the distribution of 'some' printed copies to the list of costs that can't be avoided. However, the number needed is likely to be in the tens rather than the hundreds and individual instititutions would not need to buy the subscription for their staff.

I wonder how the virtual observatory people are planning to maintain their data archives? I wonder if this publishing idea is something they'd be interested in (thinking of groups that might have small amounts of money available)?

[Sarah Bridle]

JCAP isn't open access, but it is entirely electronic. So I think these must be separate issues, and the electronic versus paper issue seems already to be solved somehow?

Good idea Chris to think about whether money/expertise could come via the virtual observatories pot.

[Antony Lewis]

Quality-stamping vs traditional journal

I don't see this is a problem. From the Open Journal web site it could look like any other journal - just when you click on [download] link it takes you to Arxiv. Each journal could apply for an ISSN number, and apply to be added to Web of Science after 3 months of publication.

I agree one would need to start with some well-known names to get the ball rolling, so that no-one can argue the quality is substandard. Ideally a bunch of well-known professors to act as the advisory board for each new journal. Starting with astrophysics/cosmology has the advantage that some of the best known names in science are in these fields. Any new journal faces the problem that it is unknown and not immediately indexed in the research assessment databases, but as long as the quality is high I don't see the new journal can be ignored.

Printed works do have the advantage that the storage media and the interface, i.e. ink, paper and the eyes, have been around for some time and this propably won't change in the near future. But what about ps/pdf/whatever? Can you expect that people N years from now can read the magnum opus you've just written and published only on-line?

This is precisely why Arxiv only accepts pure PS/PDF as a last resort. Almost all submissions are in LaTeX (and this could be made a requirement for physics Open Journals). This basically represents the source code for the PS/PDF, and could be converted into different new forms over the coming decades as required. I don't see why the LaTeX cannot form a safe information storage medium, given it is simple, human-readable and computer processable. Referee reports could be plain text or LaTeX.

As Vesa says, the easiest way to acheive long-term storage is to produce paper copies - properly printed (text from an office laser would fade over a few years). However, this is cheaper than continually renewing a digital archive (transfer to new disks every ~5 years or so) and so, reluctantly, I think we need to add the distribution of 'some' printed copies to the list of costs that can't be avoided. However, the number needed is likely to be in the tens rather than the hundreds and individual instititutions would not need to buy the subscription for their staff.

Jounals like JCAP do not have this, so I see no need for printed copies. Libraries can pay to do their own printing if they are paranoic, but I don't think it's something that needs to be part of the academic refereeing system.

The size of the Arxiv database must be trivial compared to any large scientific data repositry.

It's probably obvious but I think the main obstacle is indeed the need for editors: In addition to the scientific editors (who perhaps should be paid for their time and effort even in an Open Journal system), journals also need at least some other editorial staff, to do the administrative work (keeping the database of submissions, etc.) and the not-quite scientific aspects of editing (checking formats, proofreading, etc). Some of these would remain necessary – and these people would definitely need to be paid, even if the scientific editors were willing to do their work pro bono.

Tasks like keeping the database of submissions can be completely automated. I was thinking that the responsibility for proof-reading and non-technical editing would be borne by the submitters. This is essentially the case in practice anyway, since we all read papers from the arxiv, not journals. The quality is usually quite high, and as Sarah says it could be made a condition of acceptance that some third-party proofreader is used if it is substandard. The idea that formatting etc is standardized is a prejudice from traditional publishing - in principle there's no reason why science-quality stamping has to have anything to do with layout on the page as long as it is clearly understandable. Open Journals could if they wish apply style guidelines, but I see this as optional.

Linking to commercial third-party proofreaders/translators might be one way for Open Journal to self-fund if required.

I'd also actually quite like the refereeing system to not just say accepted or rejected, but to rate the paper,

Each journal could optionally be split into Letters, A, B representing letters, high-quality, high-interest stuff, and B- the rest. The advantage of having the assignment made by a scoring system is that it would remove the bias in the current system where prestige depends in part on author submission: pushy people submit to more prestigious journals and hence get apparently higher research rankings, regardless of whether their research is actually any better than more self-critical people.

Incidentally, typical CMB experiments like CBI, WMAP etc. typically spend on the order or $20,000 to publish their results in ApJ (which, incredibly, charges for submissions and is also not open access). By eliminating all this cost, there should be a lot of spare money sloshing around somewhere...