OK, I've watched this now. My notes on what was discussed follow (should be quicker to read than watching over an hour and a half!). There are some direct quotes, which I've referenced with approximate times in the video. I hope this is useful for people interested in scholarly publishing, particularly those that aren't familiar with the norms in mathematics.
MU Panel 2. Future of Publishing
Date & Time : 18:00
, August 19 (Tue), 2014
Moderator: Jean-Pierre Bourguignon, European Research Council, Belgium
Rajendra Bhatia, Indian Statistical Institute, New Delhi, India and Sungkyunkwan University, Suwon, Korea
Jean-Pierre Demailly, Institut Fourier, France
Chris Greenwell, Elsevier, The Netherlands
Thomas Hintermann, European Mathematical Society Publishing House, Switzerland
Nalini Joshi, University of Sydney, Australia
Ravi Vakil, Stanford University, USA
DR: My notes follow
Demailly was there to discuss Episciences http://episciences.org/
Chris Greenwell, the Elsevier representative, said he expected to get "shot at"; but really was just to listen to what people had to say, We'll see how it goes.
Bourgignon (president of the ERC) says the rigidity of the definition of OA makes it hard to work with.
3 rounds of discussion. First on forms of publishing. Second on economics of publishing. Third on involvement of mathematical community/relevant parties when faced with various models. How can we collectively get organised?
Bourgignon: other fields have completely different conception of what 'publishing' even means. Biologists, for example, care about getting results out tomorrow, who cares about availability in six months time; mathematicians are completely the opposite - they want careful vetting and long-term availability.
): helping put together a "best practices" document endorsed by the IMU that individual authors can use when dealing with publishers. Major input by/originated with Henry Cohn.
Vakil: proposal is this - publishers make all articles five years and older permanently public and distributable (DR: this last point important IMHO!!
Vakil: This should not impact publishers' income, and will address many (but not all) of the needs of mathematicians. Not a perfect solution but a start.
Vakil: Elsevier and the American Mathematical Society already do this (DR: NB Elsevier have it at four years - this is mentioned later
Demailly: books in paper form still have some time to live; e-books not yet up to scratch. Articles different. Major problem is indexation of explosion of information.
Demailly: most people expect that just clicking on references will get that paper. Open access is major issue. In favour of Green OA, but the future will bring models where the scientific community can navigate by efficient tools cf the arXiv - we need to build on that.
Demailly: Episciences to be a layer on top of that to help. Numdam has many papers that were not originally in electronic form. But full-text search missing, and this should be the future across all papers. Episciences should help with this.
Joshi (DR: earlier she said she was probably on the panel due to being head of the Australian Math. Soc., and their exercise, prompted by govt. bureaucrats, of ranking journals - this has had a longer life than one might expect, as it was officially killed after one (!) round of research impact assessment
): A new thing - mathematics progressing through collaborative enterprises. Don't know how it will link up with publishing as we now know it (MathOverflow and PolyMath get a mention). Maths is being shared on Twitter, FB, G+ even sometimes LinkedIn. Many more communication mechanisms than imagined possible. But these are collaborative channels, not publishing as currently conceived. But arXiv was originally this too.
Joshi: one might view these mechanisms in the long term as also being publishing.
Greenwell (DR: he's in charge of Mathematical Sciences division): maths is entirely different community than what I've worked with before (Engineering, for example). Longevity of data is far beyond those fields. Mathematicians use information and communicate differently.
Greenwell: Small community-run journals like EJC don't exist in other fields. Elsevier thinks future will be a diverse one. Status quo will not exist forever. Internet has lead to e.g. "open access journals that people would rather not see in existence" (~21:18
- DR: strange. I wonder who 'people' is?
Greenwell: what people want to see from "main journals" is vaildation, archiving, "being there forever". Elsevier sees collaboration as part of publishing's future ("we bought Mendeley"). Reference to E.'s four-year moving wall for open archives.
Bhatia: Focus on books. Undergrad books ok - if you write a rubbish calculus text, noone will buy it. Graduate textbooks ok. Research monographs share the same problems as journals - books that are cheap but shoddy, oversupply; likewise too many journals/articles/conferences.
Bhatia: perhaps we wouldn't write so many papers if our names were not on papers and didn't need to get respect of colleagues. Demands on us from the profession to publish. Perhaps it is good to keep us 'fit' - not all contribute at the highest level, but keeps people engaged. There's an effect like in sport: few at the top, more at the next tier, many more the next level down. The lower levels makes possible the highest level.
Bhatia: But if you have ten thousand people hitting the ball at random it doesn't really contribute to tennis. Likewise in publishing. A major danger in publishing - how do we decide a baseline for quality? Who does it? Journals supposed to certify correctness etc, but not always done. Journals sometimes publish stuff for which we would fail undergraduates.
Bhatia: (has examples willing to share). note also the recent fooling of the system by SciGen. (has examples) This is a major issue that needs addressing.
Open to the floor now. (DR: I couldn't catch the name of the first asker/commenter
). Functions in publishing are splitting and might be done completely separately (disseminated through arXiv, but not validated). Perhaps some company gets money for purely validating papers?
Response by Demailly: Join accessing papers (which is mostly through arXiv) and certification (currently through journals) in one platform. Creation of many 'epijournals' which are purely certifying bodies using the arXiv/mirrors for distribution. Invites mathematicians to create editorial boards. Functionality is easy, good, free to use and for authors (there are costs, to be discussed later). Vision is to know that when looking at an arXiv paper, see that it has been verified and this is the final version of the paper.
(Q/commenter didn't identify): Mathematical Sciences Publising (MSP) tried an episciences-style approach for G&T and AG&T etc, but stopped it.
Demailly: there are many small open access journals now, many of which are of low quality. We want to open up high-profile journals. The financial problems haven't helped, but the project (episciences) is supported by major bodies and will be so for a long time, depends on infrastructure that has been in place for ~15 years. There should be no financial issues in the short term, only if the initiative becomes extremely big, e.g. the whole world uses it. Institutions would be called to support it, but this is long-term thinking.
(commenter from Spain, on Vakil's remarks). How does the AMS support their moving wall of articles? Other parts of business? Someone from LMS wrote an article in the Notices explaining that they do the opposite: free for some time, and then closed off except to subscribers. We need to be sustainable, even if we want open.
Vakil: LMS realises how mathematicians use papers. But if there were a moving wall then opening up, libraries would still subscribe.
Greenwell: Elsevier did sums and found a four-year window was sustainable under current model.
Unidentified questioner (publisher from American Mathematical Society). Didn't notice any major financial effect on journals from moving wall and open archives. Libraries cannot wait 5 years until the article becomes free. With other models (i.e. not subscription) moving wall perhaps not sustainable.
Bourgignon: Second topic now - economic models and how costs are evaluated and covered, and how proposed models are supposed to work/be sustainable.
Bhatia: mentality in publishing that mirrors that in modern economics: an assumption that what is good for corporations is good for all of us. Mergers, acquisitions, formation of conglomerates, monopolies etc. BUT (comments all regarding publishing) #1 there seems to be no correlation between quality and price (DR:_ Amen to that_ ). in 1980, three kinds of journals: society journals (cheap), reasonably priced journals (e.g. Academic Press, Springer), unreasonably priced journals (Elsevier, Kluwer, etc). The quality of these latter was the lowest, but the price was the highest? How did this work?
Bhatia: libraries like to have everything, anyway, so it supported these latter journals. #2 - technology has brought prices of production down, but journal prices haven't reflected this, cf prices of mobile phones/clothes. #3 much labour in this industry is free - referees etc #4 companies seem to have been completely insensitive to the demands of the consumers, free workers. e.g. many editorial boards who resigned did so after prolonged efforts in negotiating with publishers.
Bhatia: e.g. 2003, the journal Algorithms, originally with AP then with Elsevier. Knuth tried to talk to E. for 2 or 3 years without success. Acquisitions do not seem to have any impact on quality (E's quality seemed to go up a little after buying AP, Springer acquiring everything, and quality perhaps going down a little). Clearly publishers would like to acquire big fancy journals e.g. Acta Mathematica, but why the little journals (DR: a name mentioned, I didn't catch it
)? Is it to make bigger bundles? Quite different an industry to the rest of the world.
Bhatia: now we have two major players and a number of smaller ones. Historically monopolies have been bad and we seem to be moving towards that - we are forced to buy a package. Pricing is opaque. In 1990 every journal would give list price. In India, for example only 5 copies of the journal are being bought: they asked how about a discount, then more copies would be sold. Tried a consortium so as to get Math Reviews. Now no one knows what is happening.
Bhatia: Some say the opaque pricing is good. eg U. Montana being charged by Elsevier 40% of what E was charging U Michigan. Would be happier if publishers were more convincing. Elsevier sued someone because this was revealed. The cost to society per article is roughly $5000 (DR: Estimate due to +Mike Taylor
, I believe). The cost of putting an article on the arXiv is $10 (Plz explain?). Annals of Mathematics
costs 13c/page, Acta Mathematica
costs 65c/page (before Springer bought it). Inventiones Mathematicae
$1.21/page (DR: for non-mathematicians, these are roughly the top three journals in maths
). Elsevier average is $1.30/page. So no correlation between quality and pricing.
Bhatia: OA APC charges: Springer $2000/paper, usually $1500/paper, but varies more widely, University presses - are they better? CUP announced the two Forum of Mathematics journals - $750/paper (DR: or GBP500). One issue of Mathematika from CUP (acquired from someone else) costs $2000. Is Elsevier really that much worse than university presses in this regard (DR: I take that as meaning, the university presses are no better than E in this regard
Bhatia: Mention of profit margins of Elsevier, Springer and Wiley (36%, 33% and 42% resp. - last years' figures it seems). Why is Elsevier being singled out?
Greenwell: a lot of numbers being thrown around, some of which probably came from the Bergstrom article in PNAS - recommend people read that article; a fairly well-thought out piece. Shows E is one of the cheaper commercial publishers. For E it comes down to what is sustainable - will be key in business model. Most of cost is in hosting and archiving. ScienceDirect has 12.6m articles (this is across all of E), adding 350k/year.
Greenwell: we add about 2k maths articles/month, or roughly 10k/year. Open Archive is sustainable. Looked at launching a Gold OA maths journal (seems to be de rigeur these days) but wasn't particularly sustainable, due to lack of funding/interest in community. Costs for hybrid taken up by funding bodies - about 50-60 articles/year. $750-2200 APC. Currently works on journal side of business. List prices are posted online, annual percentage change is also listed. Last plug for Bergstrom article.
Joshi: Have you, the audience, submitted a paper to an open access journal? (Bourgignon: you mean Green OA? Joshi: Yes - DR: this is not the right terminology, I'm not sure what was really meant
) 14 out of ~60 people. Does the cost matter? Would a lower price make it more attractive? How many would submit given a $750 APC? None. Any price? Is it price that stops?
From audience (sounding emphatic): how can I advance my career? By publishing in top ranking journals. What is a top-ranking journal? One with a top-of-market price. In OA who will refereee etc? But we referee the papers, so it's complicated. (DR: I didn't quite get the rest of this comment, something to do with pricing
Hintermann: talking of sustainable models, EMS still selling journals. Don't have a complete overview of all the models. From my perspective I haven't seen a convincing viable alternative model. Gold and hybrid have serious flaws and issues that have not been satisfactorily addressed. No pressure for these from the community. There has not been even a single serious request as to whether our journals will use an author-pays model. So see no reason to engage in Gold OA. Will only do so if it pushed by community. Not promoting in general 'free' access, but rather 'affordable' access. Don't see anything fundamentally wrong with subscriptions, just that pricing has gone haywire and is out of hand. Any system would be attacked by those who can.
Hintermann: Scholarly publishing is not operating in a consumer market - no substitutes for products - supplying scientific information to the community. It shouldn't be in the hands of someone with a commercial interest. Almost all problems come from not using not-for-profit publishers. If it is not-for-profit - society publishing - everything you pay goes back to community, e.g. used to support publications not able to support their own costs, and support the society behind the journals. Publishing should be in the hands of societies, not in the hands of aggressive commercial publishers. This is how I see sustainable models. On the statement that labour in publishing is essentially free - this is false (things the publisher does is not free). People usually think of the price of a particular product. But even as a not-for-profit need to build up a substantial organising before even producing a product. Every product needs to not only support its own production, but its share of these other costs. These should be minimal/only as high as they absolutely need to be.
Demailly: In my university, the cost of buying journals on the order of 100k euros. Multiply by all the universities - lots. Given that refereeing is (mostly) free by mathematicians, the cost of electronic communications is so little, this is really a misuse of funds in the unique purpose in how we evaluate research. The main question is pricing this evaluation and exploiting the evaluation for the sake of one's own career.
Demailly: these can easily be addressed. Have OA journals that have editorial boards that are experts in their fields. Eg. exists a current episcience board with four Fields medals, president of IMU, members of CEIC (Committee on Electronic Information and Communication). Want the first journal to be very high profile. OA doesn't mean low-quality - just need right experts in right place.
Demailly: cost now is in $100m's dollars. Imagine episciences is a huge success and takes a large % of world's papers. There is a team of 12-15 people, incl. 8 computer experts. Need secretaries, computers, bandwidth. This already exist. But there are huge costs because the system is not uniform. Episciences can provide a uniform platform - the system will create the journal website etc - don't need to do all that work from scratch in each situation. In the long term there will an impetus in the long term to save money (cf developing counties that cannot afford journals).
Demailly: why will this work? Take an optimistic view: We already have the archival through the arXiv - the other necessary functions (secretarial work, IT support) will turn out to be cheaper than it is currently.
Vakil: The community has complained a lot, but not done a lot. Now things are being done. How do we encourage people to get on involved? Talk of evaluating articles by social media, comments etc, but no one's actually done it for real. How do we find some new way to evaluate research, beyond having stable established journals? How can we drive a stake through the heart of predatory journals?
Vakil: If a major cost to Elsevier is archival, could some agreement be reached with the arXiv so that old articles could be hosted there? The cost could be offloaded from E to the community.
(missed name, publisher from USA) Professional societies - two models. Saw panel before ICM about OA publishing in maths. MAA said main income flow is publications. Most don't want societies to suffer - we benefit from them. Business models might still be changed. Perhaps a lot ("leaner"). Compare the music industry. No one purchases a whole journal issue for their own consumption, but a single paper. Why not 99c/paper rather than $40/paper? Second example - an institutional library. Librarians are hurting. Libraries some of the first places that budgets are cut, with continuous cutting - is not sustainable.
Libraries are getting excited about being a publisher and archiver of research. Wages are already paid, as librarians, and they can work with scholars as editors etc. Working with Journal of Humanistic Mathematics, and OA works for it, but perhaps not all.
(another comment) There are several costs that haven't been discussed and should be taken into account: e.g. in France in the big consortium (DR: for negotiating subscriptions, I assume
), there are 60 people paid at full time at a high level - a huge cost. We should seriously estimate the time we all spend on refereeing and editor "for free" and see how much we cost given our hourly rate - and make this widely known. Another hidden cost: they pay some of the editors, and this is not known even among other editors on the same board. This should be transparent who does what and who is paid what and why. Lastly, the secret contracts negotiated for journals should be open - so we know the overall business model in which the scholarly communication system operates. Be open on both sides of the system. Are you willing to do this? (Directed at Elsevier).
Greenwell: We've seen this in +Timothy Gowers
investigation (cf http://gowers.wordpress.com/2014/04/24/elsevier-journals-some-facts/
), Bergstrom's articles. This is likely to happen more and more, people wanting to know what things cost, what consortia pay. "The original argument about keeping those contracts secret was that certain universities managed to negotiate better deals than other universities and didn't want to advertise that fact" (~1:13:00
) Now the numbers out in the end. Those universities that have been paying more may be able to negotiation harder. E does pay its editors-in-chief (=academics) and sometimes associate editors - doesn't go all the way to reimburse them for the time they spend. Q from floor: where are these figures published? A: "We don't generally make that available, mostly because the individual editors probably don't want their colleagues to know" (~1:14:30
) Q: this is unfair A: depends on editors. There's nothing in the contract stopping them from telling people. Most of them probably wouldn't want to tell you. Averages out at about $100 per paper handled. A reasonably big number, but varies quite widely between editors.
Bourgignon: important point, shows that not all the costs are on the table, it will be difficult to have a consistent description of the cost so as to compare new models.
Last section, what can we as a community do about this? Publishing for mathematicians can't be detached from other fields? Policymakers don't make such fine distinctions. We may be caught in a difficult situation if caught up in changes driven by others. eg. Any project funded by EU must be OA, and the definition of OA is important. What delay is allowed if any? For biologists, the delay should be zero. Publishers say 'it must be Gold'. But for mathematicians we can get access to some version from the arXiv and can work with that, or from the author. We need to be careful as to the position we defend. How can we make sure our needs in publishing mathematics is taken into account in across-the-board discussions?
Vakil: There are many things we would like from publishers. Many of the things we already can get/do, but we have a hard time following through on it (some things from CEIC's perpective). 1) Author rights - even if not explicit in the contract, most of the serious publishers will give the things you want if you ask. If people knew what they were and knew they could ask this would be trivial. People don't do this so often. If there were just some way to get the community to think of this as a matter of course.
Vakil: 2) it is easy to find out if the journal is good or bad (Sherpa/Romeo), best practices that journals can follow - trivial, it's just one click of the mouse away. How do we get our colleagues to do this?
Demailly: clearly a lot of political issues around this. 1) Q of OA is innately related to evaluation of research and the use of IF. Happy that IMU will take a stand against bibliometrics in evaluating researchers. In some countries where there are not enough experts, administrators may find it easier to count pages of articles, add IF of journals. IF published by commercial companies, not freely available etc. IF do not make sense. Mathematicians should explain this to other disciplines.
Demailly: 2) need to get support from national governments to support OA journals. In initiatives like episciences become very big, will need support from institutions. If it works, it will become indispensable, so institutions etc will be happy to supply money. "The question is about having a serious evaluation of research, and not having it in the hands of administrators and people counting numbers" (~1:23:15
Bourgignon: small correction regarding European govt. support for OA. British govt was explicitly for Gold OA, Dutch similar. Need to be careful when "open access" is used as shorthand for "Gold OA".
Hintermann: this (DR: the panel?
) is not about political issues, but how mathematicians can get their voice heard. The societies play an important role in this. They have committees that are explicitly in contact with political bodies that are relevant - through these mathematicians can have influence on political decisions.
Joshi: Discussion is on several things at once. How validated original research is made available to the world. But evaluating journals - we haven't talked about predatory journals/those that spam inboxes. Signed up for thecostofknowledge.com
. Lot of movement is based on seeing mathematicians as victims: of publishing companies, of other people evaluating us via bibliometrics, of being funded less than other areas therefore can't afford certain things.
Joshi: Turn this around. How do we change to what we really want for all these things? Good that (some) publishers have a moving wall, but we can go beyond it. Wy doesn't the IMU create a publishing charter to which community and national societies can agree on? We should be proactive in this area. Community-driven not top-down. Not-for-profits/societies in at ground level.
Vakil: IMU does have a best-practices document, but national societies don't necessarily follow it.
Greenwell: mathematicians have done a good job at making themselves heard at E - pity it had to come to a boycott/editorial board resignations. You have got our attention now. A lot of people are quite frustrated how they are evaluated - you need to convince people to do it in a different way. One of E's journals was closed because it wasn't good quality based on community feedback. Mathematics council - mechanism for knowing what community is thinking. E will do the best to help mathematics community to do what it decides to do.
Bhatia: before 1950 most of publishing was done by universities plus a few altruistic publishers. In 1960s publishing took off due to expansion of research - why didn't universities/societies also jump on board? Closed mindedness? Areas neglected by societies and publishers jumped in? Bias by society/university publishers for certain areas.
Bhatia: A few yrs ago Indian acad. sciences got an agreement with a big publisher, who is not allowed to be named. Editorial control will stay with the acad.; (C) will stay with acad.; publications will be available from the acad. website for free. It seems to have worked for 5 years. Academy got about 1/4 of its income. Now publisher wants to change. Other journals are the same. Acta
and Pub. IHES
are distributed by commercial publishers, but the reputation has been built up by academic editors - how can we keep things favourable for us?
(missed name) Aside from people wanting to make money through OA, there's the issue of rapid publication vs months and months of waiting for a referee report. If we could, as part of the best-practice descriptions, ask that journals do things in a reasonable time? Would perhaps give people something for the money?
Vakil: it's not the journal, it's us, taking time to referee!
(Final question) (Chuck L... New Jersey) 1) Any journal that can guarantee publication in under 3 months - I don't trust the quality of the referee. 2) We need journals to be quicker in rejecting papers that should be rejected and spend time evaluating the papers that will be accepted, so that the former can move on quickly.
Bourgignon: As an EiC for some time, there are papers that are rejected quickly, it's the papers that look interesting and are carefully refereed, but someone points out a weakness and it can't be accepted. It's embarrassing that people have been kept waiting.
------------DR: That's the end of the panel. Lots to digest there #openaccess #publishing