The question of costs returns to the fore, arising from a proposal for a specific project. The question is taken up of what and whether editors should be paid. A university press journals manager contributes some current, real-world information to the discourse on editors and editorial offices. One of the undoubted inefficiencies of the present journal system is the delay and redundancy introduced by a distributed and publication-linked practice of peer-review, resubmission, and limited acceptances.
Date: Sun, 24 Jul 94 15:26:49 EDT
From: "Stevan Harnad"
Dear Andrew,
Below is a discussion of an important side-issue that has arisen in discussions regarding a confidential proposal concerning electronic publication. The side issue is: In what do the true residual costs of electronic-only periodical publishing actually consist?
I am circulating this to a wider group. At the request of the author of the proposal, the proposal itself is not circulated, and I have deleted below anything that refers to its content. Nothing hinges on what is removed, however. The present discussion concerns what the real functions and real costs of the editorial office of a peer-reviewed journal are.
I will try to itemize quite explicitly what comprises that residual "<25% of paper per-page costs" for quality control that will continue to need to be covered in electronic-only periodical publication. Apart from spelling this out explicitly, I also note in passing what might be some cross-disciplinary differences (especially between highly technical and symbolic texts, as in mathematics, and more prose-intensive disciplines -- the latter constituting the vast majority of the esoteric scholarly/scientific periodical corpus). In certain important respects, my own discipline of "cognitive science" (a mix of experimental psychology, theoretical psychology, brain science, biology, computer science, linguistics and philosophy) is perhaps better positioned than mathematics to provide a representative model that would apply to most of the rest of learned publication (though there may well be other views on this).
Andrew Odlyzko wrote:
ao> From: amo@research.att.com ao> Date: Fri, 22 Jul 94 22:39 EDT > ao> Stevan, > ao> A few remarks on your comments on [the anonymous] proposal. ao> I agree with you fully that the full [text of any published article] ao> has to be certified, and that this certification has to be performed ao> by the scholars who are editors and referees. I assumed that this is ao> also what [the author] had in mind. > ao> I was a little confused by your discussion of what scholarly ao> publishing ought to cost. Aside from the scholar's time ao> in doing the research and writing a paper, we have ao> the following stages in publishing it: > ao> (a) Typing or typesetting the manuscript. This essentially ao> always takes place at the author's institution, and ao> is increasingly being done by the author, since ao> technology has made that alternative attractive. > ao> (b) Peer review. This is done by scholars who are editors ao> and referees, and who are almost never paid. Secretarial ao> assistance is usually provided by these scholars' ao> institutions, and sometimes is reimbursed or provided ao> by the publishers. > ao> (c) Typesetting, copy editing, printing, distribution, etc., ao> by the publishers after the peer review and author revisions ao> are completed. > ao> It seems safe to assume that the costs of (a), which I estimated ao> at $ 200-400 per paper, will continue to be shouldered by the ao> authors' institutions in those increasingly rare cases that ao> the scholars do not typeset their own paper.
Andrew,
I agree about (a) and its costs. But I have to point out that in over 15 years of editing Behavioral and Brain Sciences and 5 years of editing PSYCOLOQUY, I have never once encountered a paper where the author's final draft could be published verbatim! In any case, this is not the real issue, as you will shortly see; the real issue is the cost of generating a publishable peer-reviewed text, and that consists (relatively seamlessly) of all of (b) plus copy editing (i.e., one component of (c)).
ao> In discussing economics of future scholarly journals it seems ao> worthwhile considering (b) and (c) separately. Here is where ao> I do not fully understand what you advocate. Perhaps it is ao> because our fields have different practices and different ao> expectations. In one place in your message you say > sh> But what about the costs (and responsibility) of implementing peer sh> review for the "free texts"? THOSE costs, plus some subsequent editing sh> and copy-editing, are the only TRUE costs in electronic-publication... sh> I estimate those true, essential costs of purely electronic quality control sh> at (well) below 25% of per-page paper costs (i.e., current journal page sh> costs). > ao> In that passage you seem to imply that in the electronic world ao> both (b) and (c) should cost below 25% of the current figure. ao> Later on, though, you say > sh> Copy-editing (which is what is really at issue here) sh> is such a minor part of the function (and the cost) sh> of publishing that it hardly seems worth talking about. (If that were all sh> there was to it, Universities could easily hire a staff copy-editor sh> to vet all final texts, and that would be the end of it.) It's the REST sh> of the quality control (implementing peer review and substantive sh> editing) that's the real work, and it's not clear from this proposal sh> who is to see that that's done, who's to do it, and how its true expenses sh> (a per-page cost I estimate at under 25%, but not zero) are to be paid. > ao> Here you seem to be saying the 25% is to go for peer review and ao> "substantive editing." Perhaps what we need here is your definition ao> of "copy-editing" and "substantive editing." Also, what do you mean ao> by the costs of peer review? Your answers might clarify ao> what you really have in mind in the passages above. In the meantime, ao> I'll explain how I see the situation. > ao> Both of us agree that (b) is indispensable. The only part of (b) that ao> I expect will continue to cost money is the secretarial assistance, ao> which I estimate in my essay should cost a maximum of $ 100-200 per ao> paper. In mathematics, computer science, electrical engineering, ao> and the few other fields that I know, the only editing that is provided ao> at this stage is what the referees and editors do gratis. Sometimes ao> this editing is extensive, and might be called "substantive editing" ao> by any reasonable definition of the term. I have had a few referees ao> completely rewrite some particularly interesting papers by Chinese ao> or Russian writers whose command of English was practically non-existent. ao> We also have the example of Walter Gautschi, which I cite in my essay, ao> who does extensive editing of manuscripts in his unpaid job as editor. ao> (Most of his work is copy editing, but some I would classify as ao> substantive.) In the overwhelming majority of cases, though, the editing ao> at this stage is trivial, such as referees pointing out the most egregious ao> mistakes. I expect this situation to continue, at least in my field. > ao> The editing in stage (c) that I am used to can be classified as ao> copy-editing. This involves correcting typographical mistakes, ao> formatting the paper, providing running heads, page numbers, ao> making sure references follow the journal's standards and are ao> actually invoked in the text, etc. I would not call any of this ao> "substantive editing." Furthermore, neither I nor any of my ao> colleagues that I have ever discussed this with would want anything ao> more than what is provided. The risk of getting the mathematical ao> substance of the paper damaged is just too great. There are all ao> too many horror stories of newly hired employees at publishers ao> trying to "improve" the presentation in a math paper only to ao> mangle it hopelessly. > ao> What do I see in the future? Well, (b) will be carried out as ao> before, primarily by unpaid scholars, possibly with some minor ao> assistance from secretaries. This will cost $100-200 per paper. > ao> Stage (c) now costs $ 4,000 per paper. This goes for printing, ao> distribution, and copy editing, but not (at least in the case ao> of mathematics) for "substantive editing." When we move to ao> electronic publishing, the only thing in stage (c) that ao> I feel will be worth preserving will be copy editing. It is ao> not all that minor a part of the publishing process, as it ao> seems to account for the bulk of the present cost. The reason ao> it is so costly is that it involves several layers of specialists. ao> It used to be that publishing involved the extremely expensive ao> steps of typesetting and printing, and it was not possible to ao> lower their costs below a certain level. Thus there was ao> an absolute floor under the cost of stage (c). In the future, ao> when (c) consists basically just of copy editing, I expect we ao> will be able to operate it at any price we choose, and my guess is ao> that an expenditure of $ 200-600 per paper will provide adequate ao> quality. Whether this function will be done at publishers or ao> the authors' institutions, I am not sure.
Let us say we agree, roughly speaking, about copy-editing, the only expense in (c) worth mentioning, once we move to electronic-only publishing. But my estimate of < 25% of paper per-page costs in electronic only periodicals was definitely NOT based only on the cost of (c); (b) costs money too, and is far more important than copy editing.
First let me itemize what (b) entails (and I will argue that the copy-editing component of (c) is probably best assimilated seamlessly with (b), but not much hangs on that):
A journal has an Editor. Editing takes time -- time that would otherwise be devoted to research, teaching and publishing. Refereeing takes time too, but the difference is that refereeing is done on a voluntary, as-time-is-available basis, whereas someone who accepts the commitment to edit a journal (or to subedit a section of a journal) must give requisite time to process the entire manuscript flow. What does that time consist of?
(1) Submitted manuscripts must be processed; this is done by an editorial administrator and secretaries who report to the Editor (and must be paid by someone).
(2) The Editor (or Subeditor) must read or at least skim all submissions and select referees (sometimes with the help of an Editorial Board, sometimes even with formal weekly real-time meetings, for journals with high submission rates and large annual page-counts).
(3) The editorial administrator and secretaries must then see to it that the referees are invited, receive the manuscripts, submit the reports in time, get followed up, get replaced if delinquent, etc.
(4) For each manuscript, once the reports are in, the Editor (or Subeditor, or Board) must read or skim the manuscript (conscientiousness varies -- as does the rigor of the peer-review provided by a given journal) as well as the referee reports and prepare a disposition letter, indicating whether the manuscript is rejected, accepted (rare without revision, at least in my fields), conditionally accepted contingent on minor revision, or requires major revision and re-refereeing (if the latter, go back to (1) and start again when the revised draft is submitted).
(5) A conscientious Editor who may only have skimmed manuscripts until they reach the possibly acceptable stage, will become more actively involved in the manuscripts that are likely to be published, not only in making the substantive judgements about which referee recommendations need to be followed, and when they have been successfully met, but in the finalizing of the manuscript itself. This is what I call substantive editing (it is not the checking of format and references) and it is an essential part of the peer review process -- indeed, without it, the Editor is not really Editing but simply doing box scores on referee reports (and the quality of the journal will reflect this).
(6) Finally, when the Editor judges it is ready, the paper is accepted, with the prior and subsequent negotiations between Editor and author, then copy editor and author, and finally the proofing of the final text mediated by the editorial administrator and secretaries, all reporting to the Editor.
(1) - (6) is, in broad strokes, the relatively seamless stream that leads to a peer-reviewed publication. It takes time (the time of the Editor, editorial administrator, editorial office secretaries, and copy-editors; hence it also costs money. (Note that I have NOT reckoned in the time contributed by the referees, which is a voluntary service we all perform when we have time, and perform gratis; these are editorial, and hence publishing costs only).
There are many different ways that journal editorial offices are structured. One common model (the one used, for example, by the American Psychological Association, which publishes most of the leading psychology journals) is to have an Editor appointed for from 4-6 years who receives an honorarium and/or some of his time is bought from his University, and the editorial office receives a budget to pay for the editorial help (editorial administrator and secretaries -- copy-editing may be administered by the editorial office or the publisher, depending on which is more efficient).
In brief, I think your misconstrual of the true functions and costs of generating peer-reviewed publications is based on assuming that Editors and editorial office staff can be thought of in the same way as referees, donating their services gratis whenever time is available: They cannot be, however, because editing a journal is a calendar based, unrelenting, obligatory workload (and time-consuming in direct proportion to a journal's manuscript flow and annual page count) rather than a voluntary, ad lib function such as refereeing. It is an ongoing commitment that takes time from other things scholars and scientists do, and takes it systematically, on a daily, weekly basis.
Perhaps in some disciplines the Editor's function can be decentralized and distributed -- but some centralized entity still has to keep up with the manuscript flow without developing arbitrary lags -- and I for one think peer review, for better or for worse, is best filtered through an Editor's unitary judgment rather than an anarchic system with only local answerability (I could be wrong on this); but either way, SOMEONE has to make the commitment to exercise editorial judgment, a key component in peer review (peer review is not simply referees, voting).
Disciplines like mathematics may require less copy editing or less substantive editing than others; fine. If they needed less in paper, they'll likewise need less on the Net. But many (most) disciplines do need substantive editing and copy editing, and I'll bet my < 25% figure adjusts for this across disciplines: The less prose-intensive disciplines probably already had lower copy-editing costs, so their 25% of paper costs will simply be a smaller absolute figure (unless other special costs counterbalance it) than the 25% for more prose-intensive disciplines. In any case, as you see, I've stressed copy-editorial functions less than editorial ones in all of these considerations.
So there you have it. In my view, it's (1) - (6) that underlie the true per-page costs of electronic-only publication, and I think the < 25% figure will derive mostly from (1) - (5).
Best wishes, Stevan
From: amo@research.att.com Date: Tue, 26 Jul 94 07:08 EDT
Various email discussions that involved Stevan Harnad and myself, as well as others, have uncovered an important question, namely, whether it is customary to pay scholars who work as editors. (Relevant payments might take the form of a stipend on top of the editor's regular salary or might be paid to the editor's university to lessen the editor's teaching duties.)
It appears that practices vary between fields. Harnad says that in his area, cognitive science (which he describes as "a mix of experimental psychology, theoretical psychology, brain science, biology, computer science, linguistics and philosophy"), such payments are common, and conjectures that his area is typical in its publishing practices. I have not seen any need to pay editors, because this is simply not done in the areas I know. I have served in the past, or am now serving, on the editorial boards of 18 different journals. These journals are published by several learned societies (AMS, IEEE, SIAM, etc.) as well as by some commercial publishers. Slightly over half are in mathematics (both pure and applied), and the others are in computer science, cryptology, electrical engineering, and one that is partially in physics. Not a single one of these jobs involved any financial compensation for me. The editors do work for free in these areas (*).
Are there any studies that address the question of how often editors are paid, and how much? Any information in this area would be helpful.
Payments to editors should be considered separately from paying for secretarial assistance. The latter is rather common, it seems. Information on how much it costs would be useful as well.
Answers to the questions posed here will be helpful in assessing the costs of future electronic journals. If it is customary in an area to provide substantial payments to editors, and this practice persists, this will alter calculations of how much scholarly journals will cost, and might restrict the choice among various models for future electronic journals.
Andrew Odlyzko amo@research.att.com
(*) I am aware that there are journals with paid editors. For example, Physical Review Letters, which I cite in my essay "Tragic Loss ...," has about four full-time senior physicists in charge of the peer-review process (which also involves unpaid volunteer editors). Other Physical Review publications have a mix of paid and unpaid editors in charge. Mathematical review journals, such as Mathematical Reviews, also have paid staffs of professional mathematicians. Such situations are easy to identify. The main question, though, is how often are scholars who work part-time as editors compensated financially? In the areas I know, this is uncommon, and when it occurs, is minor. (For example, one journal of whose editorial board I now serve is paying its two managing editors $1,000 per year each.)
From Stevan Harnad Date: Mon, 25 Jul 94
My prediction is that the cost of Editor's services and editorial office staff will vary with:
(1) Discipline and specialty (and perhaps degree of interdisciplinarity) (2) Journal size (annual published page count) (3) Submission rate (and rejection rate) (4) Journal prestige (perhaps) (5) Subscribership and readership size (perhaps) (6) Esotericity of contents (readership per ARTICLE as opposed to subscribership per journal) (7) Prose-intensiveness of contents
Because refereeing is a volunteer, ad-lib service we all perform when we happen to have the time and interest for a particular paper or grant proposal, whereas Editing (and, a fortiori, editorial office work) is a fixed, timetable-bound commitment, I believe they will continue to be represent part (indeed most, along with copy-editing costs) of the true per-page costs of quality control in electronic-only periodical publishing (which I estimate will be < 25% of the per-page costs of paper periodical publishing). -- SH.
From: amo@research.att.com (Andrew Odlyzko) Date: Mon, 25 Jul 94 08:18 EDT
Stevan,
I've sent out a request for information on the costs of peer review to several mailing lists and also a diverse group of individuals. If you have sent your message of yesterday afternoon to some other lists, please do forward my request for information to them as well. Since we are talking of concrete data, we should be able to obtain some estimates of what is going on.
Below I comment on a few of your remarks. It would be fascinating to study in detail why various fields differ so much in their publishing practices.
Best regards, Andrew
[Ed. Note: sh> indicates Stevan Harnad's message of Sunday, July 24, addressed to Andrew Odlyzko and several electronic mailing lists]
text that is left-justified: Andrew Odlyzko's comments of July 25
sh> Dear Andrew, sh> sh> I will try to itemize quite explicitly what comprises that residual sh> " < 25% of paper per-page costs" for quality control that will continue sh> to need to be covered in electronic-only periodical publication. Apart sh> from spelling this out explicitly, I also note in passing what sh> might be some cross-disciplinary differences (especially between highly sh> technical and symbolic texts, as in mathematics, and more sh> prose-intensive disciplines -- the latter constituting the vast sh> majority of the esoteric scholarly/scientific periodical corpus). In sh> certain important respects, my own discipline of "cognitive science" (a sh> mix of experimental psychology, theoretical psychology, brain science, sh> biology, computer science, linguistics and philosophy) is perhaps sh> better positioned than mathematics to provide a representative model sh> that would apply to most of the rest of learned publication (though sh> there may well be other views on this).
Yes, the differences do seem to be related to the different disciplines that we operate in. As to which is more representative of scholarly publishing, that is something that we should ascertain. My experience is based not just on mathematics, but also computer science and parts of engineering, as I will explain later.
sh> I agree about (a) and its costs. But I have to point out that in over sh> 15 years of editing Behavioral and Brain Sciences and 5 years of sh> editing PSYCOLOQUY, I have never once encountered a paper where the sh> author's final draft could be published verbatim! In any case, this is sh> not the real issue, as you will shortly see; the real issue is the cost sh> of generating a publishable peer-reviewed text, and that consists sh> (relatively seamlessly) of all of (b) plus copy editing (i.e., one sh> component of (c)).
I agree with you about the real issue. However, while you have never once encountered an author's final draft that could be published verbatim, I have even seen a few initial drafts that were published essentially verbatim (aside from minor copy editing that turns a reference listed as "7. S. Harnad, ...." into "7. Harnad, S., ...", say). Further, in the areas I know, it is very common for final drafts (after revisions requested by editors and referees) to be published verbatim (again subject only to minor copy editing).
sh> Let us say we agree, roughly speaking, about copy-editing, the only sh> expense in (c) worth mentioning, once we move to electronic-only sh> publishing. But my estimate of < 25% of paper per-page costs in sh> electronic only periodicals was definitely NOT based only on the cost of sh> (c); (b) costs money too, and is far more important than copy editing.
This seems to be the crucial difference in our views. In my fields, (b) costs very little, and (c) is where almost all the cost resides. (Note that my cost estimate of $ 4,000 per article was for mathematics and computer science, although it seemed to be consistent with the data I had for physics as well. We might need to get cost data for other areas.)
sh> First let me itemize what (b) entails (and I will argue that the sh> copy-editing component of (c) is probably best assimilated seamlessly sh> with (b), but not much hangs on that):
As a slight distraction, let me say that I feel that the copy-editing component of (c) is probably best assimilated seamlessly with (a), not with (b). If we are going to move to the kind of electronic journals that I foresee, with a continuum of publication and refereeing, it will make sense for authors to devote more effort into making their preprints easy to read.
sh> So there you have it. In my view, it's (1) - (6) that underlie the true sh> per-page costs of electronic-only publication, and I think the < 25% sh> figure will derive mostly from (1) - (5).
Broadly speaking, I accept your description of steps (1) - (6) as describing the peer-review process. (I have some minor quibbles, but they are not worth bothering about.) I also agree on the desirability of having secretarial assistance for editors. In my essay, "Tragic loss ...," I estimated this can be provided at a cost of $100-200 per published paper in the electronic world (where less time will be needed, since all the physical work of making copies, sticking correspondence into envelopes, etc., will be eliminated). With present print journals, this type of assistance is likely to cost $200-400 per paper, if it were provided to all editors.
The big difference in our views is on the issue of costs of scholar editors. As I said in the message I sent out a few minutes ago, my experience is very different from yours. We'll have to gather some data to decide which model is more common. If we don't obtain good enough information from the message I sent out, I propose that we do a quick survey. Since Princeton is likely to be much more representative of the broad spectrum of scholarly activities than Bell Labs, how about contacting the department chairs at Princeton and asking them what the common practices in their fields are? I would be happy to help in this work.
Do unpaid editors produce acceptable quality? They seem to, at least in my areas. Moreover, this is achieved in various ways. One way is by having small journals. Mathematics has slightly under 1,000 journals, as I recall, and many are small, publishing perhaps 400 pages per year. Thus the editors do not have too heavy a load. However, there are also large journals. Some, such as Proc. AMS, used to have a totally decentralized approach when I was on their board. There were a large number of editors, each with an associated specialty that was listed in the journal, and authors sent their papers to the editor covering the area of the paper. Simplifying a bit, each editor could accept a paper on his or her own. A still different approach is in force at Math. Comp., where all submissions go to the Managing Editor, who then decides which of the associate editors is most appropriate for that paper, and forwards the paper for handling to that paper.
Andrew Odlyzko
From: Stevan Harnad (harnad@princeton.edu)
Dear Andrew,
I only circulated the "Itemized Costs" posting to the lists to which you sent your query (and they are also the lists most actively involved in these questions, hence perhaps best posed to answer them): vpiej-l, serialst, globalsci. I now send this to one further list, my own internal alias list of individuals who are interested in electronic publication. Some will be on the big lists, but just in case, could you please re-send me your survey and I will branch it to them.
A systematic survey across disciplines certainly needs to be conducted -- and I wish I had the time to do it, but I'm about to move myself and my lab to the UK (University of Southampton) to take up the Directorship of a new Cognitive Sciences Centre and a Professorship in Psychology, so in the next couple of months I will not even be able to do the Princeton survey you suggest below, unfortunately. (If there were a Departmental Chairs email list, I could do it quickly, but there isn't, and I unfortunately haven't the time or the staff to do it piecemeal right now, or to process the results).
From: amo@research.att.com Date: Mon, 25 Jul 94 08:18 EDT
I've sent out a request for information on the costs of peer review to several mailing lists and also a diverse group of individuals. If you have sent your message of yesterday afternoon to some other lists, please do forward my request for information to them as well. Since we are talking of concrete data, we should be able to obtain some estimates of what is going on.
Your survey is an excellent idea. I too would like to have real data on these questions. Besides discipline differences, you'll probably also find differences as a function of the size/page-count of the journal, whether it is narrow-specialty or multi-specialty, and perhaps even as a function of the size of its readership (esoteric vs. broad spectrum).
Below I comment on a few of your remarks. It would be fascinating to study in detail why various fields differ so much in their publishing practices.
Yes, the differences do seem to be related to the different disciplines that we operate in. As to which is more representative of scholarly publishing, that is something that we should ascertain. My experience is based not just on mathematics, but also computer science and parts of engineering, as I will explain later.
I suspect that my distinction between ad-lib and fixed, clock-linked commitments will be an important variable, as will the submitted and accepted manuscript-page counts.
I agree with you about the real issue. However, while you have never once encountered an author's final draft that could be published verbatim, I have even seen a few initial drafts that were published essentially verbatim (aside from minor copy editing that turns a reference listed as "7. S. Harnad, ...." into "7. Harnad, S., ...", say). Further, in the areas I know, it is very common for final drafts (after revisions requested by editors and referees) to be published verbatim (again subject only to minor copy editing).
I suspect this will vary with the prose-intensiveness of the field.
This seems to be the crucial difference in our views. In my fields, (b) costs very little, and (c) is where almost all the cost resides. (Note that my cost estimate of $ 4,000 per article was for mathematics and computer science, although it seemed to be consistent with the data I had for physics as well. We might need to get cost data for other areas.)
My experience is that the lion's share of the cost is (b) rather than (c), and that that cost is fairly well predicted by the number of hours called for from the Editor(s), the Editorial Administrator and the secretaries. If you include (c), you just add in the hours of the copy-editor(s). Don't worry too much, though, because I continue to hold to the < 25% figure for all these costs together.
As a slight distraction, let me say that I feel that the copy-editing component of (c) is probably best assimilated seamlessly with (a), not with (b). If we are going to move to the kind of electronic journals that I foresee, with a continuum of publication and refereeing, it will make sense for authors to devote more effort into making their preprints easy to read.
Authors can be relied on for a lot, but no author can be a self-referee, editor, or copy editor. Even if (as I think they should and will), academic departments take on staff copy-editors to help with this, the individual needs to be someone other than the author. With journal publication, though, I suspect that some centralized monitoring will continue to be necessary, over and above the enhanced role of author, word-processor and house copy-editor. I think this is true of all human quality-control enterprises: People cannot be relied upon to do their own quality control in any sphere -- and there will always be a (marked) difference between relying on someone else to ensure that certain quality standards are met or instead leaving this up to each individual. (Andrew, you cannot get a sense of this from introspection on your own conscientiousness, or that of particular colleagues; we are talking about a Gaussian population of frail mortals here; quality control means answerability.)
Broadly speaking, I accept your description of steps (1) - (6) as describing the peer-review process. (I have some minor quibbles, but they are not worth bothering about.) I also agree on the desirability of having secretarial assistance for editors. In my essay, "Tragic loss ...," I estimated this can be provided at a cost of $100-200 per published paper in the electronic world (where less time will be needed, since all the physical work of making copies, sticking correspondence into envelopes, etc., will be eliminated). With present print journals, this type of assistance is likely to cost $ 200-400 per paper, if it were provided to all editors.
Remember, our estimates of the orders of magnitude (by proportion) are about the same; we just seem to be itemizing it differently.
The big difference in our views is on the issue of costs of scholar editors. As I said in the message I sent out a few minutes ago, my experience is very different from yours. We'll have to gather some data to decide which model is more common. If we don't obtain good enough information from the message I sent out, I propose that we do a quick survey. Since Princeton is likely to be much more representative of the broad spectrum of scholarly activities than Bell Labs, how about contacting the department chairs at Princeton and asking them what the common practices in their fields are? I would be happy to help in this work.
Alas, my move leaves me too overloaded to undertake this now (and especially to process the replies). By the way, I did not say that editorial honoraria and/or time-buy-outs were large, just that they were non-zero. But it is really in the administrative and secretarial back-up support (which you describe as "desirable" whereas I would call it absolutely essential) that the significant expenses occur.
Do unpaid editors produce acceptable quality? They seem to, at least in my areas. Moreover, this is achieved in various ways. One way is by having small journals. Mathematics has slightly under 1,000 journals, as I recall, and many are small, publishing perhaps 400 pages per year. Thus the editors do not have too heavy a load. However, there are also large journals. Some, such as Proc. AMS, used to have a totally decentralized approach when I was on their board. There were a large number of editors, each with an associated specialty that was listed in the journal, and authors sent their papers to the editor covering the area of the paper. Simplifying a bit, each editor could accept a paper on his or her own. A still different approach is in force at Math. Comp., where all submissions go to the Managing Editor, who then decides which of the associate editors is most appropriate for that paper, and forwards the paper for handling to that paper.
There are many variants; page-counts are important. One cannot legislate that there should be many small journals (that may not even be a good idea). And the question of centralization is also partly a practical one: In principle, one could divide all submissions into 6-paper modules and assign them to a different "Editor," but what would that do for the reliability of quality overall, and for that journal's "imprimatur" when people are trying to calibrate their reading and browsing on the basis of reliable quality-tagging?
Managing Editors (I've avoided the term) are another matter; usually they are involved in the business end of the journal. But where they professionalize the Editor's traditional functions (referee selection, referee report assessment, dispositions, revision assessment) another risk is run, namely, that of taking quality control out of the hands of the peers themselves. In my view, an Editor should be a Primus Inter Pares rather than a professional Manager of some sort. It would be interesting to hear statistics here too.
Stevan Harnad
Date: Mon, 1 Aug 94 20:44:12 EDT From: "Stevan Harnad" harnad@princeton.edu
Responses to Andrew Odlyzko's Questionnaire about Electronic Editorial Costs follow below. -- S.H.
Date: Fri, 29 Jul 94 08:34:56 EDT From: Janet Fisher FISHER@MIT.EDU Subject: Editorial Costs To: Andrew Odlyzko AMO@RESEARCH.ATT.COM Cc: Stevan Harnad harnad@princeton.edu
Thanks for sending me a copy of your questions about editorial payments. I agree with Stevan's description of the process that an editor (and support staff) go through to review papers. Yes, it is common for publishers to pay a portion or all of the expenses of the editorial office. In the humanities this is less the case, but it varies tremendously depending on the editor, the editor's institution, and the competition for the journal. When publishers compete for a journal, this is where the deal can either be made or broken. These costs have increased dramatically for a large percentage of our journals in the last five years due to tightening funds at universities. We now have editors wanting us to buy them computer equipment and software, editorial tracking packages, etc.
In addition to editorial office support, some editors do indeed receive royalties (in our case, usually after the journal has reached a break-even position and the Press has recovered its initial deficits). Or the bottom line can be split with the editor and/or the editor's institution (if they own the journal). Patricia Scarry (U of Chicago Press) and Jill O'Neill (Elsevier) gave a presentation on editorial office costs at the last Charleston Conference. Perhaps they might have copies of their presentations to share.
We have a diverse list of journals in disciplines from humanities to social sciences to computer and cognitive science, with a diverse range of arrangements across disciplines. Many, many disciplinary and institutional factors come into play in the way journals are supported and financed. There is no single formula.
The other point I would make is that most journal editors accept between 25% and 35% of the papers actually received. Possibly 10% are rejected outright, but the rest of the rejected do go through the review process and possibly through a revision stage also. These take up the time of the editor and the editor's staff also, and this time has to be paid for too. Only in a very few fields (like economics) are submission charges common.
(Note from S.H.: The acceptance rate varies greatly from discipline to discipline. The acceptance rate in physics and mathematics is more like 75-80% and author page charges are less uncommon there.)
Most of our editors have at least a half-time assistant to handle the clerical parts of the editorial tasks (acknowledging manuscripts, contacting potential reviewers, sending manuscripts out for review, dogging reviewers, writing authors, etc.), and if the assistant is a "managing editor type" and also does copyediting, this is more likely a full-time position. These costs can be anywhere from $12,000 to $30,000 per year just for that staff position. (Not including equipment, phone, fax, postage, office space, university overhead [yes, really!] that universities often try to recuperate.)
Editorial board members are usually not paid, but this doesn't mean that Editors are not. A few of our journals given token payments to Associate Editors (the usually three top people under the editor) but not to editorial board members. The fact that you have been a member of the editorial board of 18 journals and never been paid is consistent with our experience. But you cannot conclude from that fact that editors are not paid and editorial offices are not paid for by the journal or publisher.
I would also argue that there will still be some "typesetting" cost with electronic journals. I do not believe that authors -- even in the most highly sophisticated fields -- will ever do all the formatting required to take manuscripts directly without some intervention. "Typesetting" will really become formatting, I guess, but there are costs associated with this. We should know more about what these are once Chicago Journal of Theoretical Computer Science begins publication.
I guess that's all for now. Let me know what editorial costs you are interested in and if you have questions -- or disagreements -- over anything in this message.
JANET H. FISHER ASSOCIATE DIRECTOR FOR JOURNALS PUBLISHING MIT PRESS, 55 HAYWARD STREET, CAMBRIDGE, MA 02142 FISHER@MIT.EDU PHONE (617) 253-2864
Date: Tue, 26 Jul 94 12:51:25 EDT From: VMONTY@VM2.YorkU.CA (Vivienne Monty)
Hi: I shall look for harder data but in Library Science and History, two fields that I am familiar with, I have never known a scholar/editor to be directly paid. It is mostly in terms of release time or such remuneration that I have known. Even these release time arrangements are hard to get now in Canada at least.
Often the scholar/editor however has a "bureaucracy" to call on at the Association sponsoring the journal or the publisher who take care of the day to day administrative operations. The key word is often and NOT always however.
As stated earlier in your discussions, the world of academe "sponsor" academic publishing to a large degree through the granting release time, research leaves and the personal time of scholars. Universities have a tremendous dollar value investment in editorships, writing etc. whether some realize it or not. And some count such time as zero.
Date: Tue, 26 Jul 94 13:02 CDT From: Jack P Hailman JHAILMAN@macc.wisc.edu Subject: Re: Odlyzko Editorial Survey
Things might be changing on the subject of paid editorships, at least my own views have changed. I served as editor of Animal Behaviour for three years (or was it five?), and never again would I devote that much of my life uncompensated. I wonder if other former editors of major international journals feel the same way?
Date: Tue, 2 Aug 94 11:30:00 EDT
From: "Stevan Harnad"
From: amo@research.att.com (Andrew Odlyzko) Date: Tue, 2 Aug 94 06:07 EDT
I cannot speak about physics with any confidence, but the acceptance rate for a typical journal in mathematics is usually considerably lower than 75-80%. Judging from my own experience in serving on a variety of editorial boards, I would estimate that rate is perhaps around 50%. The rate of acceptance to all the journals in mathematics is higher, and might easily approach 75%. If a paper is rejected because of a serious mistake, or else because the results had been published previously, then that usually ends the matter. However, when a paper is rejected because of the much more subjective judgement that it is not of sufficient quality, the author typically submits it to another journal.
It would be extremely useful to get exact figures across disciplines. Apart from the social science acceptance rates (25%) cited by Janet Fisher and the physical science rates (75% in physics, perhaps 50% in math), there are the biomedical sciences (low acceptance rates), the humanities (probably varied), and interdisciplinary journals (Science, Nature, etc.) with very low rates. And the acceptance rates probably rise as one descends in the prestige hierarchy (except where self-selection keeps submissions to prestigious journals at a high level of likely acceptability, as in the top physics journals). All these data would be useful to have. The articles I cited (Cichetti, Hargens) report some of it.
In evaluating the costs of running a journal, it is the 50% acceptance rate that is the significant one, not the 75% rate. The work involved in handling a rejected manuscript is usually comparable to that of an accepted one.
I agree completely. And another figure that needs to be calculated field by field is the ULTIMATE (cross-journal) acceptance rate: It is my belief that in one form or other, just about EVERYTHING gets published eventually, if the author is persistent enough, even if it's in the unrefereed vanity press. Having approximately the same manuscript refereed repeatedly for different journals is a drain on resources, but I'm not sure how to get around it: the prestige hierarchy is based in part on (intellectual) competition.
The other remark is that page charges have essentially disappeared in mathematics. They have been bringing in less and less revenue, and the American Mathematical Society, for one, has eliminated them completely.
This has to be re-thought. Page charges made little sense in paper publication, since the publisher needed to take copyright and charge subscribers anyway. Author page charges were usually just a voluntary supplement, sometimes offered as a way of speeding publication. But in the electronic medium, where total page costs would be so much lower (75% lower) and reader access would be so rapid, global and free, it should be re-thought whether it would not in fact be to EVERYONE's benefit (especially the author's) if the requisite advance subsidy to cover the FULL minimal costs per electronic "page" came from a combination of learned society, university, library, and author-publication-grant sources.
Stevan Harnad
From: Ann Okerson ann@cni.org Date: Tue, 2 Aug 1994 12:19:03 -0400 (EDT)
About the topic of acceptance rates:
There is not a great deal of real data published about this particular question. There are a lot of general, informal speculations and assertions. The acceptance rates for individual journals are certainly not the same as the overall acceptance rates across all journals in a field.
According to a presentation I heard a couple of years ago from the Editor of the PMLA, a major journal in the modern languages area published by the Modern Languages Association, the rejection rate for PMLA is in the low 90% range. The journal is relatively small, highly prestigious, and has not grown commensurately in physical size with the growth in the size of the literature of the field. The editorial board works diligently to select the small proportion of submitted articles that can be published, but the Editor-in-Chief affirmed that "much" of the work that is rejected is of high calibre, and "the great majority" of it ends up published elsewhere, often in more specialized journals. He presented no data beyond that.
According to figures from Louise Addis, formerly Librarian at the Stanford Linear Accelerator, slightly over 70% of the high energy physics preprints that are accessible via SLAC's major database of same, are eventually published somewhere in the print physics literature. That is, the finished products are recognizably close to the original preprint and thus the librarians can indicate with the preprint record that the work has appeared in Journal XYZ with a standard citation to it in its "finished" form.
In attendance at many, many meetings of societies, publishers, and libraries on the topic of scholarly journals and communication, I have heard many generalizations about the rejection and acceptance rates in the sciences, social sciences, and humanities. Yet in each of these broad areas, the range of acceptance/rejection across journals must be very, very great, for the "averages" rarely resemble the specific data quoted for any individual title, such as the PMLA.
Absent real data, this suggests that one should be careful of making generalizations. What does seem true is that a great majority of work is eventually published somewhere. In high energy physics, we know it's close to at least 3/4 of all submissions. I've always thought that passing an article through two or three or more editor's or publisher's hands wastes some of the time of the system overall. (Note that this competitive process is also the way that book submissions, particularly in the trade market, work and the mechanism by which work is rewritten, revised, and improved.) I hope, possibly naively, that some of the current duplication of editorial and reviewing effort can be reduced as the process of scholarly communication is more and more electronically supported.
Ann Okerson/ARL ann@cni.org
From: amo@research.att.com (Andrew Odlyzko) Date: Tue, 2 Aug 94 12:46 EDT
I just realized that I did not express myself clearly in the message I sent early this morning. (I'll blame this on a rush job, as I am leaving on a trip this afternoon.) The 50% acceptance rate I estimated for mathematics was for the average journal. The 75% rate was an estimate of what you call "the ULTIMATE (cross-journal) acceptance rate," namely the fraction of all preprints that ever get published. In mathematics, a substantial fraction of papers are discovered to be incorrect or not to be novel, so that not everything gets published.
The problem of papers going from journal to journal and referees having to redo each other's work is one reason I was proposing making referee reports public (possibly after some modifications by referees), and also having referees assign significance grades to papers, so that there might in effect be only a few journals. However, this might not be acceptable psychologically.
From: AJ Meadows A.J.Meadows@lut.ac.uk Subject: Electronic Journal Costs Date: Wed, 3 Aug 94 13:38:25 BST
I have been following with interest your discussion of this topic. Just one or two minor comments:
(1) The best source I know for data on journal use and economics is: D. W. King et al., Scientific Journals in the United States (Hutchinson Ross Pub Co; 1981). It refers to print journals only and is a little out of date, of course.
(2) Comparative data on journal rejection rates was published a good many years ago by R. K. Merton at Columbia. My colleagues and I have looked at these from time to time and they still seem to give comparative rates fairly well.
(3) We are experimenting with an electronic journal that assumes distribution via a library. From this viewpoint, you have to add on the costs of the library getting ready to receive such journals. We estimate that, for the first such journal, the cost is of the order of $3,500.
Jack Meadows