SciX Open, self organising repository for scientific information exchange

Transcription

1 IST SciX Open, self organising repository for scientific information exchange D2: Scientific Publishing: To-Be Business and Information Model Responsible author: Bo-Christer Björk Co-authors: Turid Hedlund, Jonas Holmström, and John Cox Access: public Version: 1.0 Date: September 26, 2003

2 version of 29-Sep-03 11:44 page 2 EXECUTIVE SUMMARY: This study continues from the baseline established in the SciX report Scientific Publishing: Asis Business and information model. The report deals with the open research questions highlighted in the first report concerning new business models, cost analysis comparing current publishing to open access publishing and an analysis of the barriers to change. Firstly the report includes a walk-through of the expanded and revised version of the SPLCmodel. The model now includes the evaluation of the researcher based on his/her research activity. The evaluation is crucial for the researcher s career development and future research funding and affects the publication strategy of the researcher both as an individual and as a member of a an institution. The important activity, copyright negotiations, which are carried out between the publisher and the author of a research paper, is added to the model. The most common form of open access is today self-publishing by authors who put copies of their own publications on their own home pages. This form of publishing requires that the author retains the right to publish on a web page or in an institutional repository. Secondly the report analyses the cost structure of current publishers, commercial as well as noncommercial. The results of an empirical study on the characteristics, funding sources and cost structure of the new business model, the open access journal publishing are reported. This section also includes synthesis and analysis of the cost structure of the open access journal publishing compared to current mainstream publishing. The results indicate that the costs per article for open access journals are clearly lower than for mainstream print+electronic journals, but not as radically lower as some proponents of open access have suggested. In comparing the costs one also has to remember that there are differences in the level of service provided by the journals, most open access journals do not have extensive copyediting. The author charge model for open access journals could be a solution, but there are still many open questions. Thirdly the barriers to change to new business models are discussed not only for journals but also for subject-specific repositories and institutional repositories. The different business models are discussed in the context of which type of access they provide (paid access, open access, or limited access). Institutional repositories in principle offer many advantages for parallel publishing (archival security, sustainable financing) but the copyright challenges need to be resolved. The central lever for change is the point at which the author of a publication decides where to submit it (and also weather to upload a copy to a repository).

3 version of 29-Sep-03 11:44 page 3 RELEASE HISTORY Date September 26, 2003 Changes Final version, SciX template applied

4 version of 29-Sep-03 11:44 page 4 TABLE OF CONTENTS: EXECUTIVE SUMMARY:...2 RELEASE HISTORY...3 TABLE OF CONTENTS: INTRODUCTION SCIENTIFIC PUBLICATION LIFE-CYCLE MODEL (SPLC) INTRODUCTION AIM AND SCOPE OF THE MODELLING EFFORT MODELLING METHODOLOGY OVERALL ORGANIZATION OF THE MODEL MODEL WALK-THROUGH A-0 Do Research, Publish, study and Implement the Results - Context Diagram A0 Do Research, Publish, Study and Implement the Results breakdown A1 Perform the Research A2 Publish the Results A22 Perform Publishing Activities A223 Publish as Scholarly Journal Article A2231 Do General Publisher s Activities A2232 Do Journal-Specific Activities A2233 Do Article and Issue Specific Activities A22331 Do Article Specific Activities A22332 Prepare Issue A23 Archive and Index A231 Make Publication available A2313 Make Electronic Copy available A2314 Integrate Meta Data into Search Services A232 Perform Value Adding Services A3 Study the Results A31 Find Out about Publication A311 Search for Publication A312 Be alerted to publication A32 Retrieve publication A4 Apply the Results CONCLUSIONS SCIENTIFIC PUBLICATION LIFE-CYCLE MODEL (SPLC): VALIDATING THE MODEL BY COMPARISON WITH CURRENT PUBLISHER COST STRUCTURES CONTEXT THE COST STRUCTURE OF JOURNAL PUBLISHING ESTABLISHING THE IMPRINT THE JOURNAL EDITOR, EDITORIAL BOARD AND PEER REVIEW THE PRODUCTION EDITORIAL PROCESS AFTER ACCEPTANCE MARKETING JOURNALS SUBSCRIPTION MANAGEMENT AND DISTRIBUTION...43

5 version of 29-Sep-03 11:44 page OUTSOURCING AND FREELANCERS PRODUCT DEVELOPMENT: BUILDING THE JOURNAL LIST COMMENT ON THE SPLC MODEL THE COSTS OF THE CONVENTIONAL PUBLISHING MODEL SCIX SURVEY ON THE CHARACTERISTICS AND COST STRUCTURE OF OPEN ACCESS SCIENTIFIC REFEREED JOURNALS METHODOLOGY AND RESEARCH SETTINGS THE QUESTIONNAIRE THE RESULTS THE REVIEW PROCESS AND THE QUEUE TIME A PROFILE OF AN OPEN ACCESS JOURNAL, BASED ON SURVEY RESULTS AND FREE COMMENTS BY THE EDITORS COST ANALYSIS OF THE SCIENTIFIC PUBLICATION LIFE-CYCLE MODEL: COSTS FOR PUBLISHING THE RESULTS THE OBJECTIVE OF THE COST STUDY RESEARCH SETTINGS Identification of publication models Cost data categories and cost measures Data collection methods RESULTS OF THE COST DATA STUDY Prevalence of scientific refereed open access journals All open access journals Journals using author charges Library costs ANALYSIS AND DISCUSSION OF THE RESULTS Impact of the different publication models on the costs for the publish the result activity Impact of the publication models on library costs...79

6 version of 29-Sep-03 11:44 page 6 6. OPEN ACCESS TO SCIENTIFIC PUBLICATIONS AN ANALYSIS OF THE BARRIERS TO CHANGE? INTRODUCTION WHAT IS OPEN ACCESS? EXPERIENCES OF THE PIONEERING YEARS BARRIERS TO CHANGE LEGAL FRAMEWORK Open access journals Subject-specific repositories Institutional repositories IT INFRASTRUCTURE Open access journals Subject-specific repositories Institutional Repositories BUSINESS MODELS Taxonomy of business models for scientific journal publishing Paid access Open access Limited open access Open access business models INDEXING SERVICES AND STANDARDS Open Access Journals Subject-specific repositories Institutional Repositories ACADEMIC REWARD SYSTEM Open Access Journals Subject-specific repositories Institutional repositories MARKETING AND CRITICAL MASS Open Access Journals Subject-specific repositories Institutional Repositories CONCLUSIONS CONCLUSIONS REFERENCES: Appendices Appendix 1.Model operating costs Appendix 2.P&L per journal Appendix 3.The SPLC model Appendix 4. Current Literature on Economic Issues Related to journal publishing and Emerging Publication Models

7 version of 29-Sep-03 11:44 page 7 1. INTRODUCTION This report constitutes deliverable two of work package one (process and social-economic analysis) in the SciX project, funded by the European Commission under the contract IST The first report titled Scientific publishing: as-is business and information model, included a description of the current state of scientific publishing. Is also introduced the process model of the scientific publishing life cycle (SPLC-model). The first part of this report includes a walk-through of the expanded and revised version of the SPLC-model. Note that the version of the model presented in this paper is the second first draft and that the model is continuously evolving based on the feedback we receive. The SPLCmodel has so far proved to be a useful framework for studying different aspects of scientific publishing. The second part presents three studies of both mainstream publishers and open access publishers. The first study was conducted by the international publishing consultancy John Cox Associates. The study is based on cost data from five publishers and describes the key features and economics of journal publishing as mainstream publishers currently practise it. The second study was conducted in May 2003 as a web survey. It consisted of a survey sent to editors of open access scientific refereed journals. The study focused on acquiring general information about the journal and the publisher as well as relevant cost data and intellectual property rights. The third study of part two is a synthesis and analysis of the cost data acquired from the second study and cost data reported in the literature. Barriers and business models are discussed in the third part. The barriers and business models are discussed not only for journals but also for subject-specific repositories and institutional repositories. The different business models are discussed in the context of which type of access they provide (paid access, open access, or limited access.) Those who are interested in broadening their knowledge of the issues in scientific publishing may consult some of the references listed in appendix four. In the third and final report titled Recommendation, model comparison we will compare the economies of the as-is and to-be models, take into the account the experiences with the prototypes, the surveys, the studies of the barriers to process change and issue the recommendations for the near future. We encourage interested parties to provide feedback on this report.

8 version of 29-Sep-03 11:44 page 8 2. SCIENTIFIC PUBLICATION LIFE-CYCLE MODEL (SPLC) 2.1 INTRODUCTION The scientific publishing process has during the past few years undergone considerable changes, due to the possibilities offered by the Internet for fast delivery and interlinking of publications which refer to each other. The socio-economic structures have, however, not changed much, and many academics and librarians view the current situation as sub optimal and highly unsatisfactory. This has triggered a number of initiatives to set up e-print repositories and electronic peer reviewed journals, which usually offer the full text for free on the web. The label Open Access is gaining popularity for describing such efforts, in line with the Open Source term used in the software development community. Despite the obvious advantages it offers, Open Access hasn t become as widespread as expected, and there is a need for both successful demonstrator projects and serious research studying the way the scholarly communication system is affected by the Internet. As a part of the EU funded SciX project the cost implications of different publishing models are being studied. A basis for the cost studies is the formal process model of the scientific publishing process, which is presented in this paper. The model has been developed using the IDEF0 modelling method, a method which allows the breaking up of a process into its parts connected by output and input arrows. The Scientific Publication Life-Cycle Model models the life cycle of the single publication, in particular the refereed journal article, from the research leading to it and writing it into it being read by other researchers years later or used as an impulse for practical implementation. This part presents the 22 hierarchical diagrams of the model including 64 separate activities. Conclusions are drawn about the usefulness of the modelling methodology for this particular purpose as well as of future uses of the model itself. In addition to providing a basis for cost studies the model could function as a road map for different types of Open Access initiatives. A breakdown of the costs of producing and delivering a typical refereed journal paper indicates that perhaps as much as 90 % of the cost consists of the actual research work preceding the writing of the paper. The research work is usually financed by public bodies and the costs are in no way recuperated through the sales of the publications (as would be the case for commercial products such as books, music CDs, movies etc). Even if we look only at just the costs of preparing, reviewing, distributing, archiving and retrieving scientific articles, thus excluding the actual production costs of the knowledge reported on, almost all the costs are in the end born by the universities, public libraries etc. Because of the commercial interests of one group of stakeholders, the journal publishers, which incur a very small fraction of the total life-cycle cost, the access to scientific publications is highly restricted and expensive and the process as a whole is highly inefficient [for a recent overview cf. Guédon 2001]. Publishing parallel electronic versions restricted to subscribers only partly remedies the situation and leaves much of the potential of the Internet untapped.

9 version of 29-Sep-03 11:44 page 9 The dilemma is that it would be in the interest of the researchers and the public to have all this information published for free on the Internet, facilitating global access and hyper-linking of research publications. Nevertheless it is in the legitimate interest of the publishers to make a profit from selling this information, which leads to restricted circulation, pass-word protection schemes for digital versions of traditional journals, bundling of electronic subscriptions to so called big deals with library consortia. Pioneering work to try out new methods of scientific publishing has already been started, usually by enthusiasts from subgroups in the scientific community. Examples are found in chemical science as well as mathematics (Town, Vickery, Kuras & Weeks 2002; Wegner 2001) Examples include preprint repositories, open access scholarly journals and journals offering a review process where the readers perform the peer review after a manuscript has been posted on the web. There are, however, many psychological, legal and institutional barriers to change the process and these have been underestimated by the pioneers and enthusiasts. Consequently only a small part of the overall volume of the scientific communication process has so far been affected by isolated efforts involving e-journals and preprint archives. A recent study made at the Swedish School of Economics and business studies indicates that only about 0,7 % of peer reviewed journals offer open access on the web. The SciX (Open, self organising repository for scientific information exchange) project, which is financed through the IST programme of the European Commission, aims at demonstrating that the Internet enables new business models for the scientific publishing process which are much more cost and time efficient to the scientific community than the current practice [SciX 2002]. The SciX project group will create services on the Web that will enable scientists as well as practitioners from the fields of architecture and construction easy and free access to relevant research publications. In addition existing publishing practices will be analyzed systematically and business models for re-engineering the scientific publishing process will be developed, taking into account also the legal, social and psychological barriers to change. The model presented in this paper is one of the deliverables of the theoretical work done within the SciX project. 2.2 AIM AND SCOPE OF THE MODELLING EFFORT The aim of the modelling is to help us understand the scientific publishing process and how it is affected by the Internet, in order to provide a basis for a cost and performance analysis of various alternative ways of organizing it. The model can also work as a roadmap for positioning various new initiatives, such as e-print repositories and harvesting tools, within the overall system of scholarly communication. The model explicitly includes the activities of all the stakeholders involved in this system, including the activities of the: Researchers who perform the research and write the publications Publishers who manage and carry out the actual publication process Academics who participate in the process as editors and reviewers Libraries that archive the publications and provide access to the them

10 version of 29-Sep-03 11:44 page 10 Bibliographic services which facilitate the identification and retrieval of publications Readers who search for, retrieve and read publications Practitioners who implement the research results directly or indirectly In the model the unit of observation is the single publication, how it is written, edited, printed, distributed, archived, retrieved and read, and how eventually it may affect practice. The viewpoint taken is life-cycle cost per publication. Thus at later stages all cost and time data which is collected will be translated to a per publication basis. The aim is to clarify the process and to find a way of measuring the total life-cycle cost of a publication, since the objective of the reengineering efforts in the SciX project is to try to investigate business models that minimise the total life-cycle costs, rather than the cost of some particular stage. This is clearly in the interest of the public authorities that finance research as well as the researchers who are primary end users. Another aim of the modelling effort is to try to clarify the functions that the demonstrator parts of SciX try to make more effective. The model depicts publishing and value added services using both paper and electronic formats. Pure electronic or pure paper-based publishing could be described by subsets of the model. The same goes for free publishing on the web ( open access ), which resembles traditional publishing, but where certain activities such as negotiating, keeping track of and invoicing subscriptions can be almost entirely left out. The current version of the model has some limitations, which should be kept in mind. It only includes the publication and dissemination of research results in the form of publications that in the end can be printed out and studied on paper (irrespective of whether the publications are distributed on paper or electronically). Thus forms of communication such as oral communication, unstructured use of and multimedia, sharing of data sets and models, which all are essential parts of the scientific communication process, are out of scope. These could be added at a later stage, but would also add to the complexity of the model. The model includes some activities, which would be typical for a scientific publisher publishing several journals, allowing for economies of scale. The activities of single-journal publishers could be described by a subset. The reason for including activities such as the general activities of a publisher is that these significantly influence the cost of running individual journals in the form of the general overhead costs that publishers add to the subscription prices. How easier access to scientific publications might influence the quality of later research and industrial practice, which use these publications as input, is clearly also an important issue, but such qualitative effects of a more efficient process are very difficult to model and even more difficult to measure, and have not been attempted in this model. The same applies for the effect of the publishing on the careers of the authors, which is an important aspect for the choice of where to publish and has created a very strong barrier for change to pure electronic journals from established brand name journals.

11 version of 29-Sep-03 11:44 page MODELLING METHODOLOGY Formal process modelling techniques are frequently used in process re-engineering efforts to clarify a process and propose changes in it. Using a formalized tool helps in communicating about the process. The IDEF0 method was originally developed in the 1970s for the US Army to help them in large procurement projects. In order to read the model the reader needs some familiarity with the IDEF0 modelling methodology. IDEF0 stands for Integration Definition for Function Modelling [NIST 1993]. The IDEF0 method is a formal graphical modelling method. (Figure 1). Control Input Output Activity Mechanism A0 A1 A2 A3 Diagrams are hierarchical More general A21 A22 A23 A24 More detailed A2 Figure 1. The basic concepts of the IDEF0 method [Karhu 2000]. The main concepts are the activity and the flow. The flow can be used as input, output, control or mechanism. An input represents something which in an activity is consumed to produce an output. Typical inputs could be raw materials, energy, human labor, but also information when the purpose of the activity is to transform the information to provide added value. Outputs can be reused as inputs to further activities, and feedback loops are possible. Activities are controlled by controls. Typical examples could be laws, guidelines and instructions for carrying out an activity etc. Mechanisms, which point at activities from below, are persons, organisations, machines, software etc. which carry out the activities. The presentation of the IDEF0 diagrams is hierarchical in that diagrams on lower levels provide detailed breakdowns of the those from the higher ones, Figure 1. In the SciX project HANKEN has used a particular software tool called BPwin for making and editing the IDEF0 model. Compared to a simple drafting tool BPwin enhances the speed and consistency of the modelling work, especially for larger models and when changes are needed.

12 version of 29-Sep-03 11:44 page OVERALL ORGANIZATION OF THE MODEL The current version of the SPLC-model includes 22 separate diagrams, arranged in a hierarchy up to seven levels deep. There are typically three activity boxes on each diagram, although there are a couple of diagrams with more activities and some with only two. Although official IDEFO guidelines recommend using up to six activities per diagram, it was felt that models with three activities are easier to read and understand. There are altogether 64 activity boxes and around 200 arrows. The overall hierarchical breakdown of the model is shown below in table 1. Only the separate diagrams are shown. Some diagrams are further broken down into separate activities. In the following each diagram is explained separately. The diagrams are numbered using the standard IDEF0 numbering scheme, which helps keeping track of the hierarchical position of each diagram.

13 version of 29-Sep-03 11:44 page 13 A-0 Context Diagram A0 Do Research, Publish, Study and Exploit the Results A1 Perform the Research A2 Publish the Results A21 Write Manuscript A22 Perform Publishing Activities A221 Publish as Monograph A222 Publish as Conference Paper A223 Publish as Scholarly Journal Article A2231 Do General Publisher s Activities A2232 Do Journal Specific Activities A2233 Do Article and Issue Specific Activities A22331 Article Specific Activities A22332 Prepare Issue A22333 Publish Article. A224 Publish in Miscellaneous Form A23 Archive and Index A231 Make Publication Available A2311 Secure Access Rights and Subscription A2312 Make Paper Publication Available A2313 Make Electronic Copy Available A2314 Integrate Meta Data into Search Services A232 Perform Value-Adding Services A233 Archive Securely A3 Study the Results A31 Find out about Publication A311 Search for Publication A312 Be Alerted to Publication A32 Retrieve Publication A33 Read Publication A4 Implement the Results Table 1. Hierarchical breakdown of the model, only the diagrams are shown

14 version of 29-Sep-03 11:44 page MODEL WALK-THROUGH A-0 DO RESEARCH, PUBLISH, STUDY AND IMPLEMENT THE RESULTS - CONTEXT DIAGRAM This is the diagram for depicting the overall model, which traditionally is the starting node of all IDEF0 models, and which contains only one activity describing the overall process. The philosophy of this diagram is to show how science can help in solving problems of everyday life by a combination of research (which creates scientific knowledge) with publishing, dissemination and exploitation activities (which ensure that results help in the advance of our collective body of knowledge). The main stakeholders in the process are shown as mechanism arrows coming into the activity box from below, and the main drivers controlling the behavior of the stakeholders are shown coming in from above. Note that from an academic viewpoint the main result is new scientific knowledge. From the viewpoint of society that funds research the most important outcome is improved productivity and quality of life. Scientific Curiosity Economic Drivers Scientific Problems Do Research, Publish, Study and Implement the Results New Scientific Knowledge Improved Productivity and Quality of Life 0 Stakeholders in Research and Technology Transfer Figure 2. Diagram A-O Do Research, Publish, Study and Exploit the Results, Context Diagram

15 version of 29-Sep-03 11:44 page A0 DO RESEARCH, PUBLISH, STUDY AND IMPLEMENT THE RESULTS BREAKDOWN This diagram is crucial for understanding the life-cycle view adopted in this modelling effort. The whole life-cycle is seen as consisting of four separate stages. The fifth activity is evaluate researcher, performing as a control mechanism of the perform the research activity. The perform the research stage is probably the most expensive part, usually consisting of several man-months of work effort per resulting publication, but the one least affected by the reengineering efforts facilitated by the Internet (at least directly, indirectly the effect can be substantial in terms of better quality of the research). The publish the results and study the results stages constitute the main object of study in this project. This part of the model tries to clarify the dual nature of the publication process. From the perspective of the public bodies that to a large part finance research it is the efficiency of the total process, including both the production and consumption of publications, that should be optimized. The important thing is that in a life cycle analysis, the cost and efficiency of both the publish the results activity and the study the results activity are important. Optimizing only one of these may lead to a sub optimal solution for the total process. Here Internet has changed the situation dramatically, as it has for any information goods that can be delivered in a digital format. Scientific Method Publishing Practice Appointment & grant decisions Industrial R&D Policies Scientific Problems Perform the Research Reading Habits New Scientific Knowledge 1 Publish and Disseminate the Results Performance measures Evaluate the researcher 4 2 Retrievable Publication Study the Results Funding and Academic bodies Disseminated Scientific Knowledge 3 The Researcher Improved Productivity and Quality of Life The Researcher Implement the Results The Publisher Libraries, indexing services etc. Readers 5 Industrial Problems Society and Industry Figure 3. Diagram AO Do Research. Publish, Study and Apply the Results, Breakdown

16 version of 29-Sep-03 11:44 page 16 The end result of these two stages is called disseminated scientific knowledge, reflecting the viewpoint that scientific results which have been published, but which are not read by the intended readers are rather useless. In the last stage, implement the results, the disseminated scientific knowledge is transformed to an improved performance of society and industry A1 PERFORM THE RESEARCH This diagram shows a highly simplified view of a typical research project. Note that one important feature of IDEF0 diagrams is that the consecutive activity boxes do not necessarily imply a strict order in time as in scheduling methods. Thus the activity study earlier research can go on after the other two activities have started. The important thing is that it provides input to these. Clearly this is only one possible way of looking at the research process. The reason for choosing this view is that it clearly distinguishes the study earlier research activity, which also is part of a later stage of the whole model. Here it is seen as providing input to the research that produces a publication, whereas the later stages of the model show how other researchers utilize the results for their own separate research projects. If this activity can be made more efficient, the overall productivity of the research can be increased. Scientific and Industrial Problems Study Earlier Research Knowledge Acquisition Habits Scientific Method Existing Scientific Knowledge 1 Do Experiments and Make Observations 2 Empirical Data Research Questions and Hypotheses Analyse, Draw Conclusions New Scientific Knowledge 3 Figure 4. Diagram A1 Perform the Research

17 version of 29-Sep-03 11:44 page A2 PUBLISH THE RESULTS This part of the model has been split up into three distinct activities, which to a large extent are carried out by different stakeholders. Based on the results of the research, the researcher writes a manuscript, which then in the next stage through a number of transformations is changed into a publication (on paper or electronic). The last activity is extremely important from a life-cycle viewpoint and involves the activities of libraries, bibliographic services etc. to make the publication easily available to researchers and practitioners worldwide. Scientific Writing Style New Scientific Knowledge Write Manuscript Publishing Practice Library and Indexing Practice 1 Manuscript Perform Publishing Activities 2 Publication Archive and Index Retrievable Publication 3 The Researcher The Publisher Libraries and Bibliographic Services Figure 5. Diagram A2 Publish the Results A22 PERFORM PUBLISHING ACTIVITIES At this stage the model is split into four parallel tracks which all take the generic Manuscript as input. The term monograph is in library science used to denote scientific publications which usually are published by the university of the researcher and which are not part of a scientific periodical journal or conference proceedings. Typical examples include working papers, research reports and Ph.D. theses. Conference papers are subjected to some sort of external review either for the abstract or the full paper, and are usually presented orally in addition to the printed version. Conference proceedings are published as one-off books or typically annually.

18 version of 29-Sep-03 11:44 page 18 Articles in scientific periodicals are subjected to rigorous peer review. It is important to note that periodicals articles have a much higher likelihood of being referenced in bibliographical services than the other types. Also journals are usually available by subscription whereas the access to monographs and conference proceedings is predominantly acquired on an individual basis. The fourth track involves publication of all other types, for instance as popularized articles targeted for an audience of non-academics, self-publishing of multi-media content etc. Of these four only the publish as journal article has at this stage been further detailed. This is because of its relative importance in scientific publishing and also because the cost modelling effort will be concentrated there. Manuscript Publish as Monograph Monograph Journal Review Policy Publication 1 Publish as Conference Paper 2 Conference Paper Journal Article Publish as Scholarly Journal Article 3 Miscellaneous Publication Publish in Miscellaneous Form 4 Figure 6. Diagram A22 Perform Publishing Activities

19 version of 29-Sep-03 11:44 page A223 PUBLISH AS SCHOLARLY JOURNAL ARTICLE This diagram may at first sight be difficult to understand. The idea is to show all the activities which are carried out by the publishing organization, and thus have a direct cost implication for them. This is the reason for separating activities such as do general publisher s activities, do journal specific activities. Both of these demand resources, which cause overhead costs, which then are added on top of the basic variable costs caused by the processing of each individual article ( in the activity do article specific activities). For instance setting up and maintaining the IT-technical infrastructure for a portfolio of journals could be such an overhead causing item. The main pipeline in the model is, however, the input arrow manuscript, which directly enters the activity do article and issue specific activities. Publisher's Business Strategy Do General Publisher's Activities Plan for Running Journals Journal Review Policy 1 Do Journal Specific Activities Issue Schedule 2 Manuscript Do Article and Issue Specific Activities Journal Article 3 Infrastructure for Running Journal Figure 7. Diagram A223 Publish as Scholarly Journal Article

20 version of 29-Sep-03 11:44 page A2231 DO GENERAL PUBLISHER S ACTIVITIES General publishing activities are typical for most commercial publishers and such professional associations which publish several journals. Activities can include general management and financial functions as well as the setting up of the IT-technical structure for the production of journals (both on paper and the web). Typically such infrastructures are shared for a number of journals because of the savings due to the economies of scale. Publisher's Business Strategy Do General Administration Company Infrasteructure 1 Plan and Monitor Publishing Activities Plan for Running Journals 2 Manage IT infrastructure Infrastructure for Running Journal 3 Figure 8. Diagram A2231 Do General Publishing Activities

21 version of 29-Sep-03 11:44 page A2232 DO JOURNAL-SPECIFIC ACTIVITIES Like many of the diagrams in this model, this model represents a choice of viewpoint. Here an important aspect is that commercial journals may spend a lot of money on marketing, and also on the management of subscribers (invoicing, setting up ways of checking access to electronic versions). For open access electronic journals, the latter activity is almost non-existent. Note the output arrow issue schedule, which is later used as a control of issue-specific activities. Plan Journal Policy and Issues 1 Journal Scope and Target Audience Issue Schedule Market Journal Demand for Subscriptions 2 Subscriptions Manage Subscribers Subscribers List 3 Infrastructure for Running Journal Figure 9. Diagram A2232 Do Journal-Specific Activities

22 version of 29-Sep-03 11:44 page A2233 DO ARTICLE AND ISSUE SPECIFIC ACTIVITIES This diagram shows the two major modes for publishing scientific journals. In the paper-based world prior to 1990 articles were as a rule bundled into issues and had to wait for publishing until the whole issue was ready. Electronic publishing does however provide the possibility to publish each article as soon as it is ready. Today many journals are printed in both print and electronic formats but still retain the issue-based structure. Journal Review Policy Manuscript Article Specific Activities Issue Schedule 1 Article Accepted for Publication Prepare Issue 2 Issue Ready for Publishing Publish article Journal Article Article Accepted for Individual Publication 3 Figure 10. Diagram A2233 Do Article and Issue Specific Activities

23 version of 29-Sep-03 11:44 page A22331 DO ARTICLE SPECIFIC ACTIVITIES This part of the model depicts the activities carried out during the peer-review process as well as after the peer review to technically format the paper for printing. Note the use of a feedback mechanism, where the reviewers comments become a control of the subsequent revise manuscript activity, and where the revised manuscript is feed back into the review process. Before the advent of personal computing the typesetting activity used to be an activity incurring considerable cost, but nowadays most researchers produce text which already is formatted according to the needs of the journal. Also the actual final typesetting is much easier since almost all information is acquired in a digital format. The negotiation of copyright usually takes places after the manuscript is accepted for publication. The output is a signed copyright agreement where the author and publisher agree to the terms of publishing rights. For open access journals a copyright transfer does not usually take place since no economic rights are involved. Journal Review Policy Manage the review process Choice of reviewers Rejected manuscript 1 Review Manuscript Accepted Manuscript Manuscript 2 Reviewers Comments Negotiate copyright Copyright agreement Revise Manuscript 4 3 Copyedited manuscript Copyedit Article Publisher The Editor Reviewers The Researcher Publisher 5 Figure 11. A22331 Do Article Specific Activities

24 version of 29-Sep-03 11:44 page A22332 PREPARE ISSUE This diagram includes a very significant activity which might escape modelling in many modelling exercises, which focus solely on cost or the publisher s activities. Once an article is accepted for publishing, it enters an activity called queue for publishing, which typically takes from half a year to a year for traditional issue&paper-based journals (the worst case this author has experienced was three years). Waiting does not imply a direct cost, but there may be an important opportunity cost involved from the viewpoint of the researcher and society, since the results are poorly spread before the actual publishing. This opportunity cost is different for different domains of science. It might be low for the humanities but is usually higher in the STM (science, technology and medicine) domain. In particular this is the case for IT research, where developments are extremely fast. Issue Schedule Queue for Publishing Article Accepted for Publication 1 Add General Issue material Issue Ready for Publishing 2 Figure 12. Diagram A22332 Prepare Issue

25 version of 29-Sep-03 11:44 page A23 ARCHIVE AND INDEX This is the part of the overall process which traditionally to a large part has been handled by research libraries, with public funding. Note also that from a cost viewpoint, hundreds of libraries from all over the world have been performing the same archiving function for each paper version of an article. The primary activity is here make publication available, which secures that a publication is available either in print or electronically within a particular organization (such as a university), as well as that the publication can be found in different bibliographical search services. In the perform value-adding services a third party analyzes the data to calculate citation indexes, impact factors etc., or writes news bulletins about research results that practitioners can digest more easily. The archive securely activity is currently receiving increasing attention, since the archiving of electronic versions of journals for decades implies a number of difficult problems. Library and Indexing Practice Formats for Long Term Digital Storage Publication Make Publication Available 1 Retrievable Publication Perform Value-adding Services Value-added Services 2 Archive Securely 3 Long-term Archival Copy Libraries and Bibliographic Services Information Brokers National Libraries etc. Figure 13. Diagram A213 Archive and Index

26 version of 29-Sep-03 11:44 page A231 MAKE PUBLICATION AVAILABLE This diagram includes both the activity of making the paper publication available (placing it in the shelves of the library) and making the electronic version available. In both cases this is preceded by the longer term activity of securing subscriptions and access rights to the material, an activity which is even more visible today due to the large library consortia that negotiate bit deals with the large publishers. An additional value adding activity is the integration of the meta data about the publication in data bases which facilitates finding out about the existence of the publication. Secure Access Rigths and Subscription Local Demand for Publications Subscription or Pay per View Facility 1 Publication Paper Publication Make Paper Publication Available Retrievable Publication 2 Electronic Publication Make Electronic Copy Available 3 Meta Data of the Publication Integrate Meta Data into Search Services Searchable Metadata 4 Alerting Messages Figure 14. A231 Make Publication Available

27 version of 29-Sep-03 11:44 page A2313 MAKE ELECTRONIC COPY AVAILABLE There are at least two major mechanisms for making an electronic copy of a publication available doing this. Firstly this can be done through standard commercial services, which necessitate that the reader or normally the local university library has secured a subscription and makes the publication visible via the university intranet. A second possibility which partly bypasses this is if the author has sent a copy of the publication to an open access e-prints repository as exemplified by the Los Alamos preprints server for physics Subscription or Pay per View Facility Make the Publication Available Inside your Organisation Retrievable Publication Electronic Publication 1 Post the Publication on E-print Server 2 Figure 15. A 231 Make Electronic Copy Available

28 version of 29-Sep-03 11:44 page A2314 INTEGRATE META DATA INTO SEARCH SERVICES Traditionally commercial indexing services have dominated this function and libraries have paid subscriptions to them. Over the past years researchers have increasingly started to use general web search engines for trying to identify interesting publications. An effort to overcome the quality problems related to this is the definition of the Open Archives Initiative standard for tagging scientific content material on the web, which will enable dedicated harvesting search engines to maintain a much more focused data base of links to relevant publications. A by-product of the heavy use of IT for these purposes is the possibility or readers to subscribe to services, which based on the interest profiles they define, can send them alerting messages when something they might be interested in is published. Subscription or Pay per View Facility Meta Data of the Publication Manage Centrally Edited Bibliographic Index Searchable Metadata 1 Alerting Messages Manage Automated Search Engine 2 Manage Local Front End to Meta Data Services General Web Search Engines 3 Bibliographic Services Dedicated Harvesting Tools The Local University Library Figure 16. A 2314 Integrate Meta Data into Search Services

29 version of 29-Sep-03 11:44 page A232 PERFORM VALUE ADDING SERVICES In this overall category some party analyses the data to calculate citation indexes, impact factors etc., or write news bulletins about research results that practitioners can digest more easily. The three categories are not exclusive and just given here as examples. Publication Compile Citation Index Value-added Services 1 Perform Impact Factor Calculations 2 Summaries, Newswatch etc. Perform Synthesis and Analysis 3 Specialised Bibliographic Services Information Brokers Figure 17. Diagram A2133 Perform Value-adding services

30 version of 29-Sep-03 11:44 page A3 STUDY THE RESULTS This diagram structures the activities of the readers of scientific activities. Note again that from a cost per publication the activities of individual readers all over the world and in different time periods should be summed up. The find out about publication activity results in the output metadata of interesting publication (including the location from which a paper or electronic version can be retrieved). This output is used as the control of the retrieve publication activity. Finally the publication is read and the scientific information in question has been disseminated. Note that as a rule researchers self-archive interesting publications they have read either as paper copies or today increasingly as bookmarks or in a data base. Searchable Metadata Find Out About Publication Metadata of Interesting Publication Alerting Message 1 Retrievable Publication Retrieve Publication 2 Retrieved Publication Distributed Paper Copy Read Publication Disseminated Scientific Knowledge Self-archived Copy 3 Figure 18. Diagram A3 Study the Results

31 version of 29-Sep-03 11:44 page A31 FIND OUT ABOUT PUBLICATION This activity is rather difficult to split up into alternative parallel options (for each identified article one of these is used). On this first level there is a split into two generic categories. The first one is active search where the reader is pulling for information. The other option is push, where the reader receives a notification through some mechanism that something interesting has been published. Searchable Metadata Search for Publication Metadata of Interesting Publication 1 Alerting Message Be Alerted to Publication 2 Figure 19. Diagram A31 Find Out About Publication

32 version of 29-Sep-03 11:44 page A311 SEARCH FOR PUBLICATION The first modelled option of the pull variety is a traditional bibliographic data base search, for instance using key words. Other possibilities include more unstructured web searches using a general search engine or just browsing from one hyperlink to the other. Less and less the traditional method of physically browsing in library bookshelves are used. Searchable Metadata Do Systematic Bibliographic Search 1 Browse the Web Metadata of Interesting Publication 2 Bibliographic Services Browse Physically in a Library 3 Interesting Publication Web Search Engines Web Browser Library Shelves Figure 20. Diagram A311 Search for Publication

33 version of 29-Sep-03 11:44 page A312 BE ALERTED TO PUBLICATION Receiving a hint from a third party could be a hint from a colleague or a supervisor, or in today s world a hint from a software based alerting service. The important distinction is that the bibliographic search is triggered by the researcher himself (pull) whereas the hint is coming from the outside (push). Notice reference in other publication is very common way of finding out about an interesting publication. This is where it would be so convenient if all publications were for free on the Internet, because the retrieval would then only entail pressing a hyperlink. The last option remember existence of publication is less evident. Alerting Message Receive Alerting Message from Service or Colleague Metadata of Interesting Publication 1 Notice Reference in Other Publication 2 Alerting Service Remember Existence of Publication 3 Figure 21. A312 Be alerted to publication

34 version of 29-Sep-03 11:44 page A32 RETRIEVE PUBLICATION The basic split has here been made between the retrieval of a copy of a paper publication and one in digital form. Although the two activities modeled here may look straightforward, they might become rather complex in reality. If for instance the organization that the researcher belongs to doesn t subscribe to the journal in question, it might take quite a long time to obtain a copy of the article through some add on service for interlibrary loans or through a service for buying individual articles over the web. Many younger researchers have become lazy for retrieving paper-based copies at all, since so much is there on the web. Metadata of Interesting Publication Retrievable Publication Retrieve Electronic Copy Retrieved Publication 1 Retrieve Paper Copy 2 Order Copy 3 Ordered Copy of Publication Figure 22. Diagram A32 Retrieve Publication

35 version of 29-Sep-03 11:44 page A4 APPLY THE RESULTS In the same way as the breakdown of A1, Do Research, this diagram is more of a contextual nature, since no further breakdown of the technology transfer process is attempted. It tries to show how disseminated scientific knowledge can be transferred by several parallel mechanisms into better industrial performance, new products and services and eventually a better quality of life. One of these mechanisms is education and training, which results in better-trained professionals who go out into working life (i.e. medical doctors and engineers). There is a rather straightforward link between research and especially university education. A second mechanism is through commercial development work which translates research results into new products, services and working methods. A third mechanism is where practitioners read research publications and are directly affected by them. Industrial R&D Policies Disseminated Scientific Knowledge Teach Professionals Educated Professionals 1 Do Industrial Development Improved Tools and Practices 2 Improved Productivity and Quality of Life Apply Results Universities and educational establishments 3 Industrial Problems Companies Figure 24. Diagram A4. Apply the Results

36 version of 29-Sep-03 11:44 page CONCLUSIONS The use of a formal process modelling language for a purpose such as this was motivated by the personal experience of the main researcher, having earlier used this type of methodology for the modelling of the construction process as well as having led an international project investigating innovative process modelling methodologies. The initial experiences have been very positive. The studied process is by its very nature rather linear (contrary to for instance architectural design), which makes the modelling easier than for processes involving a lot of networking or iterative procedures. Also colleagues to whom the model has been shown have quite easily grasped the fundamentals of the IDEF0-notation and have been able to follow the logic of the model. The model in its current shape is not yet validated in its details, but has been discussed with a number of domain experts (publishers, librarians) with encouraging feedback. Based on these discussions and on the extensive literature review done as part of the SciX project it is the conclusion of the authors that this is the first time a formal process modelling methodology is used in this comprehensive way to model the system of scholarly communication. Publishers employ methods of a similar nature to study the workflows within their organizations, but the point here is to study the whole process, including the activities of libraries and readers, and to use the process model as a basis for determining the activities which will be studied more in detail as a part the cost modelling. The cost modelling will be a synthesis task of data from several different sources. To some extent web surveys will be used, in particular concerning the economics of open access journals and repositories and reader behavior. Research work of other researchers concerning the cost of certain activities in the model will also be used [for example Tenopir and King 1998, Odlyzko 1998]. The common denominator will be the to study costs per publication flowing through the system. In addition to the cost modelling the model could also prove useful in providing a roadmap showing the place of a number of different initiatives for increasing access to scientific publications, within the overall system of scholarly communication.

37 version of 29-Sep-03 11:44 page SCIENTIFIC PUBLICATION LIFE-CYCLE MODEL (SPLC): VALIDATING THE MODEL BY COMPARISON WITH CURRENT PUBLISHER COST STRUCTURES 3.1 CONTEXT The Scientific Publications Life-cycle Model (SPLC), part of SciX Project IST , is aimed at modelling the progress of a scientific publication principally a journal article from inception to consumption. This paper puts that progress in the context of a conventional academic journal publishing business: - the development and maintenance of a publishing programme; - the culture and organisation of scholarly and research journal publishing; - the cost structure of publishing - the role of editors and editorial boards; and - the marketing and distribution that underlie the activity. While their missions may vary, most publishers, whether commercial companies or non-profit society and university press publishers, operate and structure their publishing activities in similar ways. While this may be seen as a generalisation, their objectives are to meet the needs of their chosen clientele by building a list or portfolio of journals and to achieve a return on the investment made in the activity. While in the case of commercial companies that return is transparently in the form of earnings for shareholders, most non-profit publishers seek to cover their costs and earn a surplus that can be deployed to support the organisation s other activities. Primary journal publishing is focussed on the author and the reader. Journal publishing exists to serve the interests of the scholarly and research community to publish the results of research. In this sense, it is a supply-driven business. The journal market is peculiar and dysfunctional in that the reader in the faculty or corporation may select, recommend or even insist on the titles that are acquired, without having to bear the cost of acquisition, while the purchase is made by the library, which has the budget but is driven by the requirements of its readers. Price signals do not reach the ultimate consumer the reader. In such an environment, pricing is generally geared closely to the cost of producing the journal. Some publishers have occasionally taken the opportunity to enhance profitability, as they have found that the market will bear just about any price. The publisher does not know what the right price for the product is, because he knows that, whatever the price, libraries will come under enormous pressure to buy it. This paper describes the key features of journal publishing as it is currently practised, and does not attempt to evaluate the current system against the alternatives being explored within the SciX programme. They should be read with specific reference to the diagrams and hierarchies contained within A223 and A23.

38 version of 29-Sep-03 11:44 page THE COST STRUCTURE OF JOURNAL PUBLISHING The two principal factors that drive pricing decisions are the costs involved in publishing, and the circulation of the journal. These are crucial, as costs have to be recovered by amortising them across the circulation. Reduced circulation does not affect the number of papers published, but does drive up the cost of servicing each subscription. There is much misunderstanding about what costs are inherent in journal publishing. Publishers have not until recently found it necessary to articulate what they do. Publishing is about making products out of authors ideas. The work is detailed, time-consuming and requires accuracy. It is designed to serve authors by providing a range of functions that make the author s manuscript readable, polished and packaged for the reader, and marketed effectively to libraries and their readers internationally. Incoming manuscripts often require heavy editing in order to make them clear and readable. Being published also implies quality, and that quality control procedures have been applied to the work; in journal publishing this is peer review. They make the investment and bear the financial risk, based on their independent editorial evaluation. All of this costs money. It is important to realise that up to 80 per cent of the cost of publishing a journal is incurred in processes that are not dependent on the medium of output paper or online. Paper, printing and distribution are a small portion of the subscription price. The cost factors include: - Direct costs such as review and refereeing, editorial work (including the cost of editorial offices and secretarial assistance that institutions may not be willing to pay for), preparing illustrations and typesetting/data capture, and indirect costs such as subscription maintenance, marketing and author administration. These cost factors are fixed. - New journals serving new areas of research have to be funded from existing resources i.e. from the revenues earned from existing titles. A new title in the social sciences can build up losses of up to US$ 100,000 in three years, and not reach break-even until Volume 4 or 5. There have also been changes in the pattern of subscriptions. As publishers publish more articles and more issues an increase from 6.5 issues in 1975 to 8.3 in 1995, prices increased and subscriptions from individuals, priced lower than the library subscription, declined. They began to rely more on the library. Subscription numbers dropped and the fixed costs had to be spread over fewer copies, thus increasing institutional prices at a greater rate than inflation and size increases can explain. Online publishing has not reduced costs. Those same fixed costs still has to be recovered in revenue. The savings in paper, printing and distribution are more than offset by the need to build in additional functionality links to and from cited articles and A&I services, searching power and flexibility etc. and provide twenty-four hour availability and customer service. And most journals continue to have high fixed costs and a low number of subscribers.

39 version of 29-Sep-03 11:44 page ESTABLISHING THE IMPRINT The key to developing a publishing programme is the Commissioning Editor, or Publisher. He or she will be responsible for the acquisition of journals from universities and societies, and for managing the publishing process up to the publication date, and for the creation of a list of journals that complement each other and can be marketed together. As journal publishing is focused on the needs of editors and authors who can be a significant resource of contacts, mailing lists etc he or she should have a significant and influential role in devising the marketing plan. The cost of the Commissioning Editor is an overhead to be borne by the whole programme. It is important to establish an imprint under which journals are to be published. This is not because libraries or readers will remember the imprint or choose a journal because it is published under a particular imprint, but because it signifies the discipline(s) in which it publishes and the quality of the product to editors and editorial boards. While most authors and readers will choose the journal a brand in itself, to be nurtured and developed editors and journal proprietors choose the publisher (or imprint) in which they will have most confidence. The imprint provides the context in which: - the list of journals can be developed, e.g. a list in business and management, in educational research, or in civil engineering, architecture and the built environment, to take three examples. The Commissioning Editor is responsible for populating the list with appropriate titles. Initially this is likely to be through acquisition or by contracting with societies or university presses to publish their journals. It may also involve the launch of new titles as the Commissioning Editor spots a gap in the literature, or a new area of research in the field. - A format or house style for the journals can be developed. Standardization provides consistency in design, quality control and enhances marketing both in print and online. 3.4 THE JOURNAL EDITOR, EDITORIAL BOARD AND PEER REVIEW The Editor is the key figure in the development of the journal. As well as being an expert in the field probably an academic, but could be a practitioner - he or she should be efficient and effective in selecting and leading the Editorial Board, which functions under, and depends on the efficiency of, the Editor. He or she is the principal contact for authors and the face of the journal to the author/ reader community. The selection, support and maintenance of an international editorial board with an international membership comprising scholars and practitioners of repute are vital to attracting high quality papers to the journal. Authors look at the members of the editorial board and their institutional affiliations as an important guide to the quality of the journal.

40 version of 29-Sep-03 11:44 page 40 The maintenance of the Board and the relationship with the Editor is a key function of the Commissioning Editor. The Editor operates peer review and will receive submissions from authors, select reviewers, who may be members of the Editorial Board or outsiders, and makes the final decision on acceptance or rejection. The Editorial Board meets periodically usually once a year at an appropriate conference which most of the members will already be attending. Peer review is facilitated by the publisher (usually financially, by meeting the out-of-pocket costs of the Editor and the costs of editorial board meetings), although the process of selecting reviewers and making the final decision on acceptance is very much an academic process. All academic and professional publishers use reviewers to assess the theory or practice the author has described in the manuscript. While in book publishing this is an informal process, in journal publishing it is formalised in peer review: - Articles are submitted to the journal Editor. They are often encouraged to submit their manuscripts electronically (either as attachments or on disc), in order to utilize already captured and copy-edited keystrokes, save time in a tight publishing schedule, and save money. The Editor selects peers in the field i.e. academics or other experts in the field and usually sends the manuscript to them for a referee s report under conditions of anonymity (i.e. double blind peer review ): - The author s name is withheld, so that the reviewer has no idea who has written it, and therefore has to be completely objective about the work, without any personal feelings about the author compromising his or her judgment. - The reviewers names are withheld from the author for similar reasons. - At least two, and usually three, referees are used. - The process is often an interactive one, where referee s suggest amendments to the original, and the author then revises his or her manuscript and re-submits it for further review. It should be noted that the process is not designed to discover whether the referee likes the work or its conclusions; its purpose is to assess the soundness of the methodology. This process is well described in Diagram A The publisher is responsible for facilitating this process and acting as an agnostic guardian of its integrity. 3.5 THE PRODUCTION EDITORIAL PROCESS AFTER ACCEPTANCE Copy editing the manuscript, once accepted for publication, is the responsibility of the publisher, and is designed to re-check and correct the accuracy of the information and make the work stylistically consistent and grammatically correct. Once the edited manuscript has been typeset, the typesetter will supply proofs. Proofs will be read to ensure that no printer s errors have crept in. It is usual to supply proofs to the Editor and to the author so that they may check the work prior to publication. Most publishers will allow authors to correct up to 10 per cent of the whole. This allows for minor revisions and an improvement to the whole, without inviting a wholesale rewrite.

41 version of 29-Sep-03 11:44 page 41 Most publishers sub-contract copy editing and proof reading to freelance editors. Freelancers should be selected for their copy editing skills and also their familiarity with the subject matter of the journal. Knowledge of the discipline is clearly important in medicine, mathematics, physics and other STM areas; in business and management, a degree in and general knowledge of the social sciences should be sufficient. It is important both for the publisher and the Editor to have confidence in the freelancer. Most freelancers therefore work on the same titles and often maintain their association with the journal for years they become almost an extension of the publisher s permanent staff. They are usually paid by the hour; there is a recent tendency to pay freelancers by the issue or by the volume. 3.6 MARKETING JOURNALS Journal marketing involves reaching the academics and professionals who might be interested in the journal. In most of the discussions about institutional repositories, university-based selfpublishing and authors acting as publishers posting their work direct to the Web, little or no attention is paid to the central importance of sales and marketing activity. Unless it is marketed, a journal or other resource may well remain inaccessible simply because potential readers have never heard of it. The key to success is to reach potential readers by targeted communications, in both the academic and business communities, by: - Direct mail, using targeted mailing lists. Society membership and conference delegate lists are important here. These should be complemented by mailings to academic and special libraries (preferably librarians by name). - Each journal should have a simple brochure/flyer with an order form for library and personal subscriptions that lists the members of the editorial board, describes the editorial platform of the journal, lists the contents of a recent issue, and provides information on submissions to authors. - Ensure that each journal is indexed by all appropriate abstracting and indexing services. Particular attention should be paid to ISI and to the appropriate A&I services for each discipline. The appropriate abstracting and indexing services for each journal can be identified by consulting EBSCO s The Index and Abstract Directory and by checking where comparable journals are indexed. Each of the appropriate services should then be contacted for information on how to submit titles to them. - Establish an opt-in online contents alerting service to advise readers of newly available articles in the subject areas they register. - Create a web site or web page for each journal that will provide readers with information on the editorial board, editorial platform, instructions for authors, links to online issues, pricing information and online ordering facilities.

42 version of 29-Sep-03 11:44 page 42 - Set up arrangements for exchange advertising in comparable journals published by other publishers, in order to reach their readers and vice versa, at no cost other than the preparation of artwork. - Establishing mandates with the Reproduction Rights Organisation (e.g. the CCC in the USA, or the CLA in the UK, and thence with all the major Reproduction Rights Organizations worldwide and licenses with appropriate document delivery organizations (notably BLDSC, CISTI, Infotrieve) in order to secure incremental income from authorized use. - Licensing aggregators and re-publishers that compile collections of journal literature on databases sold for undergraduate use in universities, and to vocational and tertiary colleges, schools and public libraries, thus extending readership and generating incremental royalty income. The publication cycle from acceptance of a paper for publication to dispatch of printed issue is typically much shorter than for books. As systems like Quark Express and PageMaker have enabled publishers to automate the data capture/typesetting process, typical publishing cycles have been shortened from 16 to 12 weeks. As the print production component of that cycle may be three weeks, it should be possible to post articles online as soon as the final proofs have been approved and the issue passed for press (i.e. authorized for printing). Many publishers make a feature of uploading articles online in advance of print publication. This feature is attractive to potential authors, as well as being of benefit to readers; it is an important selling feature. In Diagram A23, the SPLC indicates that libraries play a principal role in both archiving and indexing journal literature. This is not strictly true: - The library prepares the catalogue, either by creating catalogue records itself or by buying them in from library utilities such as OCLC. It should also be noted that metadata is becoming available from the publisher as part of the service to libraries: e.g. Elsevier provides XML catalogue metadata as part of ScienceDirect for uploading into the library catalogue; - The library is ultimately responsible for archiving; nevertheless, libraries look to publishers to make appropriate arrangements for archiving electronic publications. Archiving in the online environment has become a joint effort, with publishers and libraries entering into collaborative arrangements for archiving online content (a recent example is the agreement between the Koninklijke Bibliotheek, the National Library of the Netherlands, and Kluwer Academic to archive the electronic publications of Kluwer Online: some 670 journals and 600 ebooks). - Indexing is undertaken by abstracting and indexing (A&I) services which form an integral part of the publishing industry. In many cases, publishers publish both primary research journals and leading A&I services. Examples include Elsevier (A&I includes Excerpta Medica, Ei Compendex), American Psychological Association (PsychInfo), American Chemical Society (Chemical Abstracts), Emerald, etc. The submission of journals to appropriate A&I services is a vital part of the marketing undertaken by publishers, as such services are part of the navigational infrastructure used by readers.

43 version of 29-Sep-03 11:44 page 43 In the online environment, the division of labour has become very blurred so far as indexing and archiving is concerned, and Diagram A23 needs considerable amendment to reflect current reality. It is clear that many academic libraries themselves are failing to catalogue open access journals: see postings to the liblicense-l listserv from June The reasons for this significant failing revealed by these exchanges include the following: - Subject librarians or library selectors do not choose titles for collection development policy reasons; - Available records supplied by open access publishers like BioMed Central are very basic and would require significant effort to augment, or funds to acquire a complete record from a bibliographic utility; - Journals are catalogued when the first issue is received, or when the first volume is sent for binding, when the order is placed, or when an invoice is processed many libraries prepare electronic journal catalogue records from lists furnished by the Acquisitions Department. Open access journals do not trigger any of these prompts to the cataloguing department! - Open access journals are still seen by many librarians as unstable, or unreliable to access; Apart from the ISSN, most open access journals are not identifiable by the usual reference devices used by libraries and publishers (SICI, DOI etc.) and are not linked either to abstracting and indexing services or to citations via CrossRef. Establishing these links is a function largely undertaken by publishers. A large number of open access journals are not marketed at all. With the single exception of BioMed Central s titles, they are generally published and operated by university departments - some by learned societies - who operate on the assumption that interested readers will simply find the content. This assumption underlies some of the commentary in the SPLC report, with its emphasis on metadata harvesting tools and alerting services. Most experienced journal publishers would emphasise the importance of push marketing both to libraries and to readers; it cannot be assumed that even specialists in the discipline will know about a journal simply because it is available on the Internet. 3.7 SUBSCRIPTION MANAGEMENT AND DISTRIBUTION Warehousing, shipping, customer service and subscription liaison facilities must be laid on. Journal publishing is a subscription-based cash-with-order business, with most subscriptions being placed by subscription agents. Publishers normally give a discount to agents, averaging 5 per cent, for institutional subscriptions. Nevertheless, many institutional subscriptions, and all personal subscriptions, are placed direct. This requires a good subscription management system. Many publishers outsource this activity to specialist service companies.

44 version of 29-Sep-03 11:44 page 44 Scholarly journal publishers generally set their prices for the following year in May and June. This is driven by subscription agents need to have up-to-date price files well before (i.e. by the end of July) they issue renewal reminders in September for the following year. One of the perverse results of this is that the publisher is pricing for next year without having a complete picture of the subscription numbers and revenue for the current year. 3.8 OUTSOURCING AND FREELANCERS Journal publishing is a process that comprises many functions that can successfully be outsourced: - Copy editing and proof reading - Typesetting - Printing - Online production and hosting - Web site development - Subscription management - Print distribution - Preparation of brochures, flyers and advertisements - Direct mail marketing There are companies that specialize in providing a full publishing service to journal publishers from receipt of manuscript accepted for publication, through copy editing, typesetting, proof reading, printing and binding, and the provision of digital files for online hosting to the distribution of printed copies, relieving the publisher of the need to establish more than a virtual publishing unit, with energies directed at finding journal to add to the programme and managing effective marketing. Nevertheless, such activities have to be planned and costed, and the processes managed, by the publisher. 3.9 PRODUCT DEVELOPMENT: BUILDING THE JOURNAL LIST The journal list can be developed in three ways: purchase of titles from existing owners, securing a publishing contract from the journal owner or launching a new title where research patterns change, for which there is no existing journal provision. It should be noted that this often involves a long process of courtship and relationship building before a deal can be negotiated and completed. Today, the scope for starting new journals in an economically stressed environment is limited. The primary mode of list building is individual title acquisition. The principal targets are

45 version of 29-Sep-03 11:44 page 45 established, and the owners identified. These are likely to be single titles, or lists of fewer than three titles that are: - owned by universities and published from university departments outside the university press or other formal publishing arrangements it may have; or - owned and published by small learned or professional societies; or - individual titles listed for disposal by other publishers; individual journals are often bought and sold, either because they do not meet the financial criteria set by the current publisher, or they do not fit the list usually because they lie outside the principal disciplines on which the current publisher is concentrating. There is a great deal of competition for journals when they come up for sale. The usual valuation methodology is based on annual subscription revenue. Most publishers currently expect to pay between 150 and 200 per cent of the current subscription revenue as a purchase price. Smaller journals that require investment in re-design and marketing, or that require the appointment of a new Editor and/or Editorial Board, or that are losing money, may be available for less. Acquiring the right to publish under a contract with the owner should not be ignored. While outright ownership is the priority in building capital value in the business, negotiating the publishing rights to titles owned by societies or professional associations can generate significant and positive cash flow. Some publishers depend almost wholly on publishing society journals (e.g. Blackwell Publishing does not own most of the titles it publishes). Most journal publishers, including Elsevier, Wiley, Taylor & Francis, publish a mixture of titles that they own and society journals. In setting up a journal publishing programme from scratch, society journals add reputation and authority to a growing list COMMENT ON THE SPLC MODEL In view of the commentary on the journal publishing business set out in above, the following should be noted: - First, a general observation must be made. The SPLC model focuses on the process of authoring, publishing, retrieving and utilising the journal article itself. Except in the case of indexing and archiving (see 3.6 above, referring to Diagram A23) it is a very detailed and faithful representation of the processes involved. However, it gives insufficient weight to the publishing organisation s infrastructure that supports the article (see Diagram A2231) and, if taken at face value and out of the context of that infrastructure, may mislead and oversimplify the organisation and management functions that are necessary for the processes to work. Furthermore, it ignores the wider environment in which scholarly publications operate secondary publishing (see comments on Diagram A231) that provides navigational tools to readers, review journals and the costs incurred by the publisher in ensuring that its journal content is future proof.

46 version of 29-Sep-03 11:44 page 46 - Diagram A2232 is misleading; taken together, the diagram and the accompanying text imply that the need for marketing to generate demand for subscriptions is reduced for open access journals. However, marketing includes the broader range of activities described in 3.6 above. There is little evidence of active marketing of open access journals by the university departments or societies that are responsible for most of the titles listed in the Lund Directory, with the exception of BioMed Central, which itself is a for-profit publisher managed by experienced publishing professionals. - The hierarchy and Diagram A231, and the accompanying text, state that the need for commercial indexing services may reduce and that the use of general or specialist search engines will increase. This remains to be proven. For most academic researchers, it remains true that ISI s Citation Indexes remain the indexes of choice; subject-based indexes are widely used to identify and locate relevant articles. Open access journals are only just beginning to be indexed and catalogued see 3.6 above. A quick review of some journals in the Directory of Open Access Journals hosted by Lund University ( indicates that few have adopted the Digital Object Identifier (DOI) or CrossRef (the citation-linking resolver based on the DOI), all of which are common standards within the conventional journal publishing environment that are integral to the process of cataloguing and navigation within the journal literature. Moreover, few are providing reliable cataloguing data to libraries to enable them to upload a full record into the library catalogue THE COSTS OF THE CONVENTIONAL PUBLISHING MODEL The prima facie advantages of open access publishing consist in the following: - The open access model transfers the cost of publishing from the subscriber the market in which the reader operates to the supplier the community in which the author/researcher operates. As journal publishing has, ever since the founding of the first recognisable scientific journal, Philosophical Transactions, by Henry Oldenbourg in 1665, been driven by the needs of scholars to publish the results of their research, the move to transfer the costs of the system to the point of supply is logical. - Open access makes access available to all at no charge to the reader; it can eliminate the restrictions on access that disenfranchise readers who cannot afford to subscribe or whose library budgets restrict the journals to which they can subscribe. - It eliminates costs within the conventional subscription-based system, in subscription management and order processing and access control. This cannot be said of those journals that apply open access to the online version but still require pre-paid subscriptions for printed editions. Essentially, these costs can be avoided only if the journal is online only. For the purposes of validating the costs revealed by the SPLC open access journal publisher survey, and of comparing the cost structure of the open access model with that of the conventional publishing model, it should be noted that open access does not of itself affect the processes of submission, peer

47 version of 29-Sep-03 11:44 page 47 review, preparation, publication, marketing and delivery, or the costs involved in operating those processes. Model cost structures have been constructed from data supplied by three publishers, and from a study undertaken for the Association of Learned and Professional Society Publishers (ALPSP): Dryburgh Associates: The Costs of Learned Journal and Book Publishing: A Benchmarking Study for ALPSP, London Two large journal publishers (one commercial and one non-profit), a small commercial journal publisher, and two small non-profit publishers (one society and one university press) were consulted about the costs incurred in publishing their journals. Four of the five publishers consulted were reluctant to provide real data on named journals for publication, so in those cases a model was constructed based on typical operations and costs incurred; the model operating costs, corresponding with the SPLC questionnaire, are set out in Appendix 1. Appendix 2 provides a complete profit & loss statement for each one, set out in a conventional format. It should be noted that: - in the case of the large commercial publisher (which publishes some very large research journals as well as smaller titles), it was thought appropriate to model a monthly journal that would be typical of the majority of titles in its list. This particular publisher is a large international commercial house with a substantial list. In this case core revenue was derived entirely from institutional library subscriptions; this publisher s revenue from advertising and from individual subscribers is negligible, and was therefore discounted in creating this model. - In the case of the large non-profit publisher (a substantial university press) a similar approach of modelling a typical monthly journal was adopted; individual and/or society member subscriptions are an important part of the revenue. - in the case of the small non-profit society publisher, the model was based on its principal title. Revenue is derived from society members and institutional subscriptions. A small amount of advertising is sold, and included in the revenue total, but is not significant. - In the case of the second non-profit publisher, a small university press with three journals, the largest of its journals was chosen. Its revenue includes a significant element of page charges ( 34 for first 8 pages and 152 per page thereafter) and colour plate charges, and some advertising and reprint revenue, as is typical of journals in the biomedical sciences. - in the case of the small commercial publisher, the revenue model is different. Its principal title is a peer-reviewed medical title, which is distributed free to 3,500 medical libraries world-wide; the online version is available free of charge to them. Its editorial and production processes are the same as subscription-based journals, but its revenue is derived from author charges and from the sale of article reprints to pharmaceutical companies. While the journal is available on subscription to libraries not on the free circulation list, it is in essence an open access title for the core readership in medicine. - the ALPSP Benchmarking Study examined specific academic publishing operations. It was based on ten publishers: eight were non-profit, and nine published in science. The sample

48 version of 29-Sep-03 11:44 page 48 ranged from small to large publishers. While three areas of general costs included in the SPLC questionnaire to open access journals were not addressed, the remaining issues were covered and provide useful benchmarks against which the publishers' responses, and those of SPLC questionnaire respondents, can be measured. The overview of revenue and general costs per journal title can be summarised as follows: a. Journal characteristics Description (EUR 000) Large Large non-profit Small Small non-profit Small non-profit commercial commercial (society) (university press) Frequency Pages published per year 1,200 1, ,000 4,796 Articles published per year Average pages per article Institutional subscriptions ,900 2, b. Journal revenue Description (EUR 000) Large Large non-profit Small Small non-profit Small non-profit commercial commercial (society) (university press) Revenue per issue Revenue per title , ,793.2 c. General costs Description (EUR 000) Large Large non-profit Small Small non-profit Small non-profit commercial commercial (society) (university press) Admin/financial overhead ,823.0 IT infrastructure Planning issues (origination) Marketing to authors incl. in readers incl. in readers 15.0 incl. in readers incl. in readers Marketing to readers Other costs 15.0 incl. in Admin

49 version of 29-Sep-03 11:44 page 49 At the article level, a comparison can be made with the ALPSP Benchmarking Study: Description (EUR 000) Large Large Small Small non-profit Small non-profit ALPSP commercial non-profit commercial (society) (university press) benchmark Editor / review process Reviewer s work Technical editing Placement on the Web The broad headings under which this data was collected are open to a variety of interpretations. Experience in journal publishing indicates that both formal accounting policies and management accounting practice varies widely from one publisher to another. This is not to say that any of them are wrong, but only to emphasise the problems of precise definition and the differences between publishers in respect of what each regards as key management data; for instance: - IT infrastructure and Placement on the Web between them cover publishers that own and operate their own hosting systems and those that outsource hosting to a third party; - Planning of issues and Technical editing encompasses typesetting, page layout, data creation for online services, copy editing, technical editing, proof-reading, editorial management etc. It is a matter of judgement under which heading such costs should be placed. Nevertheless, the publisher surveyed and the ALPSP Benchmarking Study show comparable costs at the article level in respect of: - Planning of issues (including origination), ranging from 287 to 578; the median is 481, compared with the ALPSP benchmark of 540; - The costs involved in supporting the Editor and the peer review process, including payments to reviewers, where practice varies, were remarkably consistent, ranging from 588 to 687, with a median of 615. The ALPSP Study shows a cost of only 108, which is significantly different from the publishers surveyed; - Technical editing costs range from 287 to 632 with a media of 479, compared with the ALPSP benchmark of 346; It should be noted that most publishers integrate their information to authors and to readers in a single campaign. Most publishers typically place information for contributors, details of the Editor, Editorial Board and the aims and scope of the journal together with subscription information, citation and review information and promotional text together in the same information leaflets and web sites. Few publishers distinguish between marketing to authors

50 version of 29-Sep-03 11:44 page 50 and marketing to readers after all, they all belong to the same academic and research community in most disciplines. This lies behind the inability of most of the publishers surveyed to distinguish between the two types of marketing; it is an unreal distinction. The ALPSP Benchmarking Study provides an analysis of the aggregate data from all participants, breaking down major cost items as a proportion of total costs. The data collected from the five publishers surveyed have similarly been compared with the ALPSP benchmarks; problems of definition and cost allocation mean that each of the ALPSP categories is only roughly comparable with those used in the SPLC model: Description Large Large Small Small non-profit Small non-profit ALPSP commercial non-profit commercial (society) (university press) benchmark Refereeing 22.7% 27.4% 7.6% 24.3% 6.3% 22% Editorial / typesetting 18.9% 11.5% 7.5% 16.6% 6.7% 33% Print production/distribution 14.1% 31.4% 51.9% 30.9% 21.7% 23% Sales & marketing 7.6% 2.0% 14.4% 2.5% 1.8% 13% Promotion to authors not separated not separated 2.2% not separated not separated 2% It should be noted that these percentages vary widely because of the variation in print production and distribution costs, which assume a much higher proportion of total costs in the case of the large non-profit, the small commercial and the small non-profit (society) publishers titles, which have high circulation and require long print runs; this in turn drives down the other percentages.

51 version of 29-Sep-03 11:44 page SCIX SURVEY ON THE CHARACTERISTICS AND COST STRUCTURE OF OPEN ACCESS SCIENTIFIC REFEREED JOURNALS 4.1 METHODOLOGY AND RESEARCH SETTINGS The survey was conducted in May 2003 as a web survey. The initial contact was an sent to 310 editors of open access scientific refereed journals, where they were asked to participate in the study and to fill in the questionnaire designed as a web page. The answers were directed to an address at Hanken. The list of open access journals and the addresses to editors or the journal were collected earlier during the SciX project and the list has been used as a population in this survey. 10 error messages of incorrect address were received so the total number of s sent out was answers to the questionnaire were collected; thus the percentage of answers amounted to 20% of the population. 4.2 THE QUESTIONNAIRE The questionnaire contains three parts. The first part (Q ) provides general information about the journal and the publisher. The second part (Q ) provides cost information. Q2.3 on general costs for publishing and Q2.4 on per article costs. The third part (Q ) provides answers to 6 statements regarding publishing, queue time and rate of publishing. The fourth part (Q ) concerns intellectual property rights, such as copyright and copyright transfers. The fifth part is general comments by the editors on the questionnaire and on open access publishing. 4.3 THE RESULTS Part I General characteristics and information on open access journals Q 1.1 Title of the journal. This question was optional and is not reported in the study. Q 1.2 Subject area The subject areas of the participating journals represent a wide span of scientific work. The participants of the survey represent and correspond roughly with the subject areas of the initial list of open access online journals.

52 version of 29-Sep-03 11:44 page 52 Table 2. Broad subject (1.2) Physical sciences 10 Humanities 14 Social sciences 22 Technical sciences 3 Life sciences 10 Sum 60 Table 3. Narrow subject (1.2) Statistics 2 Mathematics 6 Physics 2 Psychology 7 Literature 1 Philosophy 1 Linguistics 3 Religion 1 Anthropology 1 History 1 Media, communication 5 Economics, business 5 Library and information 2 Gender studies 1 Education 6 Inter-disciplinary social sc. 1 Safety science 2 Material and construction 3 Medicine 5 Biology, paleontology 5 sum 60 Q 1.3 The question if online publishing has changed the traditional print feature of publishing issues towards the publishing of individual articles was the motive for question journals or 55% publish issues while 25 journals or 42% publish individual articles. 2 journals or 3 % chose not to answer this question.

53 version of 29-Sep-03 11:44 page 53 The results indicate that open access journals mostly publishing only an electronic version tend to go for a shift towards individual article. Q 1.4 The number of published articles in The large spread of the answers to this question was anticipated. The minimum number of published articles was 3 and the maximum 110. The average number of articles = 20 and the median = 16. If we compare these numbers to the ones presented in John Cox current study (for the SciX project) with commercial and non-commercial publishers (both large and small) the number of articles is considerably higher from 100 to 872 articles. Thus, a conclusion is that most open access journals publish relatively few articles compared to traditional publishers. published articles 2002 Number of observations Number of articles Figure 25 Published articles 2002 (1.4) Q 1.5 Characteristically open access journals have a large amount of funding from the publisher s own institution. In the comments to the questionnaire many of the editors mentioned volunteerism and it s a part of my normal work as other sources of funding. Grants are also one important source of funding. In some cases professional institutions provided funding for journal publishing. Advertisements, member fees and author charges were the source for main funding in few cases. The participants were able to give more than one main source for funding. Table 4. Main funding (1.5) grants 11 author charges 3 member fees 4 advertisements 1 publisher s institution 40 Other funding 14

54 version of 29-Sep-03 11:44 page 54 Main sources of funding of open access journals other funding 19 % grants 15 % author charges 4 % member fees 5 % advertisments 1 % publisher's institution 56 % Figure 26. Main sources of funding of open access journal (1.5) Q 1.6 We were interested to know whether open access publishers used the Open archives initiative standard for metadata. The standard promotes scholarly communication by enhancing accessibility to repositories. The answers indicate that the OAI initiative was not used or familiar to the participants in the survey. Table 5. OAI tagging (1.6) Uses OAI tagging 4 7 % Does not use OAI tagging 24 40% OAI tagging is not familiar 32 53% Q 1.7 The weight of indexing services to attract readers and promote the standard of a journal is generally known. The general indexing service ISI with the citation indexes on both author and on journal level is highly regarded among researchers. 6 (10%) of the open access journals participating were indexed by ISI. Subject-based indexes are also very much in use by readers to identify or locate articles. Of the participating journals, 34 (57%) were indexed in one or more subject-based indexes.

55 version of 29-Sep-03 11:44 page 55 Table 6. Indexing in general and subject-based indexes. (1.7) General indexes ISI 6 Ulrich s? Subject-based indexes Mathematical Reviews 5 Zentralblatt für Mathematik 4 Physical abstracts 1 INSPEC 3 PsychINFO 5 ERIC 3 Chemical abstracts 4 PubMed 2 Medline 1 Embase (biomedicine) 1 BIOSIS 1 ABI / Inform 1 Econlit 1 SIRC (Sports information) 1 Gale 2 ISA (information science) 1 LISA (library and information science) 1 PsychLit 1 Communication abstracts 1 Religion Index One 1 Religious and Theological Abstracts 1 MLA (language and literature) 1 LLBA (Linguistics and Language Behavior Abstracts) 1 Q 1.8 The accessibility to Open Access journals is promoted through free directories on the Internet. In May 2003 a new directory DOAJ was launched. The Lund University in Sweden provides the directory service. The participants in the survey gave their opinion on 1) whether their journal was listed in any directory, 2) whether they were familiar with the DOAJ directory of the Lund University. 30 participants, half of the population informed us that their journal was listed in a directory of open access journals. 20 participants informed that their journal was not listed in any directory of open access journals. 10 participants did not answer this question. 17 participants were familiar with the Lund directory, DOAJ. However 35 informed that they were not aware of the service, 8 did not answer the question.

56 version of 29-Sep-03 11:44 page 56 Table 7. Indexing in public directories of open access journals (1.8) Any directory of OA journals Lund DOAJ Yes No No answer Yes No No answer % 33% 17% 28% 58% 14% Q 1.9 Regarding the question on how the journal handles the long-time preservation of articles, the participants could choose from 4 given alternatives 1) own arrangements, 2) joint with library, 3) national library, 4) disciplinary-based repository, or they could provide some other alternative of their free choice. Long-term preservation Other (mostly mirror sites) 13 % Disciplinary-based repositories 9 % National library 12 % Own arrangements 58 % Joint with library 8 % Figure 27. Long-term preservation (1.8) Table 8. Long-term preservation (1.8) Own arrangements 46 Joint with library 6 National library 9 Disciplinary-based repositories 7 Other (mostly mirror sites) 10

57 version of 29-Sep-03 11:44 page 57 Part II Cost information Since open access publishing today is presumed to be volunteerism to at least some extent, we knew that exact cost data covering the whole publishing process would probably be very hard to get from all participants. The difficulties in providing cost data is also mentioned and regretted by several of the participants in the free comment section. Editing a journal is considered as part of the ordinary work at a university and a good portion of idealism is also true for carrying out the editorial work. The participants had three options for giving cost data; they could use one or all three. They could give the information as direct expense numbers or as budgetary cost. As an alternative they could give the approximate time spent on a task. The last alternative was used in most answers. It is perhaps also the most relevant measure and best reflecting the characteristic of the work of publishing open access journal. The editor puts in a lot of his own time and this is not calculated as direct or budgetary cost. Instead of giving information on individual tasks such as administration, marketing etc. the participants could give an approximation of a total direct expenses or budgetary cost for running the journal or an approximation of the total time spent running the journal. A more detailed analysis of the cost data is provided in the Section: Cost analysis of the scientific publication life-cycle model. Q 2.1 The first question in this part concerned whether the work as editing a journal is expressed in a yearly budget. 15 participants or 25% answered that they had a budget. 42 or 70% reported that they did not have a budget. 3 participants or 5 % did not answer this question. Q 2.2 Mostly reported currencies for the cost figures in the survey are US Dollars and Euro. In the report the currency is Euro. 1 US Dollar = 0,87 ; 1,00 = 1,44

58 version of 29-Sep-03 11:44 page 58 Table 9. Q 2.3 General costs / year Task direct expenses time spent budgetary cost Aver. Median Aver. Median Aver. Median Administration 3058 ** h 50 h 8078 * 0 IT-infrastructure 1770 ** 0 50 h 40 h 836 * 0 Planning issues 266 * h 50 h 497 * 0 Marketing to authors 182 ** 0 30 h 20 h 424 * 0 Marketing to readers 17 * 0 14 h 3 h 291 * 0 Other 1683 * 0 88 h 0 h 2777 * 0 SUM h 163 h direct expenses time spent budgetary cost Aver. Median Aver. Median Aver. Median Total cost and time 6367 ** h 250 h 6125 * 0 * less than 5 observations indicate a value > 0 ** less than 10 observations indicate a value > 0 Table 10. Q 2.4 Per article cost direct expenses time spent Aver. Media Aver. Median n Editor's review 56 * 0 58 h 10 h Peer review 2 * 0 40 h 6 h Technical editing 93 ** 0 41 h 4 h Placement of article 161 ** 0 10 h 2 h Sum h 22 h

59 version of 29-Sep-03 11:44 page 59 Part III Questions about submissions, reviewing, queue for publishing and the rate of published articles in percentage of submitted articles. Statements: Q Most submissions are by the authors on their own initiative Agree 52, Disagree 7, No answer Most submissions are by the authors on their own initiative No answer 2 % Disagree 12 % Agree 86 % Q We mostly invite authors to submit articles Agree 11, Disagree 46, No answer We mostly invite authors to submit articles No answer 5 % Agree 18 % Disagree 77 %

60 version of 29-Sep-03 11:44 page 60 Q Most submissions are to special issues in the journal Most submissions are to special issues in the journal No answer 7 % Agree 15 % Disagree 78 % Agree 9, Disagree 47, No answer 4 Q It's easy to receive submissions Agree 39, Disagree It's easy to receive submissions Disagree 35 % Agree 65 %

61 version of 29-Sep-03 11:44 page 61 Q Most received submissions maintain a high standard Agree 33, Disagree 26, No answer Most received submissions maintain a high standard No answer 2 % Disagree 43 % Agree 55 % Q It's easy to get reviewers to produce quality reviews in time Agree 26, Disagree 33, No answer It's easy to get reviewers to produce quality reviews in time No answer 2 % Disagree 55 % Agree 43 % Q 3.2 The time for the article to go through the review process + queue time before the final article is published, indicates the time from submission of an article to the publishing. The average is 5,1 months; the median is 4 months. Q 3.3 The rate of acceptance of submitted articles for publishing. The average is 47,4 % and the median is 50%

62 version of 29-Sep-03 11:44 page THE REVIEW PROCESS AND THE QUEUE TIME Q 4.1 The journal publisher requires a formal copyright transfer. 19 answers or 31% report that they require a formal copyright transfer. 41 answers or 68% report that they do not require a formal copyright transfer. 68% of the journals have a copyright statement on their web site, 31% again report that they do not (Q 4.2).

63 version of 29-Sep-03 11:44 page A PROFILE OF AN OPEN ACCESS JOURNAL, BASED ON SURVEY RESULTS AND FREE COMMENTS BY THE EDITORS. The form of the journal The open access journal is still mostly in the form of traditional issues but there is a transition to other options, such as publishing individual accepted articles. The median number of published articles is 16 / year, which indicates that the growth and future of the journal is depending on the steady flow of good papers. The rate of acceptance for submitted articles is around 50% and the processing time + queue time is 4 moths, which compared to traditional journal publishing is very fast. Most articles are submitted by the authors on their own initiative, the editor seldom invites authors to submit. The journal is sometimes but not regularly publishing a special issue to attract submissions. Generally the editor finds it easy to receive submissions, and they also keep a rather high standard. However, one problem for the editor is to get the reviewers to produce quality reviews in time. The editor also thinks that a formal copyright transfer is not needed but the journal has a copyright statement on the web, protecting the author s moral rights to the article. Sources of funding The main sources of funding is the publisher s own institution and the editor does not count the hours spent on editing the journal since it is a voluntary work inspired by his/her own research interests. The editor sees him/her self as a pioneer, and would very much like to inspire other researchers to change their publishing behavior towards open access publishing. However, standards, such as Open archives initiative standard for metadata is relatively unknown and not in use. Indexing of the journal The measure of the quality of a journal is today mostly based on citation indexes on a journal level. The ISI index Journal citation report is the most comprehensive today. However, the relatively slow and strict procedure for being accepted in the ISI index is troublesome for open access journals. Relatively few open access journals are therefore indexed in ISI. The journals also benefit from being indexed in subject-based indexes like PsychINFO and Mathematical Reviews and the Zentralblatt für Mathematik and many others. Being listed in directories of open access journals also helps the public and libraries in finding articles and journals. To handle the long-term preservation of articles the editor is relying on own arrangements but he/she is considering mirror sites. The open access journal lacks a yearly budget, partly because only direct expenses are counted and the editor does most of the editorial work for free, sometimes assisted by a student. The editor very often sees that editing the journal is part of his/her normal work as a professor, who is trying to promote his area of research. Copyright The open access journal does not usually require a formal transfer of copyright by the author. The author is therefore free to publish the article also on the web on his personal homepage or in an institutional archive. This possibility promotes the visibility of the research results on several media and forums. A copyright statement on the journal s web page again often includes protection of the author s moral rights, for example to be cited correctly.

64 version of 29-Sep-03 11:44 page COST ANALYSIS OF THE SCIENTIFIC PUBLICATION LIFE- CYCLE MODEL: COSTS FOR PUBLISHING THE RESULTS 5.1 THE OBJECTIVE OF THE COST STUDY Cost studies related to the process model presented in the Scientific publication life-cycle model (Björk & Hedlund 2003) aim at presenting the costs to society of the activities in the model. In this part costs for the activity publish and disseminate the results" are presented. We aim at answering the question whether open access publishing as a new form of publishing differs in cost structure from traditional publishing. In Figure 28 the activities are presented at the top level of the hierarchical model where the whole life-cycle is seen as consisting of four separate stages, "do research", "publish and disseminate the results", study the results" and "implement the results. Scientific Method Publishing Practice Appointment & grant decisions Industrial R&D Policies Scientific Problems Perform the Research Reading Habits New Scientific Knowledge 1 Performance measures Evaluate the researcher Publish and Disseminate the Results 4 2 Funding and Academic bodies Costs for publishing and disseminating of the results Retrievable Study the Results Publication 3 Disseminated Scientific Knowledge Improved Productivity and Quality of Life The Researcher The Researcher Implement the Results The Publisher Libraries, indexing services etc. Readers 5 Industrial Problems Society and Industry Figure 28 Cost study presented in the framework of the scientific publication life-cycle model

65 version of 29-Sep-03 11:44 page 65 The optimisation of cost for the whole process is interesting from the point of view of society, which to a large part is financing the production of research results as well as the publishing and use of research publications. The production of research results is probably the most expensive part, but the one least affected by the reengineering efforts facilitated by the Internet. These costs are therefore relatively stable and unchanged from the "as-is model" presented in deliverable 1 to the "to-be model" presented in deliverable 2. On the other hand the activities "publish and disseminate the results" and "study the results" have undergone considerable reengineering with the change from print publishing to electronic. We have elaborated with different business models for publishing and disseminating results. On one end we have the model with traditional subscription based journals and on the other end we have examples of models entering the market that are operating with open access to the user. The cost efficiency of both the production and dissemination of publications should be optimized from the perspective of society. 5.2 RESEARCH SETTINGS IDENTIFICATION OF PUBLICATION MODELS For the cost calculations we have characterised the journals into two main categories. The traditional business model, which is subscription based, that is based on paid access and the relatively new model facilitated by the Internet, the open access model. The concept of business models is discussed in more details in Part 3 Current business models for scientific journal publishing. The subscription-based model is divided into three sub categories based on the format of distribution. Subscription based print journals print + electronic electronic only The open access journals are typically published only in electronic form; so on this case the sub categorisation based on format is irrelevant. On the other hand open access journals can be categorised based on type of funding. The main sources of funding for open access journals are the publisher's own institution, grants, or funding from professional associations. We can also identify journals that apply author charges or membership fees as a mean of funding (see Part 2 Characteristics of open access journals). Open access electronic / funding by institutions, grants etc. electronic / funding by author charges or membership fees.

66 version of 29-Sep-03 11:44 page COST DATA CATEGORIES AND COST MEASURES The cost data for the publish the results is categorised into two main types of costs resulting from the three activities described in Figure 29. General costs for publishing, including the two activities "do general publisher's activities" and "do journal specific activities" in the SPLC model. Article specific costs, including the "do journal specific activities" in the SPLC model. This categorisation is relevant for open access journals that mainly are published as individual journals and it is also reflecting the cost data gathered in the survey to editors of open access journals. For traditional publishers general costs could be separated into general publisher's costs and journal specific costs, since a journal publisher normally publishes a number of journals. However, in this study we have limited us to the two cost categories general costs and article specific costs. Journals that publish as print only or in combination with an electronic version also have costs for printing and distribution. The cost structure for traditional publishers is reported in Part 2, Scientific publication life-cycle model (SPLC): Validating the model by comparison with current publisher cost structures. General costs is in this study according to the SPLC model split up into costs for: administration IT-infrastructure planning of issues marketing to author marketing to reader other costs As administration costs we have regarded costs for general management and financing activities and as IT-costs are regarded costs for maintaining a sufficient IT-infrastructure. For journals publishing issues the planning of volumes and issues is considered a general costs. Marketing to authors and marketing to readers are in this study considered as separate activities, however we can think of cases where the two activities are not completely separated and marketing is directed to both authors and readers. Article specific costs is in this study according to the SPLC model costs for producing an article. editor's costs for maintaining the review process costs for the peer reviewing costs for technical editing costs for placing of the article on the web

67 version of 29-Sep-03 11:44 page 67 In the study we have used three units to measure costs: direct expenses or out of pocket money (Euro) time spent (hours) budgetary costs (Eurasia) Publish the results Publisher's Business Strategy Journal Review Policy Do General Publisher's Activities Plan for Running Journals 1 General costs Do Journal Specific Activities 2 Issue Schedule Manuscript Do Article and Issue Specific Activities Journal Article 3 Infrastructure for Running Journal Article specific costs Figure 29 Cost categorisation of the activities in the publish the results diagram of the SPLC model The costs for archiving and indexing of a publication in the SPLC model consist of costs to make the publication available, proper indexing and archiving facilities (See Figure 30). The libraries are here in the focus for cost dissemination. At the moment very few libraries perform cost calculations on an activity level, that is: costs for acquisition, handling of access rights or subscription handling costs of cataloging and making a publication available inside an organisation either in the library or on the web. costs for archiving in a library

68 version of 29-Sep-03 11:44 page 68 To get hold of cost data for these activities we are relying on secondary data from other studies made on library costs when changing from print journals to electronic journals. Archive and index Library and Indexing Practice Formats for Long Term Digital Storage Publication Make Publication Available Retrievable Publication 1 Perform Value-adding Services Value-added Services 2 Costs for acquisition, making available, archiving Archive Securely Long-term Archival Copy 3 Libraries and Bibliographic Services Information Brokers National Libraries etc. Figure 30 Costs for making publication available Cost measures for the publishing of an article We apply two measures first copy cost and total cost for publishing / article for the journals. First copy cost equals the per article costs (do article specific activities in the SPLC model) and does not include general management costs for running a journal (do general publishing and journal specific activities in the SPLC model). Total cost for publishing an article equals the general management costs / article + first copy cost. Cost measures for the dissemination of a publication There are few empirical studies addressing the economic impact of the change from print to electronic. Montgomery and King (2002) start the discussion of the need of comprehensive analysis. Their study, on the changes in a library s handling costs with the shift from print to electronic, covers cost for systems and services, space, staffing costs by function etc. The effort in the project described is to capture all costs, including overheads, and fixed costs. Since this

69 version of 29-Sep-03 11:44 page 69 study did fit the purpose of approximating the costs for make the publication available activity, the results in it are used as examples. The Montgomery & King study report cost data from the years The empirical data is gathered in the Drexel University library where a large transition from print journals to electronic journals took place during those years. The operational costs are counted separately for electronic journals as well as current print journals. The items for the cost study in the Drexel University Library are not one-to-one with the activities in the SPLC-model (see Figure 31) but they reflect the same idea of trying to calculate the costs for all the activities in the publication life cycle. Secure Access Rigths and Subscription Local Demand for Publications Subscription or Pay per View Facility 1 Publication Paper Publication Make Paper Publication Available Retrievable Publication 2 Electronic Publication Make Electronic Copy Available 3 Meta Data of the Publication Integrate Meta Data into Search Services 4 Searchable Metadata Alerting Messages Figure 31 The dissemination of the make publication available activity in the SPLC model DATA COLLECTION METHODS The gathering of the data for the cost study consists of the following: Study on the prevalence of open access journals Web survey sent to editors of open access journals Interviews with selected editors to traditional journals (See also the chapter 2.1) Secondary sources for library cost data Study on the prevalence of open access journals The study was performed in summer 2002 in order to establish the amount of Open Access scientific journals on the Internet. It resulted in a list consisting of 317 Open Access journals

70 version of 29-Sep-03 11:44 page 70 and information on web-addresses, subject and country of origin, was gathered about the journals. The sources of information were two databases for journal information (Ulrichsweb ( ARL ( ) and a directory of journals provided by Alison Wells ( The web survey to editors of open access journals The survey was conducted in May 2003 as a web survey. The initial contact was an sent to 310 editors of open access scientific refereed journals, where they were asked to participate in the study and to fill in the questionnaire designed as a web page. The answers were directed to an address at Hanken. The list of open access journals and the addresses to editors or the journal were collected earlier during the SciX project and the list has been used as a population in this survey. 10 error messages of incorrect address were received so the total number of s sent out was answers to the questionnaire were collected; thus the percentage of answers amounted to 20% of the population. Library cost data for the handling of periodicals For this information we have relied on secondary sources. The articles used are: Montgomery, C. H. (2000). Measuring the Impact of an Electronic Journal Collection on Library Costs. D-Lib Magazine vol. 6 (10). Montgomery, C. H., King, D. (2002) Comparing Library and User Related Costs of Print and Electronic Journal Collections: A First Step towards a Comprehensive Analysis. D-Lib Magazine vol. 8 (10). 5.3 RESULTS OF THE COST DATA STUDY PREVALENCE OF SCIENTIFIC REFEREED OPEN ACCESS JOURNALS In the beginning of the study collecting basic data about open access journals was important. 317 refereed scientific open access journals were identified from the sources used (Gustafsson 2002). The main research question was, what proportion of scientific refereed journals are available by open access principles? The proportion of open access journals in the most comprehensive database of journals, The Ulrich s web was established. This proportion was around 1%. The actual prevalence of open access journals is higher. Today in September 2003 we have the DOAJ directory of open access journals that contains around 500 titles and the BioMed central with 100 open access journals. So the percentage today is probably close to 2%.

71 version of 29-Sep-03 11:44 page 71 Prevalence of Open Access Journals FREE Ulrich s: 1 % Total Ulrich s: 99 % Figure 32 Prevalence of open access refereed scientific journals in Ulrich s web. The subjects covered by scientific refereed open access journals are shown in Figure 33. The journals in the area of social science form around 30% of all journals followed by arts and humanities, physical sciences and life sciences with around 17% each. Technology forms around 8% and the category general is 1%. Subject Broad Technology: 9 % Arts and Humanities: 17 % General: 1 % Social Sciences: 39 % Life Sciences: 17 % Physical Sciences: 17 % Figure 33 Subject categories of scientific refereed open access journals The country of origin reflects a clear predominance for United States, however, the United Kingdom and Germany as European countries represent 11%.

72 version of 29-Sep-03 11:44 page 72 Origin Country Others: 21 % USA 53 % Germany 3 % Canada 7 % Australia: 8 % The UK 8 % Figure 34 Country of origin for open access refereed scientific journals ALL OPEN ACCESS JOURNALS Since open access publishing in its current form is presumed to be volunteerism to at least some extent, we knew that exact cost data covering the whole publishing process would probably be very hard to get from all participants. The difficulties in providing cost data is also mentioned and regretted by several of the participants in the free comment section. Editing a journal is considered as part of the ordinary work at a university and a good portion of idealism is also true for carrying out the editorial work. The participants had three options for giving cost data; they could use one or all three. They could give the information as direct expense numbers or as budgetary cost. As an alternative they could give the approximate time spent on a task. The last alternative was used in most answers. It is perhaps also the most relevant measure and best reflecting the characteristic of the work of publishing open access journal. The editor puts in a lot of his own time and this is not calculated as direct or budgetary cost. Instead of giving information on individual tasks such as administration, marketing etc. the participants could give an approximation of total direct expenses or budgetary cost for running the journal or an approximation of the total time spent running the journal. 50 participants of the total 60 that took part in the survey provided cost data in some form, mostly as an approximation of the time spent on a task. Direct expenses and budgetary cost figures greater than zero were provided by relatively few, as can be seen in the Tables 11 and 12. For most general management tasks the reported value is zero, which means that the editors state that they have no direct expenses or budgetary costs. If they could not answer the question the editors were asked to leave the answer field blank.

73 version of 29-Sep-03 11:44 page 73 Table 11 General costs / year Task direct expenses time spent budgetary cost Aver. Median Aver. Median Aver. Median Administration 3058 ** h 50 h 8078 * 0 IT-infrastructure 1770 ** 0 50 h 40 h 836 * 0 Planning issues 266 * h 50 h 497 * 0 Marketing to authors 182 ** 0 30 h 20 h 424 * 0 Marketing to readers 17 * 0 14 h 3 h 291 * 0 Other 1683 * 0 88 h 0 h 2777 * 0 SUM h 163 h direct expenses time spent budgetary cost Aver. Median Aver. Median Aver. Median Total cost and time 6367 ** h 250 h 6125 * 0 * less than 5 observations indicate a value > 0 ** less than 10 observations indicate a value > 0 Due to the low number of observation of direct expenses and budgetary costs the time spent reporting unit is used in the following calculations. We have calculated the above values for general costs time spent / year transformed to a per article base. For this we have used the survey data, the median value of the number of articles published per year. The mean value is 16 articles / year. Thus we can calculate the general time spent /article (based on median values of the time spent values split into tasks and the approximation of total time spent / 16 articles) = 13h For open access journals the article specific costs or the costs relating to processing of an article the median for direct expenses is 0. The median for the time spent on processing an article is 22 h (see Table 12).

74 version of 29-Sep-03 11:44 page 74 Table 12 article specific costs direct expenses time spent Aver. Media Aver. Median n Editor's review 56 * 0 58 h 10 h Peer review 2 * 0 40 h 6 h Technical editing 93 ** 0 41 h 4 h Placement of article 161 ** 0 10 h 2 h Sum h 22 h * less than 5 observations indicate a value > 0 ** less than 10 observations indicate a value > 0 To convert the labour time spent value into monetary value we have used an approximation of salary costs for the editor the reviewers and an assistant. The labour time spent, 22 h is divided between the editor (professor), reviewers (professor) and an assistant (student or other office staff) approximately so that the editor and reviewers work 16 h (median value) on the review process, and the assistant works 6 h (median value) on technical editing and placement of the article. The salary / hour, including social costs, is for the editor and reviewer approximated to 45 and for the assistant to 20. Thus the calculated first copy cost based on median values is: Review process 16 h (10 + 6) * 45 = 720 Technical editing and placement of the article 6 h (4 +2) * 20 = 120 Total labour cost = ( ) = 840 The median value for direct expenses = 0 The first copy cost exclusive of time spent on the review process is 120 and including the review process 840. General management cost based on median value: The median for direct expenses on general management is 0. The median for labour time spent is 13 h. Since this is mainly work done by the editor we count an hourly rate of 45. Labour cost = 585 The calculated total cost for publishing an article (see Table 13) based on median value is: first copy cost + general management cost = 1425 Table 13 Calculated costs for open access journals on per article base First copy cost First copy cost General costs Total cost / exclusive of the review including review / article article process process Open Access

75 version of 29-Sep-03 11:44 page JOURNALS USING AUTHOR CHARGES Since we could identify 2 journals that use author charges as part of their main funding in the survey we have calculated separated values for this publishing model. The sample is not generalisable but reflects the cost structure for these specific journals using this publication model. Table 14 General costs for open access journals practicing author charges General costs 2 observations Author charges Direct expenses Time spent(h) Task Median Median Administration 0 62 IT-infra Planning of issue 0 0 Mark. Auth. 217,5 0 Mark. Read. 0 0 Other 0 0 SUM 5872,5 62 Per article/ ,1 General cost per article using the same calculation method as above. Labour costs = 3,1h * 45 = 139,5 Direct expenses per article = 294 Table 15 Per article costs for open access journals practicing author charges Per article costs 2 observations Author charges Direct Expenses Time Spent (h) ( ) Median Median Editor's review 0 55,5 Peer review 0 2 Tech. Edit Placement of article 43,5 80 SUM 119,5 157,5 Editor s and reviewer s time = 57,5h * 45 = 2587,50 Assistants time = 100h * 20 = 2000 Direct expenses = 119,5 First copy cost = 2587, ,5 = 4707 Total cost for publishing = first copy cost + general cost / article = , = 5140,5

76 version of 29-Sep-03 11:44 page LIBRARY COSTS The transition from print journals to electronic journals during the years reported in the Drexel University report can be seen in Table 16. As reported there is a rapid transition from a main collection of print journals to a main collection of electronic journals leaving only print copies for some central journals. Table 16 Transition from print to electronic journal in the Drexel University Libraries (Montgomery & King 2002 p. 2) Type Print e-journal The specific cost items counted are space, systems costs, supplies and services costs, staff costs per function. The operational costs annually reported on data are for handling electronic journals $ and for print journals $ (Montgomery & King 2002). In Table 17 the amounts are transformed into Euro using the exchange rate $1,00 = 0,87. Table 17 Yearly operational costs (Euro) for the current electronic and current print journal collection ( ) for the Drexel University Libraries (source: Montgomery & King 2002, p. 4) Item Electronic journals print journals Space Systems Supplies and services Staff by function Director/Administration Communication among staff Collection dev Acquisitions Physical handling: bindery, labeling, reshelving Record creation and maintenance PR/ communications/publications Reference Teaching (users) Other Sum Staff costs Total operational

77 version of 29-Sep-03 11:44 page 77 The total unique titles for electronic journals were and for print journals 370. The operational cost / title for electronic journals would then be around 14 and for print journals 208. To be able to count the handling costs on a per article basis the number of articles in the collection of journal titles for Drexel University Library is needed. 5.4 ANALYSIS AND DISCUSSION OF THE RESULTS IMPACT OF THE DIFFERENT PUBLICATION MODELS ON THE COSTS FOR THE PUBLISH THE RESULT ACTIVITY In the studies in this deliverable the aim has been to characterise different publication models, the open access journal, the open access journal using author charges and the traditional commercial publisher and the traditional non-commercial publisher. The cost data collected are grouped into the main categories; general costs, article specific costs and library handling costs. The measures "first copy cost" and total cost per article based on median value are comparable across the studies in this deliverable. Thus it is possible to make a cost comparison of open access publishing to the current publisher cost structures. However, the reported cost structure for the open access journals varies to the one reported for traditional and current publishers. The open access journal is mainly produced as a single journal by an editor or publisher and has the main sources of funding from the institution of the editor or the publishing institution. The open access journal has a budget only in a few cases and actual direct expenses or out-of pocket money is reported in only a few answers to the survey. Therefore the median value in this study for direct expenses and budgetary costs is 0 (zero). The most reliable unit is therefore the reported time spent working on a task. The calculation of overhead costs for using the space and computer equipment and network of the publisher s or editor s institution is an alternative to the direct expenses and budgetary costs, but in this study such calculations are not used. On the other hand the time spent by the editor and the reviewers on the review process is not normally counted as a cost. This has been done for open access journals in this study. The presented first copy costs and total cost per article are exclusive the cost for printing an issue. Printing costs and costs for distribution are relevant additional costs to those journals that are publishing a parallel print version of the journal to subscribers or to members paying a membership fee. The cost for printing and distribution is therefore reported separately. The two Scix studies on costs (open access and current publisher) are compared in Table 18. See Appendix 1 (model operating data) for the current publisher study. The figures used for comparison are for the Large commercial, the Large non-profit and the small non-profit society publisher.

78 version of 29-Sep-03 11:44 page 78 Table 18 Cost comparison for different business models First copy cost excluding costs for the review process Cost for the review process (editor s share) Cost for the review process (reviewers share) Cost for printing ** General costs per article Total cost per article Open access OA with Author charges (based on only 2 observations) Current publisher , , * (139, ) = 433, , Total cost per article + printing and distribution * The costs for the time spent on the review process (the editor s and the reviewers time) is not normally counted as a cost for the publisher. However the support of the review process is often the task for the publisher and in that case the cost or expenditures are counted. In the case of the current publisher study the editor s share in administering the review process is counted as a cost but not the time spent by the reviewers on the task, except in the case for the small nonprofit society publisher (168 ). For the open access journals the time spent by the editor and the reviewers on the peer review process is counted separately. **In the current publisher study for print journals the printing cost per article varies from 372 to 763. Some open access journals have a subscription and distribution in paper form as well as the open access online form. However there are only 2 observations on the printing costs for open access journals. The cost varies from 270 to 500. The first copy cost (exclusive the review process) for the open access journal and the current publisher vary from 120 to The reported 2119,5 for the open access journals using author charges is not generalisable since it is based on only two observations reporting a high number of time spent. The total cost per article is clearly lowest for the open access journals. The result is not surprising since only time spent on the tasks is counted as costs. On the bases of these figures we cannot generalise that the open access model is clearly cheaper than current publishing models. For future research one could argue that overhead costs for space, computer and network facilities also should be included in the comparison. Also time spent on reviewing submitted articles should be counted as costs for all business models.

79 version of 29-Sep-03 11:44 page IMPACT OF THE PUBLICATION MODELS ON LIBRARY COSTS It has been anticipated (Odlyzko 1999) that library costs for handling journals will drop when a change from print journals to electronic journals take place. The largest cost savings would be in cost savings for acquisition and management of a physical collection. According to Odlyzko the internal library costs spent on ordering, cataloging, shelving, checking out material and on reference help under the life cycle of a journal are at least twice the purchase price of the journal. Most library budgets follow the income cost method. The library is provided a certain amount of budgetary resources and the library management makes the decision on how the money, staff and space resources are to be used for the library customers. King, Boyce, Montgomery and Tenopir (2003) provide a framework for library economic metrics. Their perspectives of measurement include the library, the user, the organisation, the community served and the society. The specific metrics from the point of view of the library is inputs, exemplified as resources and outputs, exemplified as products and services. The derived metrics are performance, cost-effectiveness and in a larger perspective cost-benefit. In the income cost method the journal subscription prices are central and the counting of actual journal titles obtained by individual print subscriptions are often compared to the title number obtained by bundled licensing agreements for electronic journals or for aggregate services that provide access to journals from different publishers. The price models often vary and a proper comparison is difficult. Generally, counting only titles, this way of measuring is favorable to the electronic journals. The operational costs reported / title, are considerably lower for electronic material than for print material, 14 for electronic compared to 208 for print material. The main difference in operational costs for electronic material for open access journals and subscription-based journals is that acquisition and dealing on agreements through consortia is lacking for open access journals. Neither is there any need for access rights management for open access journals. On the other hand the cataloging and record creation of open access journals into the library catalog or a web-based listing of journals is needed. This is also needed for all electronic journals but the required data is probably easier to collect for subscription based electronic journals.

80 version of 29-Sep-03 11:44 page OPEN ACCESS TO SCIENTIFIC PUBLICATIONS AN ANALYSIS OF THE BARRIERS TO CHANGE? 6.1 INTRODUCTION One of the effects of the Internet is that the dissemination of scientific publications in a few years has migrated to electronic formats. The basic business practices between libraries and publishers for selling and buying the content have, however, not changed much. Scientists have in protest against the high subscription prices of mainstream publishers started Open Access (OA) journals and e-print repositories, which distribute scientific information freely, in the same spirit as the Open Source movement in software development. Despite widespread agreement among academics that OA would be the optimal distribution mode for publicly financed research results, OA channels still constitute only a marginal phenomenon in the global scholarly communication system, and only a small fraction of the total yearly production of scientific publications are available through them. This paper discusses, in view of the experiences of the last ten years, the many barriers hindering a rapid proliferation of Open Access. The discussion is structured according to the main OA channels; peer reviewed journals for primary publishing, subject-specific and institutional repositories for secondary parallel publishing. It also discusses the types of barriers, which can be classified as: legal framework, IT-infrastructure, business models, indexing services and standards, academic reward system, marketing and critical mass. Publication of scientific content has been one of the areas to benefit most from the emergence of the Internet. A scientific publication can, as an information good, easily be delivered electronically to the end user. Scientists already have the necessary equipment and skills to access the material from their normal work environment, and have in fact been forerunners in the utilisation of the Internet for communication. In the near future scientists worldwide will access most of the texts they read over the World Wide Web. In addition to traditional publications there are many other ways in which scientific communities can use IT and the Internet to enhance their communication and collaboration processes, such as discussion lists, databases of observation data, the sharing of models and programming code etc., but these are not considered in this paper. While the delivery technique for scientific publications has changed rapidly, the economic ramifications have hardly changed at all. The concentration of the publishing of journal titles in the hands of a few large players, in combination with electronic delivery, has made the strong players even stronger and there have in fact been both an anti-trust investigation [Competition Commission 2001] and an informal investigation [Office of Fair Trading 2002] into some of the proposed mergers of the biggest publishers. The extremely low marginal costs of selling information over the Internet favour the use of sales and marketing strategies such as bundling and differential pricing [Shapiro and Varian 1999]. Publishers of scientific journals have consequently rapidly started offering electronic site licenses to universities and university consortia. The key issue is that the there is very little competition in this industry and that the pricing schemes depend much more on each customer s willingness (and capacity) for paying

81 version of 29-Sep-03 11:44 page 81 than on the production prices. It is thus today more or less as expensive for university libraries and individual subscribers to access this material over the Internet as before in paper format. The serials crises, the long period of rises in subscription prices that started in the 1970 s thus still continues [Frazier 2001]. In parallel to these developments pioneering scientists have leaped to the opportunity that the Internet offers for bypassing the costly intermediaries in the publishing process. During the 1990 s several e-print archives as well as a few hundred peer reviewed electronic scholarly journals emerged. The common denominator for these is that they offer free access to the end product in electronic form ( Open Access ). The experiences of this first wave of pilots demonstrate that the barriers for changing the commercially oriented communication system were greatly underestimated. Due to such barriers the impact of Open Access on the total volume of scientific publishing is still negligible. This paper will take a closer look at some of these barriers, which broadly can be categorized into economic, legal, social and psychological. Trying to get researchers to support the move towards Open Access, which most agree in principle would be good for the advancement of science, is like trying to get people to behave in a more ecological way. While most people recognise the need to save energy and recycle waste it takes much more than just awareness to get them to change their habits on a large scale. It takes a combination of measures of many different kinds, such as technical waste disposal infrastructure, legislation and taxation to get massive behavioural changes underway. 6.2 WHAT IS OPEN ACCESS? Open Access (OA) means that a reader of a scientific publication can read it over the Internet, print it out and even further distribute it for non-commercial purposes without any payments or restrictions. At the most the reader is in some cases required to register with the service in question, which for instance can be useful for the service providers in view of the production of readership statistics. The use of the content by third parties for commercial purposes is, however, as a rule prohibited. Thanks to the open availability the linking to OA publications via reference lists is substantially facilitated, since the reader does not stumble into barriers such as use licenses, and each reference is only a "mouse-click" away. In general the author keeps almost complete copyright and can also publish the material elsewhere. The Internet is a superb channel for the free distribution of information in the public domain. Science as a social institution has also always had as an ideal the open sharing and critique of information. Thus there is a much stronger case for Open Access to scientific content than for instance for the use of Open Source methods in the development of IT applications [Björk 2001] not to mention free downloading of music files from subversive third parties, that is counter to the legitimate interest of the producing artists. A related development is also the posting of teaching resources for free on the Internet, a phenomenon to which the label Open Course Ware has been suggested [MIT 2001].

82 version of 29-Sep-03 11:44 page 82 The four most important OA channels are electronic refereed scientific periodicals, research area specific e-print servers (in this paper called subject-specific repositories), institutional repositories of individual universities and self-posting on the authors home pages. OA scientific periodicals have been founded since the early 1990 s, usually as individual efforts. Almost as a rule such journals are electronic only, due to the need to minimise costs, but in some cases it may be possible to have a paid subscription to a paper version. OA journals are usually funded largely by the voluntary work of the involved editors and direct or implicit grants (the free usage of the hosting university s web servers could be seen as a subsidy). In more recent years a number of efforts to publish OA journals on larger scale have emerged. Their business plan is usually to finance the operations through author charges. A variety of OA journals are also traditional subscription based journals, which become freely available in electronic form after a delay, typically 6-12 months. The best-known subject-specific repository was founded already in 1991 for high-energy physics [Ginsparg 1996]. Such repositories typically target at parallel publishing of material, which is being written for other outlets (such as conferences or traditional journals), allowing earlier and more efficient dissemination. Subject-specific repositories typically have emerged in research areas where traditions for the exchange of preprints have existed prior to the Internet and where the speed of publication is an essential factor [Kling and McKim 2001]. The guiding principle of such electronic archives is, that researchers themselves upload article manuscripts, conference papers etc. into the repositories. Thus very low maintenance costs can be achieved. The custodians of the repositories usually only check, that totally irrelevant material is not deposited. Papers in a repository are available globally much earlier, than for instance the finally published versions of the manuscripts in paper-based journals. In some areas like computer science this can be a significant benefit. Already since the earliest days of the web individual researchers have put copies of their own publications on their homepages. Although there are no published surveys of how common selfposting is this authors experience is that this is the most common OA channel today. Figure 35 displays one example of self-publishing. The increasing interest of the universities to start institutional repositories brings a more systematic and long-term commitment to this activity. Institutional Repositories represents a third important OA-channel, and are relative newcomers compared to the journals and subject specific repositories. Universities and their libraries are in a better position than individual academics to guarantee that the material is available even after decades and that the collection is systematically maintained, for instance to take account of changing file formats and media. Institutional repositories represent an integral part of the longterm strategies of the universities in question, in particular as these have to redesign their publishing and library policies to take into account the totally new conditions created by the Internet. The university s own production of thesis and working papers can easily be put up on such repositories, but in the long run the posting of the central production of the university s researchers, i.e. their conference and in particular journal papers, is crucial. Although institutional repositories can be seen as useful marketing channels for individual universities their most significant impact on the global scale can only be achieved via co-operation via open access indexing services.

83 version of 29-Sep-03 11:44 page 83 Today the primary channel for finding OA-material is through general-purpose web search engines, unless the reader is a regular visitor to the journal or repository in question. Despite their deficiencies general web search engines can be quite efficient, and a recent study from the domain of computer science showed that publications, of which copies had been freely available on the web, received on the average three times as many citations as others [Lawrence 2001]. The problem with using general search engines is, however, how to distinguish the relevant publications from all the very varied material, which is available on the Internet and tends to clutter the search results from queries. Figure 35. Research papers of professor Hal R. Varian available at EXPERIENCES OF THE PIONEERING YEARS The experiences from the first ten years of experimenting show, that it is much harder to change the system than originally envisaged [Guédon 2001]. In a few years hundreds of scientific journals adhering to the OA principles were launched, but of these roughly half have already disappeared and many only publish a few articles per year [Wells 1999], [Gustafsson 2002]. There are, nevertheless, some success stories, such as First Monday, which on the first Monday of every month publishes papers on Internet research [First Monday]. It is possible to maintain subject-specific repositories with minimal costs and in some domains these have acquired a prominent position. The best known of these is arxiv, in which researchers during its eleven year old history have deposited publications from the fields of physics, mathematics and IT [arxiv].

84 version of 29-Sep-03 11:44 page 84 Despite such success stories it had become clear by the turn of the millennium, that the enthusiasm and collaborative spirit of researchers involved in OA efforts wasn't alone enough to offer viable competition to the traditional subscription based delivery channels. The Swedish School of Economics and Business Administration (HANKEN) participates in the Open, self organising repository for scientific information exchange (SciX) project, financed through the fifth framework programme of the European Commission. SciX is a demonstration project, in which a fully operational subject-specific repository is built for one particular research community, and in which the experiences of the implementation and deployment are recorded and analysed [SciX]. The role of Hanken in the project is to study the scientific publishing process and the effects of different alternative business models on the life-cycle costs of the process. In addition the barriers to the introduction of Open Access models are being investigated, which is the topic of this chapter. On the basis of the results recommendations and guidelines for policymakers will be formulated. There is disagreement as to how advantageous open access on a large scale would be. Many proponents have cited the very low running cost of some pioneering journals and repositories. On the other hand publishers and some researchers put forward that the costs of producing journals of high quality are mostly fixed costs and that the share of costs for printing and distribution is very low. One thing which is often overlooked in this debate is the cost repercussions down stream from the chosen business model. Open Access would substantially lower a lot of the transaction costs throughout the process (both of publishers, libraries and readers). This is central research topic in the SciX project, but is not discussed further in this paper. Rather the hypothesis is that a scholarly communication system state in which a substantial part of all scientific publications are available through one or several OA channels would be advantageous for the academic community and society in general. Given that this hypothesis is through, which are the barriers for moving from the current system state to the envisaged OA state. 6.4 BARRIERS TO CHANGE After a decade of experimenting there is now quite a lot of evidence about the possibilities and difficulties in making Open Access a real alternative. Quite a lot has been written about the subject, based mostly on the hands-on experience of creators of journals and repositories [Odlyzko 1998], [Walker 1998]. In addition to publications written by OA proponents there are also some more critical voices. For instance Kling and McKim discuss the differences between scientific disciplines in adopting OA channels [Kling and McKim 2000]. A number of surveys have also been conducted where both authors and readers have been asked about their perceptions and choices. This author has for instance participated in a web survey shedding light on the topic [Björk and Turk 2000]. He also has first-hand experience of creating and running an OA-journal and in setting up a subject-specific repository. The discussion, which follows is based both on personal involvement, use of secondary sources and data from some recent empirical studies.

85 version of 29-Sep-03 11:44 page 85 Table 19 below can be used as a starting point for a discussion about the prerequisites and barriers for open access publishing. The three major channels discussed in this paper are open access journals, which function as primary outlets, and subject-specific and institutional repositories, which mainly function as secondary outlets complementing the mainstream channels of journals and conference proceedings. Self-posting on the web are left outside the discussion, even though it is a rather important channel at present. The barriers and means have been classified into six different categories: Legal framework, ITinfrastructure, Business models, Indexing services and standards, Academic reward system, Marketing and critical mass. In the table the number of asterisks (from zero to three) denotes the importance of a particular item in hindering a rapid transition process. The importance is a subjective judgement made by the author. Thus in the opinion of this author there are no (or very small) legal obstacles for the proliferation of Open Access Journals, whereas this is a very central issue to be solved if institutional repositories are to take a prominent position in the academic communication system. Table 19. A classification of different types of barriers for increased open access publishing and their relative importance Open access Journals Subject-specific repositories Institutional repositories Legal framework - * ** IT-infrastructure ** * ** Business models *** ** * Indexing services and standards Academic reward system Marketing and Critical mass ** - *** *** * * *** ** *** In the following these barriers will be discussed one by one.

86 version of 29-Sep-03 11:44 page LEGAL FRAMEWORK OPEN ACCESS JOURNALS In the case of traditional journals, typically published by commercial publishers or learned societies, the author usually grants the publisher a rather exclusive copyright, in return for the services that the publisher renders the author. It must be stressed that contrary to what proponents of OA often state, authors do not give away the product for free. Rather they trade their papers without specific payment for the services that the publisher renders them (peer review, quality labelling, marketing, dissemination). The fact that some publishers have charged page charges to authors in addition to charging subscribers is one indication of this. The surrender of copyright is so total that for instance in Finland, where it is rather common for a Ph.D. thesis to consist of 4-5 previously published journal articles plus a summary, the author has been forced to ask written permission from the publishers to publish copies of his own papers as part of his thesis (which usually is published as a monograph by his own university). Many copyright forms grant the author the right to limited distribution of copies to colleagues. The emergence of the Internet has brought into light a particular problem, concerning the noncommercial distribution by posting copies on the web. In many of the copyright forms which publishers ask authors to sign, this area is not properly addressed and constitutes a grey zone. Open Access journals, on the other hand, have from the start adopted a rather liberal approach reminiscent of the licensing schemes used by the Open Source programming community (often referred to as copyleft). As a rule the author retains the copyright to the work. What the open access journals typically are interested in is that the paper, if made available elsewhere in the exact format of the journal, is attributed to primary publication in the journal, and also that nobody else (except the author) can resell the content. In conclusion the copyright issue does not constitute an obstacle for the proliferation of open access journals. Currently used copyright agreements for OA journals are quite satisfactory from both the author s and the journal s viewpoint. Figure 36 presents one OA journal.

87 version of 29-Sep-03 11:44 page 87 Figure 36. ITcon the Electronic Journal of Information Technology in Construction available at: SUBJECT-SPECIFIC REPOSITORIES A strong impulse for subject-specific repositories was the long lead-time between submission of a draft manuscript and the publication of the full paper. In some areas of science, such as highenergy physics, a tradition of scientists exchanging preprints on paper already existed and the new repositories just developed this mechanism further. The name preprint server describes this function. One of the problems with low cost subject-specific repositories is that due to the high number of papers in the successful ones, the managers of the service have no resources to check the legality of the papers posted. It is up to the authors and their discretion to take out papers once they have been accepted for publication if they have signed copyright agreements, which prohibit keeping the copies on the server. On one hand the situation resembles that of the controversial music server Napster, on the other hand there are significant differences. The biggest difference is that in the case of scientific publications it is the author of the work who voluntarily puts up a copy on the server, not a third party. The legal problems resemble the situation with institutional repositories and will thus be discussed below.

88 version of 29-Sep-03 11:44 page INSTITUTIONAL REPOSITORIES Institutional repositories will in early stages get their initial content from works of the faculty for which the university itself retains the copyright, such as Ph.D. thesis and working paper series of departments. These entail no legal problems since the authors or the university retain the full copyright. In the longer run, however, the critical mass of institutional repositories depends on the inclusion of the best work of each university s faculty, that is the journal papers published elsewhere. From a legal viewpoint this constitutes a challenge, since university administrations will be very careful not to be break any copyright contracts, in contrast to individual researchers posting copies of their work on their home pages quite carelessly in spite of possible legal obstacles. Many of the major publishers have recently, if the author asks for it, granted the authors permission to the parallel non-commercial electronic publishing on the web pages of the university of the author. In the EU-funded Romeo project several leading publishers were asked about their official view concerning the publishing of the manuscript or the finalized paper in open access servers [Romeo]. Of these 33 publishers agree in some form, whereas 49 gave a negative answer. Together the publishers who participated in the survey represented 7169 journals. When the results were weighted according to the number of titles, 49 % of the journals permitted the publishing of either or both versions. Although many publishers currently are quite liberal in their attitudes towards parallel noncommercial web posting in subject-specific and institutional repositories, there is a lot of uncertainty in the longer run. The title of a recently organised conference by the International Association of Scientific, Technical & Medical Publishers: Universal Access to STM Information: By Evolution or Revolution, tells a lot about the ambivalent feelings publishers have on this issue. As long as the publishers revenues are not seriously threatened, they advocate willingness to allow authors the right for parallel posting in institutional repositories. They even see this as additional advertisement for the primary publication. But if parallel OApublishing gains momentum and starts to have a negative effect on subscription income it is possible that the copyright agreements become tighter and also that compliance to existing agreements is more actively monitored. Thus if and when subject-specific repositories and in particular institutional repositories start to have more substantial content and threaten the subscription income for the primary publishing channels that they replicate, it is probable that the copyright issue becomes a central battleground.

89 version of 29-Sep-03 11:44 page IT INFRASTRUCTURE OPEN ACCESS JOURNALS The IT technical infrastructure of electronic peer reviewed journals can include a wide spectrum of different features. Storage mechanism for the papers and the metadata (static web pages vs. database) Format of the papers (HTML, PDF, XML etc.) Treatment of graphics and hypermedia content Management of drafts and the review process Indexing and linking to external parties Alerting and personalisation services for readers Hyperlinked discussion threads Statistics on readings, citations etc. for authors Security back up, mirror sites etc. Most Open Access Journals have so far been individual efforts created by single academics and groups of academics, often managing the journals on a part-time basis. Thus the IT technical infrastructure of these journals is quite varied, ranging from rather rudimentary static HTMLversions to quite sophisticated database driven systems, depending on the skills and resources of the creators. The platforms have seldom been bought from outside companies or larger publishers. One of the drawbacks of these systems is that they are very vulnerable, in case the person in charge for some reason or other stops working with the journal. The notable exceptions to this are provided by two major efforts utilising new business models for running portfolios of OA journals. The technical infrastructure of Biomed Central is on a par with the leading commercial publishers and includes coding of the papers in XML as well as workflow management of reviews. Biomed Central gets considerable economies of scale since they publish almost a 100 journals [BioMed Central]. Public Library of Science in the US has recently announced plans to start publication along similar lines [Public Library of Science]. Both publishers finance the operations through author charges. In the longer run the publishers of individual journals would benefit a lot from pooling resources, for instance by sharing software applications, or using collaborative web hosting. Such discussions are for instance under way in the Nordic countries for smaller national or Nordic peer reviewed journals [Kvaendrup 2003]. Another possibility is to use open source applications for running such journals SUBJECT-SPECIFIC REPOSITORIES Like OA journals most subject-specific repositories are the results of individual efforts and the corresponding IT systems have been made by the academics themselves. The best known of all

90 version of 29-Sep-03 11:44 page 90 these, the arxiv server (see figure 37), has for instance a rather simplistic web interface on top of its database but has on the other hand been quite successful, due to the fact that the service obtained critical mass rather early [arxiv]. One conclusion to be drawn is that readers for this type of services care less about sophisticated technical features than about being assured that they can find most of the stuff they are interested in at one place. Although there would be benefits of sharing IT-resources also for subject-specific repositories, this might be more difficult to achieve in practice, since such repositories often are bundled with other web-services to the research community in question, thus necessitating quite a lot of customisation. Figure 37. The arxiv e-print archive, INSTITUTIONAL REPOSITORIES Institutional repositories present a rather different picture from current OA journals and subjectspecific repositories. University libraries have considerable funds at their disposal and are used to outsourcing part of the work in building their IT-infrastructure. They also take a very longterm perspective in the setting up of institutional repositories. They need for instance to take into account already in the planning stages the periodic necessity to upgrade the storage media and the storage formats. In addition to scientific publications universities also have a need to systematically organise the educational web material produced by the faculty.

91 version of 29-Sep-03 11:44 page 91 When universities start planning for such systems they are likely to use one of the following solutions or perhaps combinations of these: Plan for joint national collaboration platforms Use well-proven open source applications Buy the software from outside IT-consultants Outsource the whole service to commercial publishers An example of the first option is the Dutch DARE project [DARE]. The currently best-known open source solution is the D-space system, originally developed by MIT for its own internal use but currently offered for use to other universities [MIT 2001]. 6.7 BUSINESS MODELS TAXONOMY OF BUSINESS MODELS FOR SCIENTIFIC JOURNAL PUBLISHING The first half of this chapter takes a look at current business models for scientific journal publishing. The second part explores the most common open access business models. The emergence of the Internet has opened up totally new business models for selling scientific publications and data. In the paper world the need to exactly identify each recipient beforehand was clear and the cost of delivery was substantial. Thus some sort of subscription or prepayment systems was obligatory. In electronic delivery over the web the marginal cost of delivering the extra copy of an article rapidly approaches zero. Thus the central question becomes how to cover the fixed first copy cost of packaging the articles on the web site. In table 20 the central business model options are shown. The discussion will be structured according to these options. In the table stars indicate the relative importance of the different models.

92 version of 29-Sep-03 11:44 page 92 Table 20. Business models for scientific journal publishing Type of Access Business Model Delivery Medium Paper Electronic Paid Access Open Access Limited Open Access Pay-per-view * * Journal subscription *** * Collection subscription ** *** Community service *** External grants * Advertisement * Author charges * Institutional author ** charges Delayed Open Access * Limited functionality * PAID ACCESS Pay-per-view The pay-per-view option has been in limited use already in the print era, in the form of specialized document delivery services, inter-library loans etc. It has complemented the mainstream subscription model for the case where readers have had an urgent need for a particular article not at hand in the library of the local university. The problem has however been the high transaction costs of handling such requests. On the surface one could imagine that this option could become quite popular for electronic delivery, since automated systems could help in lowering the transaction costs. Good parallels can be found in e-commerce. There are, however, a number of barriers to this option. The first and foremost is an anomaly in the internal budgeting of universities. Typically journal subscriptions are paid from the general library budget. Thus individual researchers and their departments experience journals as a sort of public good, and their job is just to lobby with the library to see to it that the journals on their wish list are included. On the other hand paid access to individual articles, for instance as interlibrary loans, are usually paid by the researcher

93 version of 29-Sep-03 11:44 page 93 himself from his own budget or projects and thus the threshold for using this option can be rather high. The transaction costs and practical problems of securing the document can be quite high. From the viewpoint of the seller the reliance on a large scale on individual article sales would thus lead to a few top sellers accounting for a big part of revenues, whereas most articles would have hardly any demand at all. This could have consequence on the peer review process. On the whole the prediction is that it is very unlikely that this option would account for more that a very small fraction of the sales volume to the university world. On the other hand for research institutions outside the universities and for companies engaged in R&D this option can be a good choice, since they often do not want to subscribe to a large portfolio of journals that in some cases may be in low use. There are full text delivery services, such as the British Library Document Supply Centre. Journal subscription This has traditionally been the dominating business model. For the vast majority of journals the number of subscribers is in the order of , thus the subscription prices have to be high to cover the rather substantial fixed costs. Quite a lot has been written about this subject and is quite well understood for the case of paper delivery. A generalisation of the observations of several studies and observers is that publishers often have to use the substantial profits of a few journals with big subscription bases to subsidise low circulation journal and/or new upstarts. Up until around a lot of the subscriptions where from individual scholars. Often such subscriptions where tied to membership to certain professional of academic societies, since members could get the subscriptions for reasonable prices. In the last 30 years the number of individual subscriptions has gone down quite radically (due to price rises) and most subscriptions are now from libraries, university departments etc. Collection subscription The rapid transfer to electronic delivery has made totally new pricing schemes possible using a well-known technique in e-commerce called bundling. Thus publishers or aggregators sell access to vast numbers of titles for prices, which are individually negotiated for each client. In library circles this is often referred to as the "big deal". Bundling is very profitable for publishers since it tends to obscure the prices from customers and makes comparisons difficult. Earlier, for paper delivery, the subscription price where listed and well-know. Now since different libraries subscribe to different bundles it is very difficult to compare prices. In combination with the monopolistic nature of the business this type of pricing situation is very favourable to the sellers since they can use price discrimination in order to extract the maximum revenue from each customer (the marginal costs of delivering the product are very low).

94 version of 29-Sep-03 11:44 page OPEN ACCESS Community service This model has so far been the dominant way in which open access journals and e-print repositories developed. The model is closely related to the Open Source model for developing software applications of general interest, which has been used for instance in the development of the Linux operating system. In this model all the key persons involved donate their working time for free and the journals can be run with very low pocket expenses. The necessary equipment, Internet access, servers etc can also be found through the participants organisations. The term community service is used to describe this activity because this type of activity is seen as the sort of activity in which academics should be involved without any additional remuneration except their normal salaries. This type has always been applied to activities like reviewing article submissions, acting as opponent etc. As part of the SciX project the characteristics and economics of current open access journals was empirically studied through a survey sent to 300 editors, of whom 60 answered. The results show that the main source of funding is the editor s or publisher s own institution and that the work can be characterised as voluteerism to a high degree. The key question mark concerning this model is its sustainability. It does not scale up very well from the running of a journal with only a small number of papers published per year (the median in our survey was 16) to the characteristics of a major journal with articles per year. If the editor-in-chief or some of the pioneers behind the project cannot continue with the project the risk for a collapse are real. Advertisement Open access journals could in principle be financed by advertisement, like many general services on the web, but this type of business model could only work for a few isolated fields of science, such as perhaps medicine, in which drug companies take an interest. Already now drug companies are paying the subscriptions to journals for practicing doctors or have set portals serving the medical profession, so Open Access journals could well fit this picture. External grants This is more realistic business model. For instance the founding of the journal ITcon was helped by a grant of some euros from the Swedish council of building research, and the arxiv e- print repository has been mainly financed this way. In both case grant funding has been combined with community service type of activities. The problem with grant funding is longevity. It is often possible to get a grant for a 3-4 year period on the basis of the experimental nature of a journal, but often the granting authorities stipulate that the journal has to survive on its own after such an initial period. Often their internal rules are such that they cannot give regular funding for longer periods to the same activities.

95 version of 29-Sep-03 11:44 page 95 Author charges Charging the authors for the services rendered by the journal is a very promising business model, because it could be able to combine the benefits of the open access with a secure regular monetary income to the publisher. Author charges are not even new, they have existed for many journals in the form of page charges, which have provided an additional income for the published (in addition to subscriptions). Currently there are a couple of interesting experiments with author charges. BioMed Central publishes around 100 titles in the area of biomedicine and has advertised a target level of around 500 USD as author charge. The BioMed Central journals are published commercially. In the USA the Public Library of Science organisation has received a large grant of some 9 million USD to start publishing open access journals, which also would be financed by author charges. They charge a 1500 USD author fee per article. Institutional author charges One of the problems with author charges is that it might often be difficult to persuade the authors to pay the charge. Often the grants for doing the research are already consumed by the time a paper is submitted or accepted for publication. And even if the authors had access to the necessary finance there might be competing uses, such as the cost of attending a conference. Also open access journals financed with author fees compete with traditional journals and other Open Access journal, which don't have author charges. One way out if this dilemma is for publishers of OA journals to make bundle deals with whole universities or even larger players who assume the responsibility of paying the author charges. Thus in the case of BioMed Central: - The university of Helsinki is paying a yearly fee of some 6600 euros for covering the possible papers from its faculty. - JISC from the United Kingdom has made a deal with the publisher covering 180 UK universities. - The Budapest Open Access Initiative of the Soros foundation has agreed to finance the author fees of all authors from developing countries LIMITED OPEN ACCESS Delayed open access One of the central ideas of the petition that the Public Library of Science send around in 2000 was to put pressure on commercial publishers to release articles as open access after a delay of

96 version of 29-Sep-03 11:44 page 96 say 6 months or a year. The idea is that research information loses in value as time goes by and thus that most subscribers would not drop their subscriptions because of the delayed open access. In practice this model works best in such a way that some other party than the original publisher takes on the responsibility of making the material available as open access. Limited functionality open access This model takes account of the fact that there are several levels of access to the same information. It might for instance be possible for readers to browse the full text of articles on the screen but without the possibility for printing articles or saving copies locally. Also the cross linking to other material etc. can be omitted from the OA version OPEN ACCESS BUSINESS MODELS Open Access journals Most Open Access Journals have so far been established by individual pioneers or groups of academics. The main business model has been to minimise costs and to fund the operations as a form of open source project, where hardly any transfer of money is involved and all costs are absorbed by the employers of the individuals participating. A recent web survey involving the editors of Open access journals carried out by Hanken confirmed this to be the predominant business model. This business model is very vulnerable to sustain operations in the longer term and for scaling up from a few papers per year to larger publication volumes, since that might necessitate employing staff. Its also not well suited for such journals where copyediting and layout work of graphics etc cannot be handled by the authors themselves. Other possible business models, which would provide more funding for professional-level operations (such as the employment of staff) could use advertisement, subsidies from learned societies or research funding agencies, or author charges, in order to keep the end product freely available on the web, rather than take recourse to subscription fees. All of these have and are being tried out, in different combinations. The most controversial is the one involving author charges (for instance used by the BioMed Central journals) since this reverses the role of a publisher from a seller of a commodity to consumers to a provider of services to authors. Getting individual researchers to pay sums in the order of euros for publication might be very difficult unless a journal already is regarded as a top-level journal in its field. A way around this dilemma which is being tried out by BioMed Central is for the publisher entering into umbrella agreements with universities who pay a yearly fee covering all article submissions of their own faculty. Another business model issue is that of co-operation. Through co-operation or outsourcing of part of the work to commercial companies the publishers of individual journals can obtain the same economies of scale, branding etc, which large commercial publishers have today.

97 version of 29-Sep-03 11:44 page 97 Advertisement can work in some limited fields of science such as medicine, where for instance drug companies may have an interest. A very important group of players are the learned societies, which historically were the ones to start scientific journals as we know them now. They could see Open Access as an important service for their constituency and society in general. Unfortunately many learned societies see journal publishing as an internal profit centre generating finance for other activities or an activity, which at least should generate income enough to cover its cost. From this perspective open access through author charges would still be acceptable. A further problem is, however, that many offer journal subscriptions bundled with their membership fees and fear that going open access would threaten the income from such fees. The business model issue is central to the further proliferation of Open access journals. The currently dominating volunteer work only model does not easily scale up to large-scale and sustainable operations and the other business models need yet to demonstrate their strengths. Subject-specific repositories Subject-specific repositories have evolved in a few select fields. The selection of fields has come about through a combination of existing behavioural infrastructure, individual entrepreneurship and pure chance. The best-known example of such a service is the arxiv service in high-energy physics. Despite its very low running costs, it still costs money to run and the whole service has recently been transferred to Cornell University. The main aim of the European SciX demonstration project is to set up such a disciplinary repository for the field of IT in architecture and construction and fill it with initial content. Discipline-specific repositories are usually rather tightly aligned with pre-existing communities of researchers who communicate a lot with each other, meet at regular conferences and publish in a limited number of journals. It would be very difficult for such repositories to start to either charge subscription fees or to start to levy fees on authors uploading their papers (one aspect is that the payments would be rather small per upload and the transaction costs easily could consume most of the generated gross income). Thus the main options left would be subsidies from hosting universities or advertisement. Probably the dependency on voluntary work will prevail. Institutional Repositories Discipline-specific repositories will in the longer run be "threatened" by institutional repositories, since both compete for the same material. If institutional repositories gain momentum and are indexed effectively through standards such as the Archives Initiative - Protocol for Metadata Harvesting (OAI-PMH), they will offer a parallel channel to the same content as subject-specific repositories, and have clear advantages in their business models. From the business model viewpoint the development of institutional repositories will depend a lot on the political decisions universities have to make concerning the future roles of their libraries and publishing departments in the electronic world. Since the need for storing and

98 version of 29-Sep-03 11:44 page 98 handling paper copies of material from the outside decreases very rapidly the finance thus freed could be used to finance the institutional repositories instead. 6.8 INDEXING SERVICES AND STANDARDS OPEN ACCESS JOURNALS One of the major drawbacks of Open Access Journals has so far been that they rarely have been indexed in the commercial indexing services for searching quality assured publications, which universities provide to their researchers and students. Information about the publications in the journals has instead been spread through direct marketing among select communities of academics and through being indexed by general web search engines. Partly this has been due to a view that existing scientific indexing services belong to the old establishment and that there is no need for their intermediation. Partly editors of relatively young and experimental journals have had a hard time getting their journals included in such services. Indexing services fulfil in this connection a dual role in helping the marketing of the journal and its content. Firstly they help in attracting occasional readers who may not even be aware of the journal s existence. Secondly the fact that a journal can claim being indexed in lends prestige to the journal and thus helps in attracting more and better submissions. A particularly important one is the Science Citation Index. This service regularly monitors a selection of a few thousand of the most important refereed journals and counts statistics of the citations in the articles that these journals publish. The more citations there are to a journal s articles in the other journals in this core selection, the higher a journal s impact factor is. Academic appointment and grant committees take these impact factors in consideration when ranking the output of academics and there are thus high rewards for publishing in such journals. The use of SCI by university administrations as a decision support tool has thus become one of the strongest barriers to the success of Open Access journals since it tends to strongly favour old established journals [Guedon 2001]. It is very difficult to get new journals accepted in SCI before they have established a track record, and being outside the core literature of SCI makes it very difficult to get good submissions and establish a track record SUBJECT-SPECIFIC REPOSITORIES Subject-specific repositories have usually not experienced any need to be indexed by third parties. Firstly it would be very difficult since most of the material in them is or will be published elsewhere, and thus the references should be to those primary sources. Secondly subject-specific repositories strive to compete (in terms of coverage) with commercial indexing and full-text services within the subject areas they were designed for rather than work in symbiosis with them.

99 version of 29-Sep-03 11:44 page 99 The emergence of the OAI-Protocol for Metadata Harvesting (discussed below) may however change the situation in the near future INSTITUTIONAL REPOSITORIES Indexing services and the standards that underpin them are crucial to the success of institutional repositories. Since each university contributes with only a very small fraction of the total publication output of the academic world, and since this output in addition is widely spread over a large number of disciplines, browsing through the user interface of a particular IR is relatively uninteresting, unless for very targeted searches for the output of a particular researcher. A typical usage scenario would be to use the IR to get a free electronic copy of an article published in an expensive journal to which the reader has no access, or perhaps to get access to a conference paper where it might be impossible any more to get a copy of the proceedings. The e-thesis repository at the University of Helsinki is displayed in figure 38. In the same way as for subject-specific repositories it would be difficult for Institutional Repositories to be indexed in current established indexes for any of the content of which they are the secondary outlet. The solution to this dilemma is in general web search engines or in a new type of search engines dedicated to scientific web content. If an author puts an electronic copy of his own publications on his web pages the main channel to this is already today through general search engines such as Google. Dedicated Open Access search engines for scientific content which are tagged according to the OAI- Protocol for Metadata Harvesting are currently under development. Figure 38. The e-thesis repository at the University of Helsinki, available at: