Communications of the ACM
The First 10 Years of the ECCC Digital Library
The advent of the Web created many new possibilities for the scientific community, including rapid access to up-to-date research-relevant information and quick dissemination of researchers' results. However, the Web is a huge collection of aggregated information that is largely unstructured and unfiltered. Concerning scientific publications accessible over this medium, one must choose between self-published research reports (on preprint servers) or peer-reviewed electronic scientific journals. Both options have their weaknesses. The publications of the first, unstructured method, are not reliable in the sense of quality and quotability; the second method has the drawbacks of often lengthy publication delays similar to traditional scientific journals.
The Electronic Colloquium on Computational Complexity (ECCC; eccc.uni-trier.de/eccc/) is a digital library designed and founded in 1994. Conceptually it provides a compromise and lies somewhere in the middle between the two previously mentioned types of publishing methods. An international voluntary working scientific board (see eccc.uni-trier.de/eccc/info/board.html) ensures the proper scope and quality control of submissions by a newly designed screening process (which is the term we chose to avoid already used terms like reviewing). The standards for publications in ECCC are high due to the involved board members. Successfully screened submissions are published in the series of ECCC reports, which usually occurs within a few days (never longer than two months) after submission. All the accepted ECCC reports are freely available via the Web.
ECCC did not only modify the acceptation and filtering process but also made use of further potential of the new electronic media. For example, a mechanism for ongoing discussions on ECCC reports is provided, and authors have the possibility to present the most recent versions of their work in ECCC along with previous accepted work. Unique ECCC report numbers and version control ensures bibliographic consistency. Finally, ECCC provides a portal for the computational complexity community, where books, online lectures, lecture notes, survey papers, theses, and other related links are provided.
Traditional Channels of Scientific Publication
Publishing scientific results is the central means to promote scientific progress by distributing new ideas and insights and assuring copyright. In the course of development of scientific culture different forms of publication channels have been established that serve different goals. In computer science, the most important channels are:
- Journals, where complete scientific papers are published;
- Preprints, where authors publish and distribute their results in paper-based format or electronically under own control; and
- Conferences, where scientific results are published in the conference proceedings under the control of a program committee.
Analyzing these channels, there are four central properties for scientific publications: quality assurance, up-to-dateness, quotability, and long-term availability.
ECCC is somewhere between peer-reviewed scientific journals, conference proceedings, and preprints.
Quality assurance concerns the originality and quality of the published material as well as the thematic classification of the described research work. In the current environment of an increasing amount of publications, readers become more dependent on reliable and pre-sorted sources of relevant information. The most popular method of quality assurance is peer reviewing by specialists.
Up-to-Dateness. Although seen as the ultimate publication channel, due to the time-consuming reviewing process the delay between submission and publication in a highly ranked journal may involve years. Even in the case of conferences the review process often lasts months. This endangers the up-to-dateness or currentness of the scientific results published in an article, which contrasts with an important advantage of the publication form of preprints: preprints reflect actual status of ongoing research, and have little latency between writing and publishing, at least if they are published in electronic form.
Quotability. In addition to the quality assurance and the currentness of published research results, the traditional academic publishing methods differ in the quotability of the provided information. This is the most persuasive reason for paper-based publishing. Researchers need to be able to cite other publications in an unambiguous and lasting way, which is not at all guaranteed by a data repository such as preprint servers run by a single person or organization. Journals fulfill these characteristics perfectly. The printed text is unchangeable, the text remains readable for many years, and journals are archived in libraries. Conference articles published by a professional publisher in printed conference proceedings provide a comparable but more restricted situation for the quotability of the articles due to limited distribution.
The long-term availability of electronic publications suffers from the fast development and evaporability of storage media and of software. It is often the case that after a few years computer systems and data formats of electronic publications become outdated and therefore useless. Archiving the data by regular electronic backups does not guarantee the long-term availability of the information. Consequently, the electronic counterpart of paper-based preprints lacks in terms of long-term availability and uniqueness. (Unsigned) electronically published information can easily be modified and changed: the information is therefore ambiguous.
Electronic Colloquium on Computational Complexity
In 1994, a group of top researchers in the field of computational complexity (which later formed the ECCC scientific board) discussed the possibility of using the Internet for creating an "online meeting place" for their research community to exchange recent results in timely, yet still somewhat filtered, manner. As a result of these discussions, the ECCC was designed and founded in late 1994. ECCC builds a compromise between the traditional publication methods mentioned previously. It is somewhere between peer-reviewed scientific journals, conference proceedings, and preprints. To combine the advantages of these publication methods ECCC implemented some new concepts, which have proved to be successful during the last 10 years.
To shorten time between the submission of a research paper and its publication, and therefore enhance the up-to-dateness of the material, the time-consuming process for quality assurance was reduced. On the other hand, this process should be good enough to provide a functioning filtering of the publications based on quality and topic. In contrast to the time-consuming review process in traditional journals, ECCC papers are merely screened by one member of the scientific board of ECCC. Before explaining this screening process in detail we describe the overall design of ECCC.
In ECCC there are two states for a submitted research paper: the waiting state and the published state. When an author electronically submits his work, it enters the waiting state. Submissions in this state can be seen only by the members of the scientific board of ECCC. A single member of this group is necessary to move the submission after a short review and filtering process into the published state, where the paper is visible for all ECCC readers.
The screening process of a paper is processed by one member of the scientific board. Compared to the classical review process this leads to a considerable time gain. Experiences with this simplified filtering process and the broad acceptance of ECCC during the last 10 years has shown that with this methodology a surprisingly high-quality standard can be achieved. The combination of the following factors makes this possible. First, the scientific board of ECCC comprises 40 top researchers in the relatively small field of computational complexity, most of whom are committee members in many conferences and editors of journals. Therefore, the scientific board has a good overview of recent works and ongoing research developments. Second, submissions to ECCC are brought to the attention of all members of the scientific board. A member selects a submission for screening according to his or her personal preferences, experiences, and on a voluntary basis.
The combination of the screening method and the immediate electronic publication of research results (papers are published as soon as they are accepted) provides a very short time interval between submission and publishing of a paper (up-to-dateness). If after two months a paper is not selected for screening by a member of the ECCC board, it is rejected due to a lack of interest (the lack of interest is seen as the paper is not in the scope of ECCC).
Another important property of a publishing method is the quotability of the published articles. Since ECCC reports are only published electronically some decisions concerning the long-term availability and the uniqueness must be made. To guarantee that articles published in ECCC are available over a very long time interval we:
- Regularly back up all data and the system;
- Annually produce an archive CD that is sent to several libraries and to all members of the ECCC board; and
- Print out in regular intervals all electronic reports and revisions, and archive this material in different locations.
Each accepted report is assigned a unique report number. The number of the paper and some additional information are included as a header (like a watermark) in the published work. In order to guarantee that the cited article is unique, some regulations and conditions are attached to the publication of a work. After publishing an article in ECCC it cannot be changed, even if this would be technically possible. Authors desiring to modify or correct an article must submit a revision that is treated almost like a new ECCC report and is published beside the original report of the same report Web page.
ECCC in the role of a colloquium. The electronic platform of ECCC provides readers the possibility to discuss recent research work. The discussion mechanism is very similar to the submission mechanism: a discussion is submitted to the ECCC server and is published together with the ECCC report under consideration. In order to prevent the abuse of this mechanism and to ensure high quality, the discussions are moderated.
In addition to the discussion mechanism, ECCC offers the possibility for a continuous "life" of already published papers. To this end, ECCC implements a revision mechanism where corrected or modified versions of a report can be submitted and published. However, although it would be easy to replace a published report by a new version, this would destroy the quotability of this report. Therefore a revision of a report does not replace it, but it is published together with the report. To make revisions quotable, they receive a unique number that refers to the number of the original report.
ECCC in the role of a portal. In addition to the publishing channel, ECCC provides a Web portal for the community of complexity researchers and has become the first place where scientists and students look for all types of relevant information on computational complexity. The ECCC portal currently provides (links to) monographs, lecture notes, online lectures, survey papers, and a number of theses and dissertations. Moreover, ECCC provides a place where complexity theorists can publish links to their personal Web home pages. A newsletter, various pointers, and links to additional information sources of interest for complexity people, such as conferences, books, and journals complete the offerings of ECCC.
In addition to the content provided by ECCC, search mechanisms have been implemented, which allow the search for keywords in author names, titles, and abstracts.
One major potential problem for the acceptance of ECCC in the community was that some journals or conferences require that a published paper is not published elsewhere. While preprints and preprint servers are usually not considered publications in this context, ECCC might well be considered as such, which made users reluctant to actually submit papers to ECCC. In discussion both with users and publishers, ECCC decided not to claim copyrights of any work (so the author may later transfer the complete copyright to a publisher). Consequently, ECCC allows authors to remove a report if this is requested by another publisher (this is the only case in which ECCC allows withdrawal of a report) but the number assigned to the removed report will not be reused to ensure proper citability. A removed paper is replaced by a notice that the paper has been removed together with the new bibliographic data.
Acceptance and Usage by Computational Complexity Researchers
After 10 years ECCC is widely accepted as a publication method for research papers in the field of computational complexity. The quality of the papers is high and is comparable to material from top conferences or top journals in the field. During the last 10 years more than 900 ECCC reports were published. On average, 300 ECCC reports are downloaded daily. The ECCC Web server averages 3,000 hits per day and transfers approximately 3,000MB of data each month. ECCC users are located in more than 100 different countries, with the majority of the users located in the U.S. and Germany. ECCC reports, comments, and revisions are provided in both .ps and .pdf formats. ECCC also provides a mailing list, where the newly published reports are announced monthly. At present, 457 users are registered; of this group, 140 researchers have registered their home pages in ECCC.
©2006 ACM 0001-0782/06/0100 $5.00
Permission to make digital or hard copies of all or part of this work for personal or classroom use is granted without fee provided that copies are not made or distributed for profit or commercial advantage and that copies bear this notice and the full citation on the first page. To copy otherwise, to republish, to post on servers or to redistribute to lists, requires prior specific permission and/or a fee.
The Digital Library is published by the Association for Computing Machinery. Copyright © 2006 ACM, Inc.