KEYWORDS: BLEND; QUARTET; ADONIS; ELVYN; Training; Electronic journals.
Unfortunately, UK involvement in EIES was "prevented by a Post Office embargo at that time on extensive transatlantic computer-based message transition" (Shackel and Pullinger, 1984). Denied of the opportunity to participate in the American study, Shackel resolved to set up a UK project along similar lines. Although Senders was later to say "I have seen the future and it doesn't work" (Senders, 1981), the digital libraries research started at Loughborough continues to the present day (and beyond).
While Brian Shackel and colleagues were working at the HUSAT Research Institute in Loughborough, fifteen miles away at Leicester University the Primary Communication Research Centre (PCRC) headed by Jack Meadows was also pursuing a programme of research on `new technology and developments in the communication of research'. The PCRC was concerned with a variety of electronic information formats including the electronic journal (e.g., Singleton, 1981) and the moderated bulletin board (e.g., Booth, 1988). When Jack Meadows moved to Loughborough, it served to strengthen the site's reputation as a major UK contributor to digital libraries research and development. The present paper outlines the major digital libraries projects undertaken at Loughborough.
In at least one respect, CHF proved potentially superior to a traditional paper journal. Although each actual article was `read-only' once issued and could not be altered, there was space allocated for comments to be entered on each article and these comments could then be seen by subsequent readers of the article. The fact that the articles' authors were also part of the `electronic community' meant that they too could read -- and respond to -- the comments. The resulting dialogue created much more of a feeling of `live' research than is possible in the paper medium where it is not uncommon for an 18-month period to elapse between submission of an article and publication in the journal.
While it might be tempting to think that the electronic medium speeded up the process of publication, this is not necessarily the case. Indeed, Shackel (1991) reported a median publication time of `just over 32 weeks' (i.e. about 8 months) for articles in CHF and it is possible that this could be attributed to a `novelty effect'.
What seems more likely is that the electronic medium more easily supports comment and dialogue than the paper medium, as has more recently been shown by current electronic journals such as Postmodern Culture with its associated discussion group.
One of the lessons BLEND was able to learn from EIES concerned authors' attitudes to submitting papers to experimental electronic journals. In the US study, authors had shown some reticence to submit to such a journal, not unnaturally since their tenure might well depend on publications and electronic journals did not really count. Hence, in the BLEND system, papers were considered to be `archived' rather than published and authors were expressly permitted to publish them in the traditional literature.
Although considered a success in many ways, CHF was not without its problems. For example, it suffered from the technology of the day -- poor quality screens, poor quality telephone lines, low transmission rates and so forth. Movement through the articles was slow and so, not surprisingly, users preferred to print copies of articles that they wanted to read in their entirety. The articles themselves were restricted to plain ASCII text and `typewriter graphics'. Also, at the time of the BLEND project, terminals were not so readily available. It was not unusual for suitable terminals to be located in the university's computer centre (which could well be in a different building to the academic's office) and would frequently be in use by other users. This is in contrast to the present position where most academics have some sort of computer on their desk (Shackel, 1990) and in many cases it is connected into the campus local area network.
HyperBIT offered the user several advantages over the paper version. For example, it was always available on the desktop (whereas the paper version might well have been borrowed by a colleague -- even personal subscriptions `go missing'!) The entire contents of the journal could be searched in order to locate, say, all articles which mentioned `screen' or referred to work by `Maguire'. The ability to move between related articles using the hypertext links was also advantageous, as was a pop-up window facility which provided instant access to the bibliographic details of references without leaving the text. This facility was provided on the basis of observations of many users who would keep a finger permanently in the References section of the article when using the paper version, turning to the section when they encountered a reference in the text and then returning to the text. In this sense the facility provided what Brian Shackel called an `electronic finger'.
The chief disadvantage of HyperBIT compared to the paper version concerned graphics. Although the Macintosh system on which it was implemented allowed the display of quite sophisticated graphics, animation and sound, the screen resolution was far lower than the resolution available to produce the average paper journal. Typical screen resolution was 72 dots per inch (dpi) whereas a typical typesetting machine has a resolution of 1200 dpi -- roughly 16 times better! In many cases this is not a real problem -- the line art typical of diagrams and graphs presents no difficulty. However, the display of half-tone photographs, for example, was not really feasible on a standard Macintosh screen of the day.
As part of Project QUARTET, colleagues and I undertook various investigations of the ADONIS system. For example, the feasibility of requesting a document via electronic mail and having it delivered over a high-speed telephone line to a local fax machine was demonstrated. More interesting in the context of the digital library were the studies we made of the system as a resource for direct use by academics. At that time it seemed that such a system might offer the scholar a great deal and the reasons why it failed to do so provide valuable insights into some of the problems of electronic journals.
Like HyperBIT, the ADONIS system suffered the problem of screen resolution. Although the system used a high resolution screen, biomedical journals make frequent use of photographic material which could not be displayed adequately. In addition, although the system software allowed for searching on normal bibliographic details, the fact that the journals were stored as bit images meant that a full text search was not possible. Display of the pages was extremely slow (ironically, partly due to the need to redraw a high resolution screen -- modern hardware and software would not necessarily suffer so) and movement through an article was only possible one page at a time so users found it frustrating as a way of viewing journals. Also interesting was the finding that 219 journals were simultaneously too many and not enough. They were too many in the sense that no single user was interested in more than a small proportion of the 219. However, as a proportion of the biomedical literature as a whole, 219 is only a small part -- Medline, for example, covers about 3200 journals -- and hence for most users the system did not contain their favourite journals.
Potentially the project offered a learning experience for three different parties: the publisher; the libraries; and the users. However, in practice too few users were available to allow a full-scale evaluation of the various implementations. Nevertheless, the project provided valuable insights for the publisher and libraries. In almost all the cases, the original format requested by a site library was not the one which they eventually made available. For example, the Loughborough systems librarian had originally thought that PostScript would be the most desirable format in which to receive the journal. This was subsequently changed to a tagged text format with TeX being the first specified. In the event the material was finally requested in SGML so that it could be converted to HTML, with X-Mosaic being used as the access client software. Further details of all the implementations and economic considerations can be found in Rowland et al. (1995).
The Training Electronic Journal (TEJ) is our answer to this quandary. As an integral part of our Information and Publishing degree program, it allows students to experience all aspects of an electronic journal while providing a `safe' environment in which to do so. The project also involves collaboration with other UK library schools and with sites in Australia and Sweden. Each site is responsible for an `issue' of the journal and consequently students in all the countries concerned can see how other students have approached the various tasks. The TEJ project is due to finish formally at the end of 1995. However, first indications are that its evaluation will show it to be a successful contribution to the degree program and it will therefore continue to exist as a teaching and learning resource.
The technology now exists to support quite sophisticated electronic journals (see, for example, Pullinger, 1994). However, issues regarding the most effective use of such technology still remain. For example, although Pullinger was able to demonstrate browsing of full-colour journals at a remote site across the wide area network, the demonstration depended on end-to-end direct access to the UK's experimental SuperJANET high speed network. As more and more universities connect to this network, will the effective bandwidth still support such remote access or will it prove more effective to distribute the journals to local sites, as in project ELVYN?
For effective desktop delivery across the wide area network, the scholar's local area network must also support high speed transmission rates and this will require most UK campuses to upgrade their existing LANs. Centralised systems have been proposed (see, for example, Gardner, 1990) and Project Muse based at the Johns Hopkins University is making the full text of journals available in World Wide Web format (Kelley and Lewis, 1994). However, the journals involved are largely text-only and even they can be slow to access in the afternoon from the UK.
Similarly, current technology is now fast enough to allow the display of page images and searching of the displayed text, as demonstrated in the TULIP (Zijlstra, 1994) and CORE (Landauer et al., 1993; Olsen, 1994) projects. However, whether the page image is the appropriate format for the electronic journal is an empirical question. Landauer's results suggest that the optimum format depends on the task for which the reader is using the journal, a result which is not surprising in the general human-computer interaction context. A related question concerns whether public domain interfaces (e.g., most World Wide Web browsers, some TeX viewers, some PostScript viewers) will continue to be used extensively or whether proprietary interfaces (e.g., Adobe's Acrobat, Farallon's Replica, No Hands' Common Ground, AT&T's RightPages) will gain significant use. How many interfaces will scholars have to learn in order to access their favourite journals? The usability of such interfaces will also determine user acceptance.
It was often assumed that electronic publication could be achieved in a much shorter timescale than traditional paper publishing. This view can now be seen to be naïve, based on a view of the technology rather than the human component. In parallel publishing ventures such as ELVYN or TULIP, the technological procedures are much the same for both paper and electronic versions and hence there is little time advantage for the electronic version. Even in electronic-only publishing, it is the essentially human components which cause the biggest delay -- papers sit on referees' desktops for a long time, no matter whether the desktop is real or virtual!
One thing which BLEND demonstrated and which continues to be true today is that the electronic medium allows for the linking of papers and the discussion of them. Once a paper is published, discussion can take place immediately, without waiting for the next issue of the journal. Furthermore, such discussion can be archived along with the relevant paper so that future scholars can benefit from it. Unfortunately, when the BLEND project ended, so too did the Computer Human Factors journal. The only publicly available record of the four issues of the journal which now exists is, ironically, the paper version which was published as one of the project reports (Shackel, 1986) and which does not contain the discussions.
As both authors and readers, the scholars are the most important part of the communication chain and hence their requirements and attitudes will continue to be important. Publication serves the scholar's career needs as well as the information dissemination function. It is not surprising, therefore, that the reticence to submit to electronic journals experienced by the EIES project and circumvented by BLEND has more recently been evident in the Online Journal of Current Clinical Trials (Wilson, 1992). A recent report on information technology in libraries in the UK (Joint Funding Council, 1993) specifically recommended that funding councils give equivalent weight to publication in refereed electronic and paper journals, a recommendation which will hopefully help to erode authors' reticence and strengthen institutional confidence in electronic journals.
The author has also benefitted immeasurably from working with Professors Brian Shackel and Jack Meadows, without whom much of the work described above would not have taken place.