Lotus Eaters On-Line: Forming an Alliance Between Groupware and the Web in a Pre-Prints System

Among the facilities available from the scholar's workstation of the future will be access to pre-prints collections with added value. Precisely what form such 'added value' will take remains to be seen, but it will address the current tendency for on-line pre-print depositories to become largely untamed wildernesses of relatively unstructured material, with wheat and chaff mixed in frustratingly unpredictable proportions. This is not universally the case, but pre-print collections can become dumping grounds for all manner of stuff and the bigger they get the more unreliable they often become as sources of efficiently accessed and valuable material.

Of course, that is also one of the important things about the very concept of on-line pre-prints: what is valuable to one user is rubbish to another. Nobody is making definitive judgements on a user's behalf about what is of value. Print material, and electronic journals that follow the peer review and editorial procedures of print, will always tend to impose such judgements in advance (evoking all sorts of professional and disciplinary criteria). Indeed, as electronic publication struggles to achieve respectability there is an in-built pressure to follow established quality-control mechanisms. Pre-print systems can and should be different.

'Formations', a project in the eLib Pre-Prints Programme Area, was established at the end of March 1996 to explore some of these issues. The project will set up a live trial pre-print system (in areas such as performance research, cultural policy and media) in early 1997, with the target of July 1997 for a stable implementation of that system, based on six months of iterative development. A first detailed evaluation will be produced a year later. 'Formations' has a further 'exit phase' of eighteen months set aside to capture and communicate as much as possible of the learning that will have taken place inside the project and to plan for life after eLib.

'Formations' is deliberately positioning itself within the current debate about the relationship between 'groupware' and the Web. That relationship could map fairly neatly onto the relationship between the editorial side of electronic publishing, on the one hand, and free-for-all pre-prints on the other, except that most collective editorial work is typically done off-line (supported by email) rather than as an integral and visible part of the particular electronic system. When electronic publications are launched onto the world's wild Web (sic), they have already been highly selected and shaped. What if 'groupware' could introduce some form of selection and shaping into a pre-prints system but integrally to it and flexibly enough to preserve the characteristic 'lucky dip' experience of pre-prints?

'Groupware' at its best is currently represented by LotusNotes release 4. Lotus , now part of IBM, has built up a base of some 3 million users, or 'seats', and well over 10,000 'business partners' worldwide who develop Notes-based applications and bolt-on enhancements to the Notes system. The latter allows asynchronous communication among users centred on an 'object store' of messages, documents, threaded discussions etc. Lotus has specialized in handling free-form information rather than rigidly structured databases (although these can be plugged in to a Notes system easily). Lotus SmartSuite 96 provides desktop applications, including wordprocessing, that integrate seamlessly with Notes. Does this start to sound like the beginnings of an integrated scholar's workstation? Maybe, but Notes is proprietary and expensive - two very real drawbacks in the academic world.

A Notes-based system has to be carefully designed to do a particular job (with application development tools that tend to be demanding of time and skill). Users need dedicated 'client' software to access the system. Once set up, a Notes system can be a very powerful tool but development and deployment costs have made it a high-level corporate asset. That is now beginning to change in principle. Lotus have begun steering Notes very deliberately towards the Web and, therefore, towards a much wider range of potential users and applications. In practice, however, Lotus would need to make some real pricing concessions to the educational marketplace. Lotus Notes might then be one important means of adding value to Web-based working outside closed corporate environments, especially for academics and the electronic library.

Or at least, as the most highly developed groupware platform, Notes will be the first place where the groupware/Web convergence takes concrete shape. Others may follow, but Notes is a good place to begin. For this reason, 'Formations' will attempt to integrate Notes-based components into its Web-based pre-prints system. Subsequent evaluation and technological changes may reveal other, better options but a Notes/Web combination looks like the best place to start learning about these things now.

There are five technical areas of immediate concern (these general issues are also touched on in a useful Byte article by Bill Roberts, 'Groupware Strategies', Ju ly 1996):

(1) Interoperability. Our scholar's workstation ultimately needs to support an easy pick-and-mix of tools and services. As software for doing interesting things on the Web begins to proliferate, it is becoming clear just how difficult such interoperability is to achieve. Release 4 of Notes supports TCP/IP, SMTP, HTTP and HTML, on top of its existing support for all major operating systems and networking protocols. In practice this means that a user can wordprocess in Lotus Word Pro (part of SmartSuite), add the document to a Notes application store (e.g. for annotation by a group) but also automatically generate an HTML version of the document and have Notes add it to a Web site, without worrying about any of the different standards involved. The process also works in reverse, with Web material being sucked into a Notes system, complete with live URLs. The underlying documents may take nearly any form, including threaded discussions and graphic presentations.

A tailored Notes application can provide a user-friendly front end for the whole thing. Of course you still need to have bought a fair amount of expensive Lotus software at both server and client ends to achieve such seamless integration ('seamless' is becoming something of a mantra word) but the technical interoperability exists. Microsoft, Sun, Netscape and others are all developing bits and pieces of new Web-oriented software to achieve groupware functionality via the Web but the standards involved look far from open at this stage.

(2) Application development support. In the kinds of client-server system we are talking about, there should be two levels of application development or programmability. On the server side there will be the semi-permanent applications, such as a pre-prints 'store'. On the client side, there should be enough lightweight programmability for users or groups of users to set up their own tailored workflow and access procedures if they so wish. An application development environment with that degree of flexibility is a lot to ask. 'Components', or small application modules that can be re-combined Lego fashion, look like the way forward, especially if they can be architecture-independent like Java 'applets'. Lotus Notes and SmartSuite desktop applications can function as 'containers' for Lotus' own components and eventually for Java applets as well, marking a move in this direction. But what components might one need for an improved pre-prints system?

(3) Flexible 'document store'. The HTTP server and Web browser have undoubtedly been a near-revolutionary change in how we store and retrieve documents on-line, driven by the huge flexibility of hyperlinking. However, following hyperlinks is only one kind of on-line activity. Making the Web support sophisticated searching, well organized asynchronous discussions, viewing options for users other than one-page-at-a-time, organization of material beyond simple bookmarking, etc. is all leading to strain on the basic HTTP server, or on the backstage Common Gateway Interface expected to enable such work.

A Web server and a groupware application server nested side by side, with easy interoperability, translation and communication between the two, may offer greater flexibility and robustness. In a pre-prints system, is there room for these two 'levels' of functioning within the document store and what should each offer?

As a case in point, Notes offers a highly customizable front end that can be much more flexible than a Web browser (although a browser is also now part of the release 4 bundle), and supports a range of different 'views' onto the stored material, including expandable outlines, annotations and so on. The ability to 'view' a pre-prints store in various ways, backed up by sophisticated searching (Notes offers phrase, sentence, wildcard and Boolean) may be a crucial way of adding value to such a system over and above what can currently be achieved via the Web on its own.

(4) Notification. Web browser developers are currently experimenting with better procedures to check sites and automatically alert the user to changes. Notes has already implemented an agent-based notification system, extending the usefulness of its long-established document replication procedures. The latter ensure that changes to a document are executed on-demand in any copies of the document that may be held elsewhere. In a more flexible pre-prints system, dispersed collaborative work on documents (whether by co-writers or 'editorial' groups) could be easily synchronized in this way, avoiding the obvious nightmare of proliferating versions. In fact, the more notification available the better in a pre-prints system, as an antidote to information agoraphobia. Notification does not interfere with open-ended browsing but may be an increasingly necessary convenience.

(5) Security. Particular systems may need to set up different levels of security for different purposes - from access security on some documents, through authentication procedures to verify users' identities, to full firewalls erected to keep intruders away from the sensitive inner workings of a system. Adding value to a pre-prints archive may include different levels of security to allow different degrees of access to and participation in the evolving store of material. (Some dispersed workgroups, for example, may wish to 'protect' early work in progress.) Again, Notes offers such flexibility while the Web is a tangle of often incompatible attempts at securing something which mostly wants to be anarchic and free.

So in all five of these crucial areas for 'adding value' to what can be done on the Web, especially in the case of a pre-prints system, Lotus Notes has the technical edge. But it is still a closed system at its centre, no matter how permeable it has become to the Web at its outer boundaries. To reap all of these benefits (in interoperability, application development support, flexibility in the document store, notification and adaptable security), one has to buy into the Lotus Notes world and close out those who do not. A lone user has to pay around #100 for basic desktop client software to get a seat at a Notes application (although multiple licensing can reduce this significantly). So-called 'core clients', with more functionality, cost correspondingly more. This is fine for large-scale deployment in a corporate enterprise but is a bit much to ask of academic users in exchange for adding some bells and whistles to a pre-prints server which they might have been getting free access to via the Web.

'Formations' has to explore whether many of these groupware-based benefits can be leveraged towards the academic user before the threshold of unacceptable cost and inconvenience is crossed. Fortunately, the convergence of groupware and the Web generally (and the development of Notes in particular) is tending to draw these benefits gradually out of the closed enclaves, where they are sustained by corporate budgets, into more accessible regions where cheaper hybrid options become possible. As the 'Formations' project develops, that convergence may make it easier to achieve what the project is interested in doing.

To be clear about this, 'Formations' will not be setting up a conventional closed Lotus Notes application and inviting users to buy client software with which to access it. Instead we want to see if some of Notes' functionality can be leveraged into a more open system, with primarily Web-based access. What kind of an 'engine' can Notes be, driving an improved pre-prints archive? The solution may still entail some levels of participation where having Notes client software is necessary or desirable but we want to define those boundaries, between the proprietary system and the Web, in ways that would justify the cost to those who need such access without disadvantaging those who do not. It remains to be seen whether this is possible.

To achieve that sort of balance we will need a clear picture of what users want and of the processes, or workflow, involved in using a pre-prints system in the context of scholarly work as a whole. 'Formations' is developing an overview of its area of activity based on three intersecting 'models', or representations of the factors, processes, interests and functions involved. These are a Process Model, a Value Model and an Application Model. These representations are not intended to be prescriptive but as project planning tools they offer the advantage of systematizing and reminding us of things that might otherwise get overlooked.

The Process Model represents the structure of activities and interactions that may take place around an enhanced pre-prints system, both in contributing to it and in accessing material (terms like 'select', 'inform', 'discover', 'prompt', etc. occur in the first version of this model). An interesting aspect of developing this model is the question of what to do about speech act theory. Long controversial in CSCW research (computer-supported cooperative work), speech act theory reduces interactions to 'acts' such as 'request' or 'promise'. This way of characterizing interactions in our Process Model has tended to creep back in unnoticed, despite the fact that CSCW systems such as 'Coordinator' (for highly structured email exchanges within a workgroup) ran into trouble because users did not wish to have their activity structured around such explicit 'intentions'. They preferred that accessing an on-line information or communication system should be a non-interpreted act (as distinct from explicitly 'searching', 'requesting', etc.) The 'Formations' Process Model needs to be sufficiently explicit about identifying the interactions without imposing restrictive labels in advance on what users do. (See Action Technologies for the kind of work that evolved from 'Coordinator').

This is one of several areas where the Value Model will provide checks and balances on the more instrumental formalism of the Process Model. The Value Model seeks to represent the interests and values of various groups significant to the project: users, the academic community more widely, and eLib/JISC. Thus an eLib value is 'effecting culture change' while a basic user value is 'ease of use'. The model will attempt to map out such values and plot their interactions. Like the Process Model, the Value Model will undoubtedly evolve in sophistication and usefulness as the project proceeds.

These two models will inform the third - the Application Model. 'Formations' will hold distinct the actual pre-prints application which it builds from the more abstract specification of such an application that emerges as desirable from the project's work. There is a simple version of the 80:20 rule here. A costly and time-consuming effort could eventually be put into hauling the actual application up to the ideal standard of whatever Application Model the project identifies as workable and worthwhile (i.e. crossing that last 20% gap). However, identifying that 'ideal' and learning from it may be more important, more accurately reflective of the interests and values that converge on an eLib project, than putting a 100% finely tuned client-server pre-prints application in a shrink-wrapped box. Setting our sights on an Application Model, which may or may not perfectly coincide with what we actually have time, money and energy to build, is our way of staying focused on what matters.

The trick, of course, will be to devise a project plan, backed up by appropriate project management procedures, which achieves an acceptable balance among these three 'models' while exploring the technical benefits that might be reaped by a pre-prints archive from the groupware/Web convergence. To that end our project plan attempts to set the Process, Value and Application Models against a timeline in such a way that each informs the others. For example, a particular technical benefit of groupware such as Lotus Notes should not make its way into the 'Formations' application unless it fits the process somewhere, reflects a value that is actually of concern to the project, and can be done (at least done in a 'good enough' way, if not ideally) with the resources and time available.

The triple-checking of technical features in this manner should cut down pretty quickly any purely technical over-ambition we might have! Our hope is, also, that the models will themselves evolve in subtlety as they are progressively exposed to the actual technical work and the questions it raises. Evaluation activities will be at the core of the 'Formations' project. Where our three 'models' overlap we will be developing largely qualitative research techniques to explore the interactions among processes, values and the application's functioning.

At this early stage, it is appropriate for us to end with the conclusion from that recent Byte report on groupware: 'If you tried to design a seamlessly integrated groupware platform from scratch, you might come up with something very much like the Web. Or you might come up with something resembling Lotus Notes. Then again, you might come up with a model that blends the best of both worlds.' (Byte, July 1996, p.78)