ARTICLE IN PRESS
Journal of Visual Languages and Computing 17 (2006) 288–291
Journal of Visual Languages & Computing www.elsevier.com/locate/jvlc
Foreword
Aims, achievements, agenda—where CDs stand now 1. Introduction What an honour this is for both Dr. Marian Petre and me: a special issue because our paper [1] has been so heavily cited. Invited to contribute ‘‘a few words’’, it seemed best to us, since our paths have diverged, to present such words separately. Exploring the intersections and unions of our recent thinking would be fascinating but take too much space. 2. What were the aims? Back in the 1980–1990s the dominant approach to human–computer interaction (HCI) was very mechanistic and very detailed; the chief example being GOMS, in which a complete action-plan for each supposed user task was laid out and then the interface was evaluated by considering how long each action would take. Although GOMS and its cousins met a real need and have continued to be important, useful, and influential, they were not what I wanted: I wanted an approach that would enrich the vocabulary of HCI, by finding labels for recurrent problems or phenomena. When the same problem was noted in diverse domains, perhaps in spreadsheets, simple music sequencers, and the Unix file structure, we should have a name for it, such as ‘hidden dependencies’. Physics envy! Those elegant conceptions—idealizations such as velocity, acceleration, mass—were not created by accident, but by hard thinking. Could HCI not do the same? Could we not create idealizations that would make our discussions clearer, enable us to recognize similar cases in different domains, help to understand pros and cons of design choices? Nor should the idealizations be limited to interaction with computers; there’s nothing special about the computer, except that it has been foregrounded by the information revolution. The goal should be to enrich the vocabulary of research on any information artifact. Flowcharts versus Nassi–Shneiderman diagrams; staff notation of music versus 16th-century lute tablature; the card index of an old-style library versus a computerized database; the pencil as a design tool versus the computer drawing package; any device or notation that was a manipulable information structure, whether a computer was involved or not. Further, I wanted something that would understand change of state. At that time, evaluations of the user interfaces of spreadsheets and word-processors were much in the 1045-926X/$ - see front matter r 2006 Published by Elsevier Ltd. doi:10.1016/j.jvlc.2006.04.002
ARTICLE IN PRESS Foreword / Journal of Visual Languages and Computing 17 (2006) 288–291
289
air, but the focus was on menu layout, size of button, and other details of the physical object. All important enough, but what about the problems of having to remake a table of contents for a long document if a change of font size moved text to different pages? Lastly, I wanted something that would be directly concerned with design. Design is characterized by frequent changes of plan. What were the characteristics of devices that were good design tools? Later it became clear that different devices were good for different types of activity, and the aims were broadened to try to understand the relation between activities and desirable characteristics of devices. My colleagues were full of fruitful ideas, and much of the growth of the CDs framework consisted simply of arranging their ideas, or our joint ideas, in a pretty pattern. Among the many sources were David Gilmore’s ‘role-expressiveness’, or understanding why a particular component is present, and his ‘match-mismatch conjecture’ that a notation is never good for all purposes, but only when its structure matches the structure of the user’s task; Stephen Payne’s work on consistency between syntax and semantics, with a sophisticated representation called Task-Action Grammar; Marian Petre’s observational studies with expert designers, revealing the designer’s need for ‘escape from formalism’ and a secondary notation, and the importance of what we later called ‘juxtaposability’ (thanks to Robin Jeffries for that name). Later, Alan Blackwell made it clear that more activities than design could and should be accommodated, leading to the ‘profiles’—attempts to chart which dimensions were most relevant to which activities. Numerous other ideas were discussed at various times by a roll-call of colleagues so long as to be embarrassing. Little would have happened without the originality and diversity of their conceptions. 3. Achievements This issue highlights one particular achievement, the comparative evaluation of some visual program languages (VPLs) by Marian Petre and me. We chose VPLs as a target domain for a number of reasons. For one reason, it was important for HCI to show that its compass could be extended to new domains. Another reason was that, since VPLs were being actively developed with little organized thought about their usability, we felt it was important to show how they could be systematically evaluated. And to top it off, Rachel Bellamy, Marian and I had empirical studies to explain. We were delighted to find that everything fell into place, although the resulting paper was indeed rather lengthy. An equally notable achievement, at least in my opinion, was the analysis that David Hendry and I performed on spreadsheets [2]. What is striking about spreadsheets is that they are so popular even though they have what appear to be usability problems: they cause numerous errors, and there are no abstraction facilities. Hendry and I felt that we were able to explain their popularity for individual usage, because they are almost ideal ‘incrementation’ tools—but more excitingly, we offered an explanation, one that I still accept, for the part in they play in organizational life. As depicted by Nardi [3] spreadsheets are used as communication tools across different specialties in an organization, because they are easy for both parties to understand, and that feat makes it worth tolerating their opportunities for hard-to-catch errors. But she did not explain why. Hendry and I claimed to show how their very lack of abstraction facilities, their
ARTICLE IN PRESS 290
Foreword / Journal of Visual Languages and Computing 17 (2006) 288–291
neutrality of ontology, and their suitability for exploratory design made them ideal for such purposes, and indeed led directly to the organizational role that Nardi called the ‘gardener’. Since those days, a younger generation has moved forward, and some of their work can be found in this volume. It is especially pleasing that the CDs framework has been adopted as a routine tool in at least two teams, those led by Dr. Burnett and by Dr. Clarke (see papers in this volume by Dagit et al. and Green et al., respectively). Becoming a routine tool is, to me, the framework’s proudest achievement. 4. Agenda The agenda items are perfectly predictable: keep the framework relevant, and get it used. How do we keep it relevant during a period when current technology changes so quickly? If CDs really do apply to any information artifact, then they should apply to the growth areas, such as personal music devices, voice-mail systems, on-line travel agencies, on-line auction sites, social software such as blogs and ‘‘social bookmarks managers’’, computer games, genealogy software, dating sites—the list goes on, and in most of these areas there is little published usability analysis. It would be good to demonstrate that CDs could contribute. And we should move beyond the individual level. HCI research today encompasses the place of an artifact in its social or organizational milieu, and I believe the CDs framework has contributions to make. The Hendry and Green paper made a start; a much more detailed study has been made by Hundhausen [4] with his concept of ‘communicative dimensions’ for human-human interaction. If the framework is to remain relevant, it would be good to see more such work. Getting the framework used, even if shown to be relevant, might entail improvements in its presentation. The framework ought to be to presented by somebody (volunteers welcome) in a shorter, clearer, simpler fashion, with appropriate explanation of its pros (and, of course, its cons), with plenty of emphasis on why no dimension leads to a judgement of ‘good’ or ‘bad’ until we know the activity. (Even viscosity can be desirable for some purposes.) Online interactive examples of information structures might be particularly useful pedagogic devices. To gain acceptance, any HCI tool must not only be well presented; it must offer good prospects, with evidence that the results will be worth the investment of learning it and using it. Not much investment is required to use the CDs framework, but the results are in a currency that is multi-dimensional and ‘soft’, rather than a single-valued ‘hard’ currency like predicted time. It is not always obvious what the benefits of such soft currency are. The agenda for CDs should include better presentation of what benefits it can offer, how it can prompt rethinking and redesign—how, in short, it earns its keep. References [1] T.R.G. Green, M. Petre, Usability analysis of visual programming environments: a ‘cognitive dimensions’ framework, Journal of Visual Languages and Computing 7 (1996) 131–174. [2] D.G. Hendry, T.R.G. Green, Creating comprehending, and explaining spreadsheets: a cognitive interpretation of what discretionary users think of the spreadsheet model, International Journal of Human–Computer Studies 40 (6) (1994) 1033–1065.
ARTICLE IN PRESS Foreword / Journal of Visual Languages and Computing 17 (2006) 288–291
291
[3] B. Nardi, A Small Matter of Programming: Perspectives on End-User Computing, MIT Press, Cambridge, MA, 1993. [4] C.D. Hundhausen, Using end user visualization environments to mediate conversations: a ‘communicative dimensions’ framework, Journal of Visual Languages and Computing 16 (3) (2005) 153–185.
Thomas Green School of Computing, University of Leeds, Leeds LS2 9JT, UK E-mail address:
[email protected]