A theory of presentation and its implications for the design of online technical documentation
©1997 Detlev Fischer, Coventry University, VIDe (Visual and Information Design) centre
The previous chapter has shown implications of the theory of presentation for document system design. This conclusion is not a summary of the theory—such a summary is provided at the beginning of chapter 1–Presentation. Instead, this conclusion attempts to extend the scope of the theory to conflicts that appear between presentation and its context. It then draws implications for research into on-line systems, both concerning a re-orientation of research design and document system research. The conclusion ends with comments on grounded theory method.
This study has repeatedly discussed one fundamental problem for design: that it is forced to draw a boundary which defines its scope according to projected needs (and by the same token, excludes unforeseen needs). The finite nature of the document implies that it can only account for the predicted and therefore ‘schedulable’ situation or process. The field work however has shown that the actual problem often does not fit the prediction and prescription set out in documents.
Here, emergent presentation comes in to deal with all the contingencies, exceptions, deviations, anomalies, and breakdowns in the domain. The navigated document is often just a starting point for or scaffolding of confluent presentation, which draws on a host of other situated and articulated resources to work out what is going on. In these situations, failure of presentation is often due to the lack of confluence between independent resources which are not subject to a unified design approach that would consider them as part of one activity. This indicates the importance of the scope of design decisions which in a competitive environment is narrower than it might be. Examples are the ATA-numbering conflicts introduced through different coding regimes at Airbus and Boeing, or the costly failure experienced at Lufthansa of a system that intended to use just one Document Type Definition per document type to integrate SGML-encoded documentation from different manufacturers.
Presentation deals with processes and problems as they are, not how they are planned or expected to be. This creates a conflict with the view of the world as encoded in protocols and schedules, and this conflict is recognised in the field. The power of protocols shows when they subvert presentation so that it hides or transforms problems into artefacts that simulate compliance. For example, the protocol which requires to produce documents which assess the likelihood of foreseeable engine failures and specify measures to contain them safely is regarded as a paper exercise which limits the scope of risk to technical features. The protocol according to which line maintenance follows written procedure is impossible to implement under the response pressure of the actual work context.
The signature is the artefact that links the world as it ought to be according to protocols and schedules, and the world that is. Line maintenance mechanics sign off the job and thereby the protocol that they followed written procedure although they did not. Doctors sign that they have taken the 5-hour break required by the schedule laid out in the work contract although they have been working all night [1].
The link between emergent breakdown and fitting documentation is the weak spot of system design. The fault, referent of derivative and often multiple symptoms transformed by several systems, must be understood by a human and re-coded in order to provide the right point of entry into documentation. In re-coding, the protocol constrains articulation so that the emergent resource becomes a pointer for navigation and is itself navigable. Such standardised reporting may decide about the success or failure of on-line document systems such as BISAM [2].
The theory of presentation suggests a re-orientation of design from the individual document system to the overall resource [3]. The focus on the individual resource leads to technology-oriented claims such as that of an increased ‘bandwidth’ through on-line document systems and ignores the effects both of other confluent resources and of concurrent but independent tasks in the background [4].
The overall resource can only be found in the field. Academics' document system prototypes—including the cinegram—are usually designed in a remote and protected domain. Most prototypes are never evaluated in other than artificial laboratory settings or during short stints of field work. The best way to change this situation is for academic researchers and designers to research and prototype their designs in the context and setting for which it is intended. This ensures that the interaction during presentation between design prototype and confluent other resources becomes visible. It also provides the designer with valuable suggestions and critique by users. The designer ‘going native’ may also extend the scope of design beyond the design of novel resources by experimenting with resources already available in the setting. Users may agree to temporary experiments which may be as simple as putting the interface to a rarely used on-line resource in close reach. Another experiment might involve networking users' isolated PCs in order to afford on-line confluence of already existing disparate user designs such as Excel engine history databases. Locating research in the setting would allow a longitudinal view of resource use and would reveal dimensions of the organisation's hidden situation [5] which do not show during purposeful and exceptional research activities by outsiders.
The outlined strategy of locating the academic researcher/designer in the professional setting depends on an advance agreement between the industrial and the academic partner about financial and logistical aspects. It is less of a problem for those part-time students who are already employed by the organisation where they conduct their research. But the strategy might also be beneficial for in-house system design. Locating system designers in users' setting will counter the proverbial ignorance of organisations' IT departments to concerns of their users. Putting IT professionals outside their physical department has the drawback of reducing confluence between them. However, IT professionals should be in a better position than most other designers to compensate partly for the lack of a shared physical setting. They can share their work on-line by using electronic conferencing systems and tools for computer-supported collaborative work.
Placing the designer in the setting creates the conditions in which design might become a confluent presentation between user and designer. A possible drawback is that designers may become too involved in users' immediate needs to perceive of design solutions on a higher level of abstraction. This may be prevented through job rotation, peer reviews, and supervision.
The frequent confluence of navigated and articulated resources in users' presentation indicates the importance of document systems which can mediate—technically, link—between the two classes of resource. Possible models are multi-levelled annotation mechanisms [6] and mechanisms that link between hypertext and relational databases [7].
One consequence of the apparent weakness of design prediction is to cede more design decisions to the users of document systems. One aspect of such control is the adaptability of document interfaces. The theory of presentation indicates that adaptation mechanisms should be controlled by the user's, not the system's agency [8].
Domain-specific frequencies of presentation can inform design decisions on domain, group, or individual level—for example, about points of entry to on-line systems. On the organisational level, there may be generic points of entry to the-engine documentation, which on the group level are supplemented by points of entry to this-engine's case history, and, on the individual level, by personal collections of resource pointers [9]. There is uncharted territory in exploring links between the navigable generic resource and users' emergent articulation. On-line systems have an advantage here since the emergent document can directly link into the generic system without jeopardising its integrity—but only if version control secures the integrity of the target in the generic resource [10].
In this study, the use of the grounded theory method was extended by the design of an artefact, the cinegram prototype, which was introduced into the researched domain as a catalyst. This conscious intervention makes it even more necessary to reflect the research need and integrate it as a condition of theory generation into the theory (cf. section 8.1 .3 Designers' requirements in chapter 8–Design). The opposite approach is often followed when the academic protocol turns the publication into a presentation artefact. This is expressed in the commonplace (and half-apologetic) joke that researchers massage the evidence until it fits their preconceived hypotheses. Methodology is presented as a priori framework which reifies the genesis of the publication. Writing makes extensive reference to the accepted literature [11] and avoids the first person singular in writing. Grounded theory was liberating in that it discards the authoritative stance of science while providing an alternative process-oriented approach to theory generation.
Grounded theory generation benefits from teamwork. This has also been stated for related qualitative approaches [12]. Fostering teamwork suggests a strategy of organising academic research as group activity [13]. A project and budget provides the material basis for the lasting teamwork of several researchers so that they can learn to work together productively and become sensitive to methods of data gathering and analysis.
In this study, teamwork was limited to occasional discussions with individuals involved in the evaluations, open coding by one colleague (cf. section Sources and coding in appendix VI–Grounded theory applied), and the resonance from friends and colleagues to the first and second draft of this study.
Teamwork has an important corrective potential. It speeds up the coding process since coding someone else's notes implies sufficient distance which only grows slowly to one's own notes and memos. Teamwork also helps correcting assumptions and fills in what the individual researcher might have missed through the resonance of idiosyncrasies or due to blind spots created by the lack of resonance. For example, in this study, the concept of validation context first appeared in codes such as ‘awareness of task artificiality’ suggested in open coding by a colleague. It took much time and reminders from others to make me aware of the importance of the validation context of presentation: I was too close to the validation context of the evaluation domain which I had implicitly defined in the choice of evaluation method, and too remote from the validation context of the referent domain of service engineering. The fact that given enough time, grounded theory method leads to the emergence of important concepts against researchers' predilections and blind spots is a clear indication of its strength.
The research project involved only one full-time researcher. The data generated through fieldwork and evaluation were rich and heterogeneous while they were limited in scope and in terms of variations of perspective. For example, intensive field work focused on one group within service engineering while covering customers and referent domain only through interviews with managers and documentation experts at several airlines. Variations of perspective were limited to the mediating role of witnesses, discussions with peers, and various comments to draft versions of the thesis.
However, these limitations do not impinge on the validity of the resulting theory since it is seen as hypothetical and open for change (cf. subsection 2.2.7 Scope for change in chapter 2–Methodology). In this view, a publication such as this thesis is just an intermediate step in a continuous research process which can do no more than offer a slice of a researcher's current understanding. The theory ‘closes out’ by saturating those concepts that have been generated based on the available data. It is clear that further data collection and analysis can change the emphasis of the theory and produce new dimensions and interrelationships. It may even extend or shift the boundary of the referent domain of research, as it happened in this research project (cf. section 2.1 Choice of methodology in chapter 2–Methodology). The constant grounding of concepts in the data however ensures that the slice of theory is unlikely to be invalidated even when its concepts later come to be seen as dimensions of more general concepts or when their interaction reveals new dimensions which were not seen at the time of writing.
[1] Thiesemann (1996), personal communication.
[2] Orlowski (1996), personal communication.
[3] This is supported by Newman (1995) who points out that research into multimedia systems is biased in favour of inventing new paradigms, interfaces and techniques and neglects evaluation which would reveal why many existing multimedia systems fail, and sometimes fail disastrously.
[4] Allen (1989 p61) quotes Rockwell, Giffin, & Romer (1983) who found that ‘pilots became so engrossed in their attempts at diagnosis [of a faulty auto pilot] they paid no attention to their flight position.’ Cf. also the report on the incident of the Airbus A340-311, G-VAEL at Heathrow airport, 19 September 1994, which states that ‘there was insufficient time for the crew to attempt to restore the EFIS displays’. (AAIB Bulletin No. 3/95, Ref: EW/C94/9/2. 1994).
[5] Leithauser (1986, quoted after Adler & Winograd 1992 p156) argues that ‘there exists an informal "hidden situation" in every organisation, largely unrecognized by external system experts.’
[6] Cf. Rheinsperger (1992).
[7] Cf. Lindsay's (1991) discussion of the difficulties of integrating the relational database paradigm and the object-oriented paradigm.
[8] In their discussion of self-adaptive interfaces, Browne et al (1991) point out the hunting problem which relates to the agency conflict described in chapter 8–Design: while the system is trying to establish a user model, the user is trying to establish a system model. The resulting interference mars both attempts.
[9] Cf. Monk (1990).
[10] On versioning problems cf. Halasz (1988) or Levy (1994). Davis et al (1994 p49) assume in blissful innocence that ‘changing [a] spreadsheet automatically causes the system to send e-mail to the interest group informing them of the change’. It will not be as easy as that.
[11] Cf. Glaser's (1978 p137) critique of the attitude of adumbration: ‘It is amazing how many authors try to find their best ideas in previous work in order to legitimate using it, as borrowed or derived as if they could not be allowed to generate it on their own.’
[12] Cf. Kleining's (1994 p28) suggestion to use small teams of observers/ researchers to maximise the variation of research perspectives.
[13]The academic protocol only requires that ‘the programme of research to be undertaken by the applicant for registration must, in itself, be distinguishable for the purpose of assessment…’ Research degrees handbook (1993). [my italics].