We round off this conclusion by highlighting four additional suggestions for further research:

We are not aware of other investigations that have tried to link research integration elements and dialogue methods and have been pleased to discover through our work that this is an area that has promise. We were struck by the potential of the dialogue methods we examined to focus on and strengthen particular aspects of research integration. As we discussed in chapter 5, more work is needed to tease out which methods are most suitable for particular research integration tasks. In addition, there are areas—such as the integration of visions, world views or interests—where we have been able to identify only one method and where the development of additional techniques is likely to be valuable. Furthermore, our analysis of the elements of knowledge to be integrated was simple and pragmatic and remains open for more sophisticated development.

We note that, with the exception of the Delphi technique, none of the other methods we describe could be illustrated by examples in every one of the four areas of application that we chose to investigate: the environment, public health, security and technological innovation. While this could be an artefact of our search strategy, we suggest it is more likely that formal dialogue methods are not yet exploited to their full potential in research integration. Cross-fertilisation between these four (and other) areas is likely to have at least two substantial benefits. First, it will draw the attention of researchers in a particular area, such as security, to the potential of methods new to that area. This not only increases the methods repertoire of security researchers, it could stop them from reinventing the wheel if they decide a new method is needed. Second, documentation of experiences in different areas could provide insights into useful modifications in applications of particular methods, as well as specific dos and don’ts that are key to the method’s success.

One of the other benefits of cross-fertilisation is that it can alert researchers to the importance of flexibility in how methods can be used singly and in combination. We pointed out the importance of flexibility in our introductory sections and re-emphasise it here. The hallmark of excellent experienced research integrators specialised in dialogue methods^[1]will be their ability to mix and match methods as the needs of a particular research problem require. This was drawn to our attention by two of the people we asked to review an earlier draft, Gerald Midgley and Wendy Gregory, whose systemic intervention practice (see Midgley 2000) illustrates these principles. This could be taken even further to an appreciation of the value of improvisation once a dialogue process is in full swing. While preparation for dialogue events is essential, surprises can occur once the group convenes. Flexibility and improvisation are important not only in choosing and adapting particular methods, but in the areas of making groups ‘work’, which we do not deal with specifically in this book. One of the key lessons provided by the teaching of improvisation to jazz students is the importance of making explicit knowledge tacit (Bammer and Smithson 2008; Mackey 2008) so that when the time comes to ‘take a solo’, the performer can organically draw on that store of knowledge. As Mackey (2008:107) has described, the ability to improvise in jazz is built by internalising explicit and performance knowledge about ‘accentuation, articulation, tempo, ornamentation, rubato, melody and rhythm’. We suggest that flexibility and improvisation are also areas for further development in research integration using dialogue.

Recognition that dialogue is an art should, however, not diminish attention to the twin aspect of dialogue as a science. In the description of the failure that we present above, it is clear that dialogue is not a case of ‘anything goes’. There will be parameters that determine success or failure and teasing those out is critical to the future of dialogue as an effective research integration method. Two important areas to explore are bias through participant selection and maximising the benefits of conflict.

With regard to the first of these, there is need for greater consideration of the potential for bias through the selection of participants. While some dialogue methods, such as the citizens’ jury, try to ensure that their participants are as representative as possible of the population of interest, the potential for bias seems not to be considered in the application of other methods.

The second area involves appreciating the importance of conflict and different strategies for maximising the benefits of conflict (and minimising the costs). It is likely to be an implicit assumption that dialogue should lead to consensus, but in fact that is not necessarily achievable, let alone what should be aimed for. A few of our examples (such as the future search conference on reducing the human and economic costs of RSI and the use of strategic assumption surfacing and testing in a US Cooperative Development Agency) demonstrate non-consensual outcomes, and there is substantial scope for further investigation of this area.