Description of a process of clustering of some 450 issues identified in the Brundtland Report, Agenda 21, and in sectoral declarations (see checklist). The coding procedure is described elsewhere.

Comments on Procedure

The initial objective was to devise a means of comparative processing of sectoral declarations in order to clarify the positions of agreement and disagreement amongst sectors on the range of Earth Summit issues. To this end a range of declarations relevant to the UNCED process were collected (see checklist). On inspection it was decided that they did not readily lend themselves to a detailed comparative analysis. There were two principal reasons. Firstly, the documents were of a variety of forms which would have complicated any attempt at comparison -- a number were not in fact "declarations". Secondly, the time and resources finally available did not permit any detailed review.

In the light of the work on strategic dilemmas described in the Figure 1 . The range of issues was then extended by incorporating the broader problems registered in the Encyclopedia of World Problems and Human Potential (1991) since these had been derived directly from documents of international organizations and ordered into hierarchies and networks, thus clarifying a number of terminological issues. This facilitated nesting of issues in terms of broader and narrower concepts.

The revised checklist was then matched against the chapter headings of Agenda 21 and against the issues recognized in individual declarations. The documents processed are described in the following document (see checklist).

To facilitate the match sought with Figure 1, the identified issues were coded according to the dimensions of Figure 1. Some remarks on the coding problem are given in a related document. The results were presented in two documents:

• Issues clustered by principal subject (in 6 parts)
• Issues sorted by code combinations

The coding and final form of these tables were achieved as the result of a number of iterations.

It must be remembered that the purpose of this exercise was to identify the issues and not to determine which bodies had recognized, implicitly or explicitly, which issues. For this reason the indication of source was made according to the following guidelines:

• BR (Brundtland Report) was left as an initial source code
• Items from the Encyclopedia were left uncoded
• A21 (Agenda 21) as a source replaced the above
• New issues from declarations cite the document in which they were first encountered, replacing Encyclopedia but not BR or A21
• If a more recent document provided a more articulated pattern of issues, its source may have been cited in preference to any previous non-BR or non-A21 source

It is important to recognize that no issue indicates the complete range of sources citing it.

Comments on Coding Sustainable Development Issues

The comments below relate to the coding of issues identified in the documents described elsewhere with a view to their presentation in tabular form.

1. It is important to recognize that any form of classification can never be politically neutral. Classification is a political act. Many classification systems have in-built biases against either developmental or environmental sensitivity, or both. How issue complexes and agenda topics get clustered and ordered is not a trivial matter without consequence for the sustainability of policies.

2. It is important to review how general topics get regrouped into clusters, whether for major reports (like the Brundtland Report), in agendas (like Agenda 21), or in specific declarations and policy statements. The question is how clusters "break-out" of more general topic continua.

3. The difficulty in clustering and coding is much more at the general level than at the specific level. Much work has been done at the specific level where hierarchies of topics are relatively uncontroversial. Little work has been done at the general level where clustering tends to be somewhat simplistic.

4. In any clustering exercise, there is a challenge in distinguishing (or weighting) attributions to primary, as opposed to secondary, clusters. Distortions in the allocation of issues to primary clusters in the checklist (Figure 9) are corrected by the codes which indicate relationship to other clusters.

5. There is a continuing challenge in the ambiguity of the significance of words conventionally used to label issue complexes. Some labels for agenda items are deliberately designed to obfuscate. The intended scope of a label may vary according to the political faction referring to it.

6. Some issue complexes are more clearly understood in terms of the "problems" with which they are commonly identified rather than through the strategies or "solutions" via which they are approached. In such cases the problem statements are often at a more general level than the solution statements.

7. In clustering in terms of generality and specificity, there is a challenge in shifting significance to more general or to more specific clusters.

8. It is useful to review the coding exercise in the light of the metaphors of "chemical valency" or of "genetic codes." Thus issues may usefully be seen as having as many "valencies" as there are codes attached to them (eg an issue complex coded PER is therefore 3-valent; more complex issues may be 5-valent). In the case of the second metaphor, just as there is an appropriate genetic code governing the sustainability of living organisms, it could be argued that there is an appropriate "genetic code" governing the sustainability of social structures.

9. In any coding/clustering process there is a balance to be struck between imposing order and responding sensitively to variations. This may be seen as a balance between the need for restraint and the need for freedom.

10. It is useful to perceive the issues of sustainable development as implicit in the challenge of producing a sustainable classification. What may seem credible in the short-term in the light of immediately obvious requirements may be far less appropriate in the light of broader and longer-term issues.

11. The coding process needs to be sensitive to conflicting perceptions, whether in terms of "problems" as opposed to "solutions", or in terms of "reality" vs "vision". These variants are all required to capture the challenge of sustainable development. Stressing one at the expense of another does not meet the needs of all involved.

12. The coding/clustering challenge may be seen as akin to that of designing a 3-dimensional crossword puzzle (involving features reminiscent of a good computer game such as TETRIS).

13. The identification of clusters appropriate to the dimensions of sustainable development corresponds in many respects to the division of governmental responsibilities into ministerial portfolios and government departments. Any clustering of issues and strategies should be challenged by the divisions of governmental responsibility -- and vice versa.