Projects Overview (Explanations)
World Problems Project (Explanations)
Patterning problems: Concept refinement process
World Problems Project
The above procedure was initiated in 1972 and has increasingly become a purely "administrative" matter, especially since the proportion of new problems or new classes of problems continues to diminish. The ongoing concern is therefore much more with the conceptual processes whereby the "problem" associated with any given number is clarified through the naming (descriptor allocation) process. This may involve grouping different problems under one number or splitting one problem into several different problems. An important question is the clarification of relationships to more general and to more specific problems.
There are two major constraints to be borne in mind in the following discussion:
(b) The information gathered also imposes constraints on what degree of "concept refinement" is appropriate without distortion. This is a major issue given the differences in quality of information and the differences in interpretation of "facts" in different documents, even those supposedly of equivalent quality. It is especially difficult when the original information is vague and defines the problem using a loose terminology or when the terminology used is precise but applies to a more general or a more specific problem. It is obviously also difficult in the absence of adequate information on a problem whose existence is only suggested by the pattern of information in the database.
An overview of the concept refinement processes is presented in Figures 1 and 2 in the previous note. It is important to understand that many of these processes occur in parallel or are undertaken simultaneously by editors working on the network of problems from different computer workstations. Some are discussed below.
1. Problem identification
Once the bulk of the material has been filed by problem (as described above), the filed documents are then individually re-examined to locate descriptions of other problems which are then transferred to appropriate existing files or made the subject of new files. This is an integral part of the process of identifying relationships between problems. Scanning the material brings out possible variants on the name of the problem, which have to be distinguished from names which are more appropriately associated with other problems (which may need to be entered into the database). Variant names, even colloquialisms, may be vital as providing alternative means for locating the problem through the index. Also to be distinguished are various kinds of non-problem, according to the criteria guidelines, which appear in the documentation as problems or are identified by problem names. Such poorly-formed problems are coded as low-priority items in the database, not to be included in the published volume unless there is later reason to change their status.
2. Problem clustering
Because of the fuzziness of problem names, some of which take the form of phrases of some length, detection of duplicates is a major concern. The challenge is that similar names may have been associated with significantly different problem descriptions. The question then becomes whether to merge the two problems or to associate both names with one description and rename the other problem to distinguish it appropriately. Another challenge is the question of whether variants or aspects of a root problem should be distinguished as separate problems or merely as alternative names of that root problem. Here the decision may be strongly influenced by the amount of material available on a "variant", suggesting that it merits separate treatment. This tendency may be restrained to some degree by the process of systematization discussed below.
3. Problem (re-)naming
A high proportion of the documentation on problems makes use of loose terminology, especially in the case of the less tangible problems. Problems are often identified by positive terms more closely associated with the values attached to their solution. Implicit anthropocentric qualifiers may be omitted. An effort is therefore made to sharpen the negative aspects of the name to give greater focus to it. Clearer descriptor phrases may be discovered in the literature. Colloquial descriptors are added when they facilitate access to the problem. Different names may have been engendered by the need for politically correct euphemisms, eg "unethical" instead of "corrupt"; "unsalubrious" instead of "dirty". The ability to add alternative names is also used as a way of grouping into one problem the names of problem variants when it does not (yet) seem appropriate to treat these as separate problems, although it is useful to maintain an index trace on them. The significance of a collection of names attached to a problem thus varies depending on whether it can be considered well-defined or simply as a loose cluster of concerns around an ill-defined "proto-problem".
4. Adequacy of problem names
In the case of complex problems, the varying adequacy of sets of problem names may be illustrated by Figure 3 (below). There the problem to which reference is being made, to which the number refers, and which the descriptive text is intended to encompass, is indicated by a circle. The names associated with that problem are indicated by the lines, each line representing one name.
Fig. 3B: In this case the content implied is larger than that effectively encompassed by the set of names. The set does however share a common centre of reference with the content.
Fig. 3C: In this case the set of names implies more than the content. The problem is of lesser scope than the set of names implies. The names and content still share a common centre however.
Fig. 3D: In this development of the ideal case (Fig. 3A), implications of two of the names, combined with a third, serve to delineate an additional problem, distinct from the main problem. This suggests the need to split off the subsidiary problem, removing from the main problem the name which contributes to its enclosure.
Fig. 3E: In this case the set of names no longer shares a common centre with the content, and as such is no longer adequate. There are aspects of the problem which are not captured by the names, and the names imply a problem which does not exist.
Fig. 3F: In this case the set of names is inadequate to the content, even less so than in the case of Fig. 3B.
• Case of set of names without a circle, which would indicate names referring to a non-existent problem ("problem-less names");
Figure 3. Illustration of varying adequacy of sets of problem
names (lines) in relation to substantive content (circles)
There may be a case for using computer graphics to explore such patterning, with the object of determining whether there are ways of massaging the relationship between names and content into more ordered patterns. Of even more interest is the possibility of using a three-dimensional representation to extend the power of the geometric metaphor.
The geometrical convention may also be used to explore the ways in which underlying problems are conceptually contained. It provides a visual illustration of how problems may escape containment. As such it might be used as a way of mapping "problem-space" and the conceptual response to it.
5. Responding appropriately to source material
Much material available on problems makes what can easily be treated as exaggerated claims for particular problems whilst disparaging others. A careful balance must therefore be sought between sensitivity to unusual (even seemingly ridiculous) problems and detachment from the seemingly evident importance of currently fashionable problems (however eminent the legitimating authorities). This calls for openness to unforeseen perspectives and reservations concerning institutional and ideological hyperbole. Given the disproportionate amount of information on fashionable problems, this balance must be sought so as to give useful amounts of information on problems which may be of quite different significance.
6. Elaboration of logical context
Few, if any, problems exist in isolation. Many problems are defined as much by the context of other perceived problems as by any particular textual description that can be given to them. Most problems can be considered as "part of" some broader problem. Many problems therefore group together more specific problems. These relationships form problem hierarchies. Because of the variety of ways in which problems can be so grouped, any one problem may be part of several different broader problems. Effort must therefore be made to position a given problem within a context of broader problems, if necessary using it to group more specific problems. The problem needs to be named so as to distinguish it effectively from both the broader and the narrower problems. The process of "tidying up" such relationships is a continuing one, which may be dependent on the availability of some new document clarifying the pattern. Difficulties may be created because of conflicting perspectives in the literature on how certain problems should be grouped. At any particular point in time, there is always more that could be done which would take more time than resources permit. Some hierarchies must necessarily be left in a crude state. The logical framework around any problem may be more or less subject to further change.
7. Problem classification and section attribution
Because of the potential complexity of a problem's logical context, any effort to classify problems neatly by subject according to conventional principles has been resisted. An alternative approach has been developed which is used annually to group problems and organizations in Volume 3 of the Yearbook of International Organizations in a pattern of some 3,000 categories. In large part this classification is done automatically on the basis of descriptors in the problem title and constitutes one of the online editorial tools for exploring the database. The process of editing was further facilitated by attributing a section code to each problem. This is a pragmatic effort to group problems so as to distinguish those which tend to be higher in any logical hierarchy from those lower in the hierarchy (using the letter codes A through E). Other codes, notably F, are used for exceptional cases. During the editorial process the code attributed to a problem may be modified to position it more usefully within this pragmatic system, as well as to give it a lower editorial priority. (The significance of these codes is explained in more detail in the following note on Classification and section attribution).
8. Identification of functional relationships
Just as few problems exist in logical isolation, so few are independent of each other. Most problems aggravate some others directly, and in their turn are aggravated by others. Some problems reduce others, or are reduced by them.
This information may be present in the material collected on one problem thus enabling the relationship to be inserted. It may, however, only be present in the affected problem, so that the relationship gets automatically inserted in the former (given the relational nature of the database), even when there is no confirming documentation there.
Just as with the logical relationships, the documentation may suggest functional links to other problems at a level which is too general or too specific because the database structure is already more finely articulated. Again it is a matter of editorial judgement how this is to be handled. No relationships may be indicated in the documentation, in which case it may be appropriate to insert those that are obvious. In many cases, the insertion of obvious functional links is more appropriate than the use of "see also references" (which is effectively the significance of the "Related problems" relationship).
A major difficulty in inserting functional relationships is that of ensuring that they are made between the appropriate levels of problem hierarchies. Thus it is not useful to show all the narrower problems of Problem A as being aggravated by Problem B, when it suffices to show Problem A as being aggravated by B.
Whereas in some cases the logical and functional relationships between problems are reasonably clear and distinct, in others this is far less so. Problem A may both aggravate B and be aggravated by it. It may be far from clear of what problem Problem A is a part. These difficulties emerge particularly in the case of problems allocated to Section PF, and may be one reason why they are so allocated.
9. Confrontation with new documents
Since so many problems and relationships are now present in the database, an important process is that of challenging the existing database with new problem-focused material. Each such document may raise questions as to whether it:
• necessitates additions to the descriptors, namely does it suggest additional alternative names for the problem, or a correction to the existing name, or revision of the names attached to a whole set of problems;
• calls for the creation of a more general or a more specific problem, possibly requiring that some existing problem should have its scope and descriptors modified to make it more general or more specific;
• implies the existence of a whole new category of problems previously unrepresented in the database.
Although the prime emphasis is on registering those problems detected in available documentation, as these build up in the database certain patterns are suggested. The extent to which this patterning can be allowed to influence the inclusion or exclusion of problems is discussed in later notes.
The question is then to what extent the problem names should be rationalized. Again these are matters for editorial judgement, where the editorial resources are available. Some tentative steps have been taken towards such rationalization, but much more is called for as the opportunities become clearer. This process is important in preparing sets of problems to be checked from a particular perspective by external authorities.
11. Systematization of problem relationships
As noted above, when entering problem relationships into the database it is quite difficult to avoid redundancy. Computer programmes can be used to highlight certain redundancies, and even to eliminate some of them. But it is a matter of judgement as to how far this process can be allowed to go. For example, if 75 per cent of the narrower problems of Problem A are aggravated by Problem B, can these explicit relationships be replaced by a single relationship between A and B ? But what if the percentage indicating such aggravation is only 50 or 25 ?
12. "Massaging" the database
The point to be made is that editors are attempting to identify a pattern of significance in relation to each problem number, but the question always remains as to whether that pattern is stable and well-formed or whether some portion of that significance should not be moved to a distinct number. This depends not only on editorial judgement but on whether some constituency believes that a problem is distinct from the problem where it would otherwise be filed.
It should be stressed again that the database is designed to reflect not "facts" but perceptions of facts, however questionable they may appear in the eyes of others. The editors do not attempt to determine what is the most "authoritative" view, but rather what views are representative of significant constituencies (which normally would include those that are widely considered as authoritative). Decisions as to what are representative views, even if conflicting, are assisted by the international context within which the information is obtained.
13. Constraining the anticipatory function
In the light of the above points and the discussion of patterning in the following notes), to what extent should the database perform an anticipatory function? It is clear that seemingly credible problem names can easily be "generated" by appropriate combinations of descriptors from descriptor sets (in a manner reminiscent of Ramon Llull's category generator).
The availability of such descriptor sets encourages the recognition by the international community of problems which may or may not exist. Just as there are "letterhead organizations", there may be descriptor-generated problems (for which project funding may even by sought). Such problems may emerge into the literature as a result of efforts by delegations or the media to appear innovative. Given the institutional resistance to the recognition of new problems however, systematic explorations of such a procedure suggests an approach to identifying potential problems rather than allowing them only to be registered in a haphazard manner as information is found on them.
Perhaps category generators could usefully be developed to draw attention to categories of problem which may be neglected in conventional decision-making. In this project the procedure has only been systematically adopted in the case of economic sectors and endangered species. It is as a means of ordering problems at different levels of specificity, particularly in order to avoid isolated excessively-publicised entries (eg endangered panda, or endangered monkey-eating eagle) where more general entries would be more appropriate (eg endangered bears, or endangered birds of prey, respectively).
The question remains, however, as to whether the distinctions between problems, generated using different verbal "operators", correspond to valuable nuances with which people identify, or whether such distinctions are purely contrived in order to arouse emotive effects through their novelty.
This work is licensed by Anthony Judge
under a Creative Commons Attribution-NonCommercial-NoDerivs 2.5 License.