1. Reasons for not using alphabetical order
(a) Many of the entries have names associated with them that are unsuitable as a basis for alphabetical ordering. The names may be derived from the description (of the problem, concept, etc) and may later be revised by substituting better alternatives. Some are long and contain many prepositions. Some names are artificial and not commonly accepted or recognized (eg those names in the world problems series starting with "inadequate"). Some entries have several possible names associated with them.
(b) Change of the name under which the entry is filed would result in the need to move the whole entry within the sequence of entries when an edition is revised. This is inconvenient for the user who does not have a permanent point of reference. It makes editorial control complicated, leading to cumbersome administrative procedures and to increased errors. It also changes and complicates the whole philosophy of file management when using computers, as well as increasing computer processing costs.
(c) If an edition (or an index) in another language is prepared, all the material must be reordered. This may necessitate the creation of a new set of indexes and cross-references. This leads to total lack of correspondence of the sequence of material between language editions. With the chosen policy, there no change of order between language editions, nor any change of indexes or cross-references. (Problem PA1234 in the English edition would be Problem PA1234 in a French edition).
(d) Use of alphabetical ordering artificially places entries next to one another on the basis of a chance superficial resemblance arising from the first words in the entry name. This may reinforce incorrect associations, or may imply to the reader that all related entries are together and that the problem of classification has been solved. With the chosen policy, readers are constantly reminded that the classification problem is an ongoing problem that is not solved by any particular classification scheme.
(e) Cross-references within entries, and indexes to entries, are more conveniently made to a number than to long alphabetical titles with many prepositions.
(f) Placing the emphasis on the name of an entry by the use of some alphabetical order prevents readers from being confronted with the fact that they are faced with a conceptual domain that may be ill-defined and labelled with many different names. By using a number, the same name may even be attached to different numbers used to denote different conceptual domains, or at least domains that are believed to be distinct.
2. Reasons for not using a classification scheme
It is interesting to note that a principal interest of academics in this project has been the choice of classification scheme (eg for the world problem series). A frequent argument is that it is impossible to collect information unless it is done through some classification scheme. The difficulty in this project (collecting information on world problems, for example) is that at no stage was it known what new variety of problem would emerge from the documents processed on the following day. Therefore at no stage was it appropriate to develop a definitive classification scheme. To do so prior to publication would deprive scholars of the pleasure of exposure to conceptually unordered data by supplying them with a classification scheme that it would be only too easy to criticize. Specific reasons for not developing any such scheme at this stage are as follows:
(a) The first problem in handling such varied material, is one of filing the entries and associated documents. This is a simple administrative task to which there are a limited number of viable solutions. The simpler the system, the simpler the administrative task. The classification of entries is a sophisticated intellectual exercise that ideally never ceases, since science is always in search of better categories and improved ways of structuring the relationship between elements of knowledge. There are many solutions to the problem of classifying any given series of entries. If a particular scheme of categories is chosen for a given series of entries, the filing problem is complicated and the administrative and intellectual activities interfere with each other. The conceptual exercise should follow the filing of the information, not accompany it, nor precede it (leading to the exclusion of unforeseen varieties).
(b) A particular classification scheme is always subject to criticism, pressure for revision, or the substitution of some alternative scheme. There is no consensus on a satisfactory scheme for classifying any of the series in this publication. Such revision again represents an unnecessary interference between the administrative task and the intellectual task. If a computer system is based on a non-permanent code/number scheme, considerable difficulties arise.
(c) Many of the entries included (eg the integrative concepts) have been included precisely because they are "fuzzily" defined and cross conventional subject and disciplinary boundaries and are therefore difficult to classify. The same could be said for the more complex problems. Use of simplistic classification schemes to eliminate this difficulty would only serve to disguise this and would thus be a disservice to all concerned.
(d) Structuring a publication in terms of a particular classification scheme only serves to alienate unnecessarily those who are intellectually frustrated by the idiosyncrasies of such a scheme.
(e) Even in the case of physically unambiguous entities like plants and animals, there is still considerable dispute about the allocation of particular entities to particular categories. There is even a school of thought that maintains that the classification enterprise itself is counter-productive. The situation is understandably more serious in the case of conceptual entities.
(f) As stressed at a number of points in this volume, a conventional classification scheme would reinforce the mind-set that the entries here could be classified without doing violence to the relationships between them. For this reason the emphasis has been placed on exploring ways of portraying the network of relationships and using this as a basis for addressing the question of how to identify significant patterns of relationship.
(g) By holding the entries in series on computer files, the subsequent intellectual operation of experimenting with a variety of different classification schemes is facilitated. The results of a number of such experiments can be incorporated as special indexes, regrouping the names of the entries, but without interfering with other experimental classifications or with the information as filed. Several experiments of this type have been included in this edition and are discussed below.
3. General principles
Since it may be argued that there is some resemblance between the task of classification of diseases of the human being and the task of classification of the problems of human society, it is useful to consider the general principles governing the former as printed in the introduction to the World Health Organization's International Classification of Diseases (Geneva, WHO, 1967). The section in question reads: "Classification is fundamental to the quantitative study of any phenomenon. It is recognized as the basis of all scientific generalization and is therefore an essential element in statistical methodology. Uniform definitions and uniform systems of classification are prerequisites in the advancement of scientific knowledge. In the study of illness and death, therefore, a standard classification of disease and injury for statistical purposes is essential."
"There are many approaches to the classification of disease. The anatomist, for example, may desire a classification based on the part of the body affected. The pathologist, on the other hand, is primarily interested in the nature of the disease process. The clinician must consider disease from these two angles, but needs further knowledge of etiology. In other words, there are many axes of classification and the particular axis selected will be determined by the interests of the investigator. A statistical classification of disease and injury will depend, therefore, upon the use to be made of the statistics to be compiled.
The purpose of a statistical classification is often confused with that of a nomenclature. Basically a medical nomenclature is a list or catalogue of approved terms for describing and recording clinical and pathological observations. To serve its full function, it should be extensive, so that any pathological condition can be accurately recorded. As medical science advances, a nomenclature must expand to include new terms necessary to record new observations. Any morbid condition that can be specifically described will need a specific designation in a nomenclature.
This complete specificity of a nomenclature prevents it from serving satisfactorily as a statistical classification. When one speaks of statistics, it is at once inferred that the interest is in a group of cases and not in individual occurrences. The purpose of a statistical compilation of disease data is primarily to furnish quantitative data that will answer questions about groups of cases.
This distinction between a statistical classification and a nomenclature has always been clear to medical statisticians. The aims of statistical classification of disease cannot be better summarized than in the following paragraphs written by William Farr a century ago:
"The causes of death were tabulated in the early bills of Mortality (Tables mortuaires) alphabetically; and this course has the advantage of not raising any of those nice questions in which it is vain to expect physicians and statisticians to agree unanimously. But statistics is eminently a science of classification; and it is evident, on glancing at the subject cursorily, that any classification that brings together in groups diseases that have considerable affinity, or that are liable to be confounded with each other, is likely to facilitate the deduction of general principles. Classification is a method of generalization. Several classifications may, therefore, be used with advantage; and the physician, the pathologist, or the jurist, each from his own point of view, may legitimately classify the diseases and the causes of death in the way that he thinks best adapted to facilitate his inquiries, and to yield general results. The medical practitioner may found his main divisions of diseases on their treatment as medical or surgical; the pathologist, on the nature of the morbid action or product; the anatomist or the physiologist on the tissues and organs involved; the medical jurist, on the suddenness or the slowness of the death; and all these points well deserve attention in a statistical classification. In the eyes of national statisticians the most important elements are, however, brought into account in the ancient subdivision of diseases into plagues, or epidemics and endemics, into diseases of common occurrence (sporadic diseases), which may be conveniently divided into three classes, and into injuries, the immediate results of violence or of external causes."
A statistical classification of disease must be confined to a limited number of categories which will encompass the entire range of morbid conditions. The categories should be chosen so that they will facilitate the statistical study of disease phenomena. A specific disease entity should have a separate title in the classification only when its separation is warranted because the frequency of its occurrence, or its importance as a morbid condition, justifies its isolation as a separate category. On the other hand, many titles in the classification will refer to groups of separate but usually related morbid conditions. Every disease or morbid condition, however, must have a definite and appropriate place as an inclusion in one of the categories of the statistical classification. A few items of the statistical list will be residual titles for other and miscellaneous conditions which cannot be classified under the more specific titles. These miscellaneous categories should be kept to a minimum.
Before a statistical classification can be put into actual use, it is necessary that a decision be reached as to the inclusions for each category. These terms should be arranged as a tabular list under each title, and an alphabetical index should be prepared. If medical nomenclature were uniform and standard, such a task would be simple and quite direct. Actually the doctors who practise and who will be making entries in medical records or writing medical certificates of death were educated at different medical schools and over a period of more than fifty years. As a result, the medical entries on sickness records, hospital records, and death certificates are certain to be of mixed terminology which cannot be modernized or standardized by the wave of any magician's wand. All these terms, good and bad, must be provided for as inclusions in a statistical classification.
The construction of a practical scheme of classification of disease and injury for general statistical use involves various compromises. Efforts to provide a statistical classification upon a strictly logical arrangement of morbid conditions have failed in the past. The various titles will represent a series of necessary compromises between classifications based on etiology, anatomical site, age, and circumstance of onset, as well as the quality of information available on medical reports. Adjustments must also be made to meet the varied requirements of vital statistics offices, hospitals of different types, medical services of the armed forces, social insurance organizations, sickness surveys, and numerous other agencies. While no single classification will fit the specialized needs for all these purposes, it should provide a common basis of classification for general statistical use.
4. Anti-developmental biases in classification scheme designs
In the conventional western approach to the design of classification schemes, a number of biases seem to emerge. These biases are inherently anti-developmental. The effects are particularly serious in the social science domain. The biases have been discussed elsewhere (Anthony Judge, Anti-developmental biases in thesaurus design, 1981). They may be summarized as follows:
(a) Static bias associated with noun categories: Most classification schemes are concerned solely with ordering nouns or objects (called "subjects"). This emphasizes static, structural mind-sets, whereas the fundamental characteristic of development is change and movement. Thus, in the 1986 edition, to de-emphasize this static approach, strategies (Section S) were frequently named using a gerund form to stress their dynamic, action-oriented character -- this practice is continued in current strategy profiling
(b) Low-context bias associated with western science: Western science is deeply preoccupied with specifics and classification systems are designed to handle them. Such systems are consequently not designed to reflect high sensitivity to context as is to be found in non-western cultures. And yet it is precisely such context sensitivity which is required to integrate the fragmented perspectives and engender more coherent approaches to seemingly unrelated issues.
(c) Pattern conservation bias: Classification schemes are usually designed on the assumption that they can grow by extension of a pre-defined pattern and not by transformation of that pattern. This is precisely the kind of thinking which reinforces the present non-transformative character of development.
(d) Dysfunctional bias: Most classification schemes are insensitive to the functional relationships between the phenomena classified. The fact that mercury may infiltrate food chains to affect seriously the survival of a bird species is not something a thesaurus is designed to highlight.
(e) Insensitivity to wider repercussions of classification schemes: Design of a classification scheme is seldom considered in terms of non-library users whose methods of organization will be reinforced by it, whether appropriately or inappropriately. This may be seen in: bookshop layout, agenda design, curricula design and organization charts. A mechanistic classification scheme reinforces mechanistic thinking in relating subject areas.
(f) Avoidance of top-of-hierarchy issues: Most of the effort in classification system design is directed to clarifying problems within particular domains. By contrast little effort is directed toward clarifying relationships between the major hierarchies within which this effort is made.
(g) Preference for adaptive "maintenance" thesauri: In the light of the Club of Rome distinction between "innovative learning" and "maintenance adaptive learning", most classification systems are designed after the fact in response to old issues. Every attempt is made to fit new issues (eg environment) into old frameworks. No provision can be made for the next interdisciplinary crisis that will call for information to be organized in a new way.
(h) Investment in rigid, anti-experimental systems: Classification systems, once implemented, are extremely costly to modify them. They are not designed to facilitate flexible experiments in reconfiguring patterns of categories.
(i) Concealment of contradictions: Every effort is made to render classification systems as impersonal as possible, de-emphasizing any personal or subjective elements. This both conceals the biases of the designers and avoids the need to classify biases more explicitly. And yet it is such biases that are a driving force behind developmental processes and problems.
(j) Concealment of values: The assumption is widely made that classification systems are value-free. A thesaurus that treats the real-world experience of "homelessness" as an aspect of the academic discipline of "sociology", and treats "war" as an aspect of the discipline of political science is taking a strong political position. A totally exploitative attitude towards the environment is suggested by the inter-agency institutional information system based on a macrothesaurus with categories of "fisheries", "fishing" and "fish processing", but not "fish" (as having a role in their own right within planetary ecosystems), nor over-fishing. It is no wonder that there has been so little institutional sensitivity to the ecological problems of deforestation.
5. Classification experiments
In this Encyclopedia several experiments in classification have been explored in this, and earlier, editions:
(a) Interfacing with international classification systems: For the 1976 edition four sections were organized on the basis of the adaptation of four specialized international schemes. Traded commodities and products (Section C) were organized according to the United Nations Standard International Trade Classification. Economic and industrial sectors (Section E) were organized according to the United Nations Standard International Classification of all Economic Activities. Occupations, jobs and professions (Section J) were organized according to the Standard Classification of Occupations of the International Labour Office. Human diseases (Section Q) were organized according to the International Classification of Diseases of the World Health Organization.
(b) "Pragmatic" approach: The range of international organizations in the Yearbook of International Organizations raises similar classificatory problems. For several editions, a simple pragmatic scheme has been used to distinguish international organizations by "levels" of internationality. The possibility of adapting this scheme to world problems was explored in the 1986 edition (Section PX) using qualifier codes. This system has been further developed in this edition to split the world problems into Sections PA through PJ.
(c) Patterns of concepts: In both the 1986 and current editions, small experiments have been made in relying on patterns of numbers to order concepts (Section CP, 1986; Section KP, current). The use of taoist system of ordering has also been explored (Section CP, 1986; Section TP, current). Many of the entries on human development (Section H) form part of sets, or series, of concepts or modes of awareness. As stages they are numbered in their respective traditions. In the Buddhist tradition, for example, indexes based on such sets are occasionally made. This approach has been explored in this edition for the entries in Section H in the form of a special index (Section HX) ordered by number.
(d) Levels of awareness: In the current edition the possibility of grouping modes of awareness into levels has been explored by the use of a qualifier code (a through g), which is described in the relevant Notes (Section HZ). But, for lack of space, entries have not been extracted in this order.
(e) Human values: Both in the 1986 and current editions, the results of experimenting with classifying "value complexes" are presented in Section VT. This is based on clustering values, building on the organization of Roget's Thesaurus. See commentary
(f) Subject classification: Each year a more extensive classification experiment is undertaken with entries in this Encyclopedia, especially those in the World Problems section (Section P). In order to move towards establishing the links between international organizations and the world problems with which they are specifically concerned, both organizations and problems are classified by subject in the subject volume (Volume 3) of the Yearbook of International Organizations. This classification scheme, designed in detail in that volume, endeavours to highlight the interdisciplinary relationships so necessary to the cross-sectoral preoccupations of the international community.
This work is licensed by Anthony Judge
under a Creative Commons Attribution-NonCommercial-NoDerivs 2.5 License.