February 28, 1952
It has too often been implied that machine methods in indexing necessarily mean the replacement of conventional printed indexes by “machine” indexes utilizing IBM cards, microfilm, electronic memory units, or any combination thereof, etc. The following is an effort to erase this erroneous conclusion.

It has already been demonstrated at Welch Library in the “handling” of the Master Subject Heading List of the Current List of Medical Literature, that “machine methods” in subject heading studies can be employed to great advantage, without the replacement of conventional printed subject heading lists by a “machine” list. Machines are employed, and more specifically IBM machines, to facilitate subject heading studies.* Through the flexibility of machine operations, interesting and valuable studies have been made of subject headings that would not even be attempted by conventional methods.  indeed, the paucity of good subject heading analyses and/or lists is in great part due to the huge physical effort required to conduct such studies by conventional methods. Admittedly, with machines available one is often tempted to perform certain operations, which would never be tried by conventional methods and are actually not  pertinent—but once the novelty of machine operation has been dispersed with one usually confines himself to more practical and fruitful approaches.

Similarly, in regard to the general indexing problem -- if  machines can facilitate the compilation of conventional indexes they will have served a very useful purpose——perhaps more utilitarian and pertinent to the information searching problem than we imagine. If an organization with fixed resources can utilize machines to increase its service markedly the greatest obstacle to improving indexes will have been removed. If editors must constantly strive to meet deadlines and never reach the point of “complete” coverage, how can they be expected to devote themselves properly to the problem of improving the overall character of the indexes they produce?** This does not imply that there has not been steady improvement in indexing services, but rather this improvement has taken place in spite of the great variety of everyday problems confronting editors and their staffs. The author is convinced that when the “pressures” of compiling conventional indexes have been substantially reduced, then will we find much wider interest in the further possibilities of “machine” indexes.

The possibilities of conventional type indexes have by no means been exhausted. If one considers the resulting psychological ramifications of this situation, ‘it is dubious to expect “machine” indexes to make real progress before this exhaustion point has been reached. One can reduce most shortcomings of conventional indexes to 3 basic facets — incomplete coverage of the literature, lack of currency, and time required for extracting information indexed.   If an index can be said to give complete coverage of a subject field, is up to date, and requires a minimum of time to extract information — it has certainly approached perfection. (The use of “coverage” includes the usual sense as well as the most extensive cross filing of the information included in the index.)

The problem is essentially no different than that encountered in any expanding operation. A small merchant may do his own bookkeeping employing standard bookkeeping methods. A large corporation employs the same principles of standard bookkeeping practice but employs machines to handle the huge bulk inherent in such operations. When a corporation gets so large that even ordinary machine operations cannot give satisfactory and prompt results, electronic brains with even greater abilities are utilized. In scientific computation, electronic computers were preceded by electrical— mechanical computers which were preceded by mechanical adding machines, etc. As soon as one method outlived its usefulness, most of its possibilities bad been explored, new problems were tackled requiring machines with greater and greater capacities. Those of the library world are rightly envious of the mathematicians and their computers, electronic and otherwise. However, it is difficult to imagine “machine” indexes resulting from a sort of "mutation" — rather a stepped up evolutionary process seems more reasonable. Considering the conservative, cogitating figure that still dominates the library world anything more than a quick “evolution” is difficult to imagine.

While not losing sight of the immense possibilities of “machine” indexes the author will be content if this analysis by analogy has helped to overcome the “bogey” of the machine by placing it in its proper historical setting. Like the other social sciences, library science is far behind the pure and applied sciences. Those few “premature” dreamers (mostly science librarians and documentalists) who visualize complete libraries in a vest pocket should take consolation in knowing that unlike the other social sciences there are strong reasons to feel that at least we are catching up.

Once machines have helped facilitate conventional indexing procedures to the point where we can truly say we have reached the limit of the possibilities in conventional printed indexes — once the scare of machines has been overcome — then will machine indexes come into their own—not as immediate replacements for printed indexes, but first as supplements, then possibly as alternatives, finally as substitutes, taking a form not yet contemplated by the most imaginative thinkers.

*back to text   See paper on compilation of Subject Heading List by machine methods.
** back to text The case of QCIM is a good example -- sheer overwork and bulk has
      prevented this fine  indexing service from making any characteristic
      improvements for years.