I came across an essay yesterday as I was reading the web: “As We May Think” by Vannevar Bush, originally published in The Atlantic Monthly in 1945. Bush’s central concern in the article is to encourage scientists, who had worked so well together in support of the war effort, to continue their collaboration in more peaceful endeavours, particularly in the area of information technologies, which he predicts will become increasingly significant. While he is unable, of course, to foresee the exact technologies that will enable the rapid changes in human relationships to information, he is remarkably perceptive in his understanding that these changes will produce a continual reduction in the size of information, a rise in technical languages to facilitate this reduction, the introduction of artificial information readers to make this reduced information humanly readable, and the use of recording technologies in cybernetic ways, among other things.
The article is worth reading as a whole, not least because it provides such an interesting insight into how the immanent information revolution was beginning to appear even in 1945, but I want to focus on a single phrase, by no means central to the essay, but interesting to me nevertheless. Midway through the article, in a section focussing on the processes that occur between data collection and the production of new knowledge, Bush says, “For mature thought there is no mechanical substitute. But creative thought and essentially repetitive thought are very different things. For the latter there are, and may be, powerful mechanical aids.”
What I think is interesting here is not the idea that creative thought or even mature thought, however Bush would define this, might be performed by a machine, which is now a science fiction commonplace, but the idea that the functions of the machine might be considered a substitution for these kinds of human thinking. Here, as with some of his other predictions, I think that Bush is not a little prescient. Whether or not he recognizes what he is predicting, he correctly projects that machines would not only perform the function of human thinking but become a substitute for that thinking as such.
Let me clarify the distinction I am making. It is one thing to have a machine that is capable of performing certain kinds of thought. It is another thing to have a machine begin to substitute for that kind of thought entirely, to perform it so effeciently that human thought ceases to perform this kind of thinking on its own because it allows itself to become substituted by the machine and dependent on the machine to perform these modes of thought in its place. For example, it is now the case, not only that calculators are capable of solving complex mathematical functions for me, but that they have replaced the necessity for me to do so, and have therefore replaced the necessity for me to learn to do so. The calculator has become, for most people, an absolute substitute for complex, or even not so complex, mathematical calculation.
This is not the only example of this phenomenon, of course. I might also talk about the grammar and spelling checkers that are now a part of virtually every textual interface that I use, and there are many other examples as well. The most interesting extension of this idea to me, however, is how the mode of thinking that has before now been called research has become increasingly performed by machines, almost by necessity, and has increasingly come to substitute for my own thinking in this mode. Google is the obvious example in this respect, but there are many others, one or more for almost every digital task that I perform, from searching the files on my own computer to searching routes on my GPS map system. The increase in data that digitization permits has necessitated machine search, and this in turn has forced me to abdicate my own thinking, at least in certain ways.
This is significant, I think, because there are many ways, many systems, many processes by which I might conduct my research, and these different approaches are completely capable of arriving at different results. So much is obvious. It should also be obvious that the range of research approaches available to me are limited by the tools that I use, along with other things, like disciplinary conventions and existing methods for organizing and cataloguing information. To some extent, therefore, the mode of thinking that is operative in research has always been limited, but it has not until now become substituted like it has become substituted by the search engine.
Virtually all research now begins through the search engine. It is the only practical place to begin in the face of the vast amounts of information now available to the researcher. Yet, search engines are coded to produce results according to certain criteria, most of which remain hidden from the researcher, and many of which are not helpful to serious research. To use Google again as the obvious example, the ranking of pages based upon the number of links to them and other such quantitative criteria produces a list of hits that is based more on popularity rather than on utility for any particular purpose. This kind of approach obviously privileges sites that are longstanding, easy to understand, visually appealing, or associated with powerful offline interests, rather than those that are most accurate or most thorough. While there are many exceptions to this privilege, and while there are ways to refine searches to make them more reflective of other criteria, it remains that the structure of the search engine itself has begun to perform the thinking of research as well as the function, has begun to substitute for this kind of thinking for most people.
Again, none of this is profound. Many have remarked at how scholars today have become more data miners and data analysts than researchers properly speaking, and this shift probably reflects a felt need resulting from the sheer amount of data that is being produced. What it means to me, however, is that we need to start thinking critically about the kinds of programs, scripts, and algorythyms that produce our data. To the degree that we allow machine thinking to substitute for human thinking, and this substitution is becoming increasingly unavoidable in many areas, it is imperative that we cede these modes of thought with caution and with a constant critical attention to the codified assumptions on which machine thinking is based. We need to maintain an awareness of the limits and the limitations of the thinking that machines are performing for us, even as we take advantage of the possibilities that this thinking enables. In some cases, we may also need to find alternative ways, either traditional or innovative, to counteract these limitations. To do otherwise is to risk limiting our thinking in ways that are intellectually dangerous.