This paper argues that the trend toward geographic information
systems that are more powerful, less expensive, and easier to use
seems to be unequivocally a democratizing, counter-technocracy
trend. In truth, however, the decentralization of geographic
information systems hardware and software goes hand-in-hand with
a centralization of the geographic modelling programming that
underpins the GIS. These paradoxical trends give the illusion of
growing democratization of GIS technology, while in reality,
there is a growing danger of a hidden GIS technocracy, owing
primarily to the lack of recognition that such a technocracy
exists. While the decentralization of the technology is likely to
prevent any major wide-spread cataclysms, individual
organizations implementing a GIS may experience problems caused
by the implementation of geographic modelling programs that may
be inappropriate to their needs. The remedy for such problems is
awareness of the hidden technocracy as well as of the potential
for trouble. One way to increase this awareness is to promote the
use of lineage information for the geographical models embedded
in geographic information systems in the same way that the GIS
community has pressed for lineage information on databases.
The development of geographic information systems has progressed dramatically in recent years. Systems have become more powerful, less expensive, and easier to use. As a result, GISs are proliferating at an unprecedented rate, and concern about the societal impacts of their implementation is growing. One area of concern has been the potential for control that lies in all information systems, including GIS. The critical component that underlies this potential is the existence of a technocracy.
Webster's New World Dictionary defines a technocracy as, "government by technicians; specifically, the theory or doctrine of a proposed system of government in which all economic resources, and hence the entire social system, would be controlled by scientists and engineers." Such a picture is unsettling indeed. However, one need only look around at the progress in GIS to see that this is not happening. On the contrary. GISs are becoming less expensive and easier to use. This trend suggests that concern about a GIS technocracy is unwarranted. Or is it?
This paper argues that the trend toward geographic information systems that are more powerful, less expensive and easier to use appears at first glance to be unequivocally a democratizing, counter-technocracy trend. In truth, however, the picture is mixed. While one can reasonably argue that more power and lower system costs are true democratizing trends, the greater ease in use of GISs (and, by extension, other information systems -- including expert systems) comes at a cost of the evolution of a hidden technocracy. This technocracy is hidden in the offices of the vendors that develop the hardware and software that make the technology more generally accessible. I do not mean to suggest that there is a vendor conspiracy to take over the world through spatial analysis and geographic modelling. However, it is important to recognize that what organizations gain in easier-to-use general purpose geographic information systems, they may give up in specifically tailored models that meet their own unique needs. This loss could create serious problems on the local scale.
Nor do I mean to suggest that it is desirable to eliminate experts. This would be both unwise as well as infeasible. However, organizations should be aware that the seeming simplicity in the use of modern, so-called transparent GISs belies the expertise that created them. This expertise -- and the technocrats who possess it -- lie in the areas of both computer programming languages and geographic modeling.
This paper begins with a discussion of technocracies, then
provides evidence of the trend toward greater general access to
geographic information systems. The following section describes
potential difficulties with the hidden GIS technocracy. The
conclusion suggests that the GIS community promote the use of
lineage information in geographic models embedded in GISs as a
primary way to protect users from such a technocracy.
Technocracy is not at all a new system and is based on the idea of a meritocracy. "For twelve centuries," notes Weber (1946:416), "social rank in China has been determined more by qualification for office than by wealth. This qualification, in turn, has been determined by education, and especially by examinations." The exalted individuals in the Chinese system were known not as technocrats, but rather as mandarins or the literati (Weber 1946:416). While it was theoretically possible for a plebeian to become a member of this group and thus to share in the prestige accorded to scholars, it was a relatively rare occurrence. The difficulty of learning the Chinese system of writing mitigated against this happening with any regularity (Weber 1946:417). Nonetheless, once an individual achieved the education of the mandarin class, he (they were male) was accorded great prestige and high status by society at large.
Technocracy continues to exist today, and there is evidence that it has been on the rise throughout the twentieth century. Weber (1946), for example, emphasizes the importance of expertise in the development and evolution of bureaucracies in the twentieth century. Weber's bureaucrat is frequently trained or educated outside of the institution where he or she is employed, often at a college or university. This specialized training is often coupled with some sort of certification, such as the earning of academic diplomas (the Ph.D., for example) or the successful passing of professional examinations (the bar exam, for example) and sometimes both. This coupling provides objective evidence that the individual has mastered specific professional skills.
Writing at about the same time as Weber, Frederick Taylor (1911) argued that "...the best management is a true science, resting upon clearly defined laws, rules, and principles, as a foundation" (1911, 1947:7). Furthermore, good management lies not with particular individuals, but instead resides in basic knowledge and skills that can be learned. Taylor coined the phrase "scientific management" to describe his theory of management. The principle of scientific management calls for the development of a science based on expertise to replace the "rule of thumb" knowledge of workers.
The implementation of scientific management has not been without consequences. Some scholars suggest that an important consequence has been the growth in the power of management even as there has been a diminution of the power of labor at the lowest levels of the hierarchy (Lipietz 1986; Sayer 1986; Urry 1986). At the same time -- and in contradiction -- Urry suggests that there has been an increase "in the numbers and influence of industrial engineers..." and that their relationship with management has become "increasingly symbiotic" (Urry 1986:46). Prophetically, Veblen, writing in 1921, suggested that technical experts were the group most deserving of the right to control the means of capitalist production (Burris 1993:26-7). The presumption that education is available to everyone (a debatable premise, even today) underpins the implication that such an arrangement is democratic.
In fact, Veblen’s wish has become something of a reality. For example, Giddens (1982:36) suggests that the increasing importance of expertise in the workplace has facilitated the rise of technical experts, particularly in centralized situations. He contends that, "strategically placed categories of workers are able, through the threat of withdrawal of their labour, to increase their power. What applies within the economy, or the state as a whole, applies in more specific situations also... For a highly centralised production process tends to be much more vulnerable to disruption by small groups of workers than one in which labour tasks are less interdependent"(1982:36).
The emphasis on education and technical expertise as the key means by which one qualifies for employment has become well-embedded in our social and economic structures today. This emphasis comes at the expense of previous models that held family ties and political affiliations as the key factors in hiring decisions (and by extension, respect accorded to individuals based on their employment). For example, the civil service system of the U.S. federal government requires that applicants for some positions (e.g., postal employees) take an examination to determine their fitness for government jobs. In other cases, one’s achievement in higher education and experience in other positions helps to determine the fitness for federal employment. As in the Chinese system, it is theoretically possible for anyone to achieve a position of high social status by attaining an education and landing a job in a respected profession as a result of the civil service system and other similar systems in the private sector. In reality, the huge variation in the quality of education at the primary, secondary and post-secondary levels across the U.S., combined with the limitations on access to education imposed by socio-economic status mean that the system is not as open as it may at first appear.
While we generally accept the value of education, and
implicitly of expertise, specialized education that is linked
with specific professions can also have its drawbacks. Cyert and
March (1963:105) emphasize the role of professional training:
Consequently, the kind of training a member of the
organization brings to his or her job is important (Merton 1952;
Simon 1952). This specialized training imparts essential skills
and knowledge to the individual bureaucrat, but also helps to
shape his or her values and choices.
Anyone who can remember the days of computer punch cards knows from personal experience alone that computer technology generally (and, by extension, GIS) has become more democratic over the years. Even in the mid-to- late 1970s, the experience was intimidating for the average user. Not the least of this intimidation were the inevitable meetings with "the consultant."
In those days, performing a computerized analysis required using a keypunch machine, that is a machine equipped with a typewriter-style keyboard that punched holes in heavy-stock paper cards. One then handed the cards to a person whose job it was to run them through a special machine that could “read” the cards because of the location of the holes. This machine then sent impulses to a computer that performed the specified analysis (or not, if one was unlucky). The computer, in turn, sent signals to a printer, which provided a hard-copy output of the results of the analysis. All this time, the user waited. If he or she was lucky, the hard-copy report met with expectations. Many times, however (at least in my experience), the report was a thin one, which, like thin responses to job applications, spelled trouble. The next stop was the consultant.
The consultant reigned as a high priest or priestess of the computer. He or she was all-knowing, all seeing. The consultant reviewed the thin printout, then made a pronouncement regarding the source of the error. The error might be something as simple as omitting a space where one was needed on an important card, or it might involve a more substantive mistake. In any case, the average user had no choice but to rely on the consultant: such was the complexity of computer analysis in those days. This was the computer technocracy.
Visits with the consultant could be intimidating if for no other reason than that the very need for such a visit alone was proof positive that the user was incapable of performing even the simplest of computer analyses. (Let us not forget that all this activity took place in public settings: one was always certain that everyone in the facility else performed their work without a hitch, all the while observing your difficulty and laughing secretly.) Moreover, it was not uncommon to encounter consultants who treated their visitors with disdain because of their inability to do their work without help. Of course, if the average user had been capable of performing his or her tasks with no assistance, there would have been no need for consultants -- which was an insight that the computer industry observed and to which it responded with the development of the personal computer and easier-to-use software.
The development of the personal computer and its software changed the face of computing, simplifying and, thus, democratizing it. The simplification of computing -- an important part of which was on-line, interactive systems that eliminated punch cards forever -- eliminated hordes of on-site consultants. The new systems were much easier to use and the fact that they were more affordable made it possible for average users to learn the systems in the privacy of their own homes or offices. Sheepish visits to the consultant were replaced by a telephone call to (often) friendly voices eager to help. At the very least, interactions that had previously been about as comfortable as visits to the principal became phone conversations with impersonal voices belonging to people whom you would never meet (or if you did, they wouldn’t recognize you anyway).
Still, it is crucial to recognize that the early computer technocracy existed to help average users overcome hurdles that early mainframe computers imposed. Typically, the computer user in this context had -- or was receiving -- education or training in the substantive application area (often statistics or basic writing). In the case of the geographic information system, the substantive application area -- geographical modeling -- is an area in which relatively few people have expertise. Organizations that adopt GISs must often rely on experts to guide them through both the computer programming as well as the geographical modeling components of the task.
The capabilities of personal computers have broadened greatly over the years; all the while costs have declined and user-friendliness has increased. Programs to accomplish routine tasks such as word processing, spreadsheet analysis, accounting, and tax preparation (to name a few) abound. One can even find inexpensive ($5) programs to balance a check-book on sale at the grocery store! Moreover, the range of functions has also increased dramatically. For example, The Wall Street Journal reported recently that a growing number of small-town newspapers are using a computer program that writes articles that report the results of high school football and basketball games -- just add data (Bulkeley 1994). There is, however, an important difference between most of these programs and GIS programs: presumably most people who purchase and use the programs to perform such tasks as word processing, balancing a checkbook, preparing tax forms and so on, are at least minimally capable or have a minimal understanding of how to perform these tasks without the use of a computer. In contrast, geographical modeling may be entirely outside the experience of GIS users.
The trend toward expanding capabilities is prevalent in geographic information systems as well. Turn-key set-ups using readily available data sets and canned programs have replaced the early fully customized systems. Functionality has improved dramatically. Increasingly, average users, many of whom have neither specialized education either in geographical modeling or in the programming or use of geographic information systems, can perform relatively sophisticated spatial analyses. For example, the Chicago Tribune reported in early 1994 that a Minnesota state legislator used a desktop demographics program, Census Bureau Population and state budget data to create his own maps showing that a small group of wealthy suburbs had only 25% of the area’s population but received 70% of new jobs, 83% of highway funds, and 90% of new sewer funding. This information, "set his constituents rocking" (McNulty, 1994, Section 5, p. 1).
GISs are becoming, in current parlance,
"transparent." That is, the user can focus less on the
mechanics of using the technology, and focus instead on
performing a specific task. By way of analogy, transparent
technology also enables us to pick up the telephone and make a
telephone call (be it local or international) without thinking
about what makes the telephone work, nor having an operator to
intercede for us. But this transparency works for the telephone
because users presumably know enough about the rules of grammar
and pronunciation to communicate. It is not at all clear that GIS
users always know enough about the rules of geographic theory to
understand the models that underlie geographic information
systems. The trend toward transparency in GIS, particularly as
exemplified by the Minnesota legislator’s analysis,
immediately suggests that computer technology generally (and with
it, geographic information systems) is becoming more
democratized. As GIS becomes more transparent, the technocracy
seems to be fading away as it apparently has with the telephone.
But is it really gone? And is its apparent elimination the
advantage that it seems.
The tendency to assume that there is no GIS technocracy is naive. In fact, the technocracy still exists, although it has become hidden, and is actually more centralized than it was in the days of on-site GIS consultants. I suggest that one of the reasons why we fail to see the continuing existence of technocracy is because we look in the wrong places. Orwellian predictions of a "Big Brother" central government appear throughout our discussions as a society about the outcomes of computer technology implementation. We are used to thinking about a government that sees all, knows all, and controls all via its technocracy. We are much less accustomed to paying much attention to the activities of the private sector in this sphere.
And yet, it is the private sector that brought us the personal computer. It is the private sector that brought geographic information systems into cities and towns, universities, businesses and homes around the world. It is private proprietors who develop the programs that perform the geographical analyses embedded in geographic information systems. These same proprietors also design the courses, and in many cases teach the people who use the geographic information systems once they are in place. Private proprietors play a central role in the implementation of geographic information systems. It is here that we find the hidden technocracy.
So what's the big deal? As Robert McMaster observed at the 1994 Annual Meeting of the Association of American Geographers (AAG), there have not yet been any major problems associated with bungled GIS implementations. And I would add that the trend toward decentralization of computer technology generally and GIS specifically decreases the likelihood of widespread cataclysms. However, the decentralization of GIS hardware masks the centralization of geographical modeling programming and the potential problems that may arise. While making available geographical analysis capabilities that would not otherwise be available, GIS technology is likely to be designed with a general audience in mind, and therefore may be less appropriate for exceptional cases. Further compounding this situation, people implementing GIS technology in an exceptional case may be unaware that a general set-up is not appropriate for their specific needs.
This apparent paradox (centralization of programming, decentralization of hardware; or technocracy in the midst of democratization) means that organizations implementing GISs may readily get the impression that they have greater control over their work than they actually have. While the organization implementing the GIS typically owns the hardware, and inputs its own data, the programs (that is, the geographic models that actually specify the assumptions and methods for performing the geographic analysis) are generally embedded in the software the organization purchases by the GIS vendor. The vendor, not the organization implementing the GIS, defines the problems and develops approaches to solutions. Unlike early systems, that often were built from scratch and specially tailored to meet the specific needs of a unique organization, modern GISs assume a greater level of generalizability. This has the potential to produce problems related to the application of inappropriate models in particular situations.
The potential risks of the hidden technocracy become clearer in light of Tobler's First Law of Geography: everything is related to everything else, but nearer things are more closely related than distant things (Tobler 19__). The closer (in time and place) an organization's situation is to the assumptions and models that underpin proprietary GISs, the more appropriate will be the implementation.
Compounding this potential problem is the likelihood that the organization implementing a GIS may not even think to ask the vendor questions about the geographic models embedded in the system, owing to the relatively low level of geographic knowledge among the general public (particularly in the geographically illiterate United States). Even if the organization does raise the right questions about the models, it is conceivable that they will not have the expertise to evaluate the appropriateness of the vendor's models for their needs. In a recent example related to me by a colleague who is a hydrologist, a student excitedly approached him with news that the program his course in environmental GIS applications used included a hydrological model. When the hydrologist examined the model, he pointed out to the student that most hydrologists no longer considered the model to be valid. Similarly, urban GISs that are based on midwestern U.S. cities contain assumptions that may be irrelevant or inappropriate for the urban contexts of either longer-developed European and Asian cities or newly developing cities in regions throughout the world. Again, as Tobler suggests, nearer things are more closely related than more distant things. How many other models embedded in proprietary GIS analytical models exhibit similar problems?
As noted previously, it is unlikely that we will see any major
disasters caused by the hidden GIS technocracy. There are two
reasons for this. First, with the growing decentralization of
GISs, problems that do arise will tend to cover relatively small
areas: a town here, a county there. While these problems may be
matters of grave concern for the places that they affect, the
tragedies will be small, in the big scheme of things and for GIS
as a whole. At this scale, it is relatively easy to view the
problem as an isolated problem, even if it is actually a symptom
of a more systematic problem. If the trend toward
decentralization of GIS hardware and software reverses itself, we
may have greater cause for general concern. A second reason why
we are unlikely to see any major GIS disasters caused by the
hidden technocracy is that most organization use their GISs
primarily for data storage (Huxhold 1993; Goodchild and Getis
1991). This use is much more straightforward, and much less
subject to debate over appropriate format. One size may fit all
(or at least most). However, if organizations begin to use their
systems for more substantive spatial analysis, more problems are
likely, since the risk of using inappropriate canned programs
rises.
In the short run, serious problems from the hidden technocracy are few, for the reasons described above. Two trends, however, could change this prediction. First, if the trend toward decentralization reverses itself, wider- ranging problems could arise. Furthermore, if organizations begin to tap into the geographic modelling capabilities of GIS more aggressively, they will increase the likelihood of experiencing problems caused by a mismatch between their needs and the general-purpose models embedded in their GISs. While the former eventuality is less likely to occur, the latter is more likely.
As always, awareness of these apparently paradoxical trends is the first step toward avoiding problems. GIS vendors must enlighten organizations that implement their systems of the limitations of the programs and work to maintain a close ongoing relationship with their users. They must also help the implementors learn how to ask the right questions. GIS users themselves must become more sophisticated, not just about the technology, but about the geographic models that underpin them.
Just as the GIS community has called for the formal provision
of a lineage for each data set used in a geographic information
system, this same community should be just as vocal in calling
for a lineage in the geographical models embedded in GISs. This
lineage should include at a minimum (1) the basic assumptions of
the models, (2) a description of the geographic theories and
models used, and (3) the year in which the model is developed.
Such lineage information will make it possible for GIS users to
have basic information that may not otherwise be readily
available to them as GIS becomes more transparent and the GIS
technocracy becomes more hidden. In this way, the hidden
technocracy will be less likely to create unmanageable problems.
Bulkeley, W. M. 1994. Semi-prose, perhaps, but sportswriting by software is a hit, The Wall Street Journal, v. 75, n. 116, Tuesday, March 29, pp. A1, A5.
Cyert, R. M. and March, J. G. 1963. A Behavioral Theory of the Firm (Englewood Cliffs, NJ: Prentice-Hall).
Goodchild, M. F. and Getis, A. 1991. Introduction to Spatial Analysis . Notes accompanying a workshop of the same name held in conjunction with the annual GIS/LIS Conference, Atlanta, GA.
Huxhold, W. E. 1993. Book Review: "Spatial Analysis and Spatial Policy Using Geographic Information Systems," Journal of the Urban and Regional Information Systems Association, v. 5, n. 1.
Merton, R. K. 1952. Bureaucratic structure and personality. In R. K. Merton, A. P. Gray,, B. Hockey, and H. C. Selvin (Eds.), Reader in Bureaucracy (New York: Free Press), 361-372.
McNulty, K. 1994, January 9, "The Power of Maps," The Chicago Tribune, Section 5, p. 1.
Simon, H. A. 1952 Decision-making and administrative organization. In R. K. Merton, A. P. Gray,, B. Hockey, and H. C. Selvin (Eds.),Reader in Bureaucracy (New York: Free Press), 185-193.
Weber, M. 1946. The Chinese literati (chapter 27). In H. H.
Gerth and C. W. Mills, (Eds. and Trans.), From Max Weber:
Essays in Sociology (New York: Oxford University Press)
416-444.