• Choose your language

    English French German Italian Portuguese Russian Spanish

Useless Arithmetic: Why Environmental Scientists Can't Predict the Future (Pilkey and Pilkey-Jarvis)

Useless Arithmetic: Why Environmental Scientists Can't Predict the Future. 2007. O.H. Pilkey and L. Pilkey-Jarvis. Published by Columbia University Press, New York, New York, 230 pp. ISBN 0-231-13212-3 (Cloth). US$29.50.

Click to buy from Amazon

It is difficult to provide a dispassionate review of this iconoclastic book by Orrin Pilkey and his daughter, Linda Pilkey-Jarvis. In only a couple hundred pages, the Pilkey's have probably managed to offend many readers, not to mention developers and users of mathematical models. I must admit that I approached this small book with a certain amount of trepidation, having heard so many disparaging remarks prior to having the book in hand. However, when I started reading the book, I was pleasantly surprised by the candor and insight that these authors bring to a subject that deserves comment. It would probably be helpful if readers could approach the book as dispassionately as possible, simply because they will get more out of the book if they can suppress angry flare-ups. That is what I did, and when I became annoyed, I sat back and cogitated over the remarks that annoyed me. After some careful introspection, I came to the conclusion that there were kernels of truth coming through what I perceived as Orrin's usual huffing and puffing overstatement to get the point across. Daughter Linda has no doubt picked up the trait, and together they build quite a good case of why environmental scientists can't predict the future. Agreeing with them, one might wonder how much of the future really needs to be predicted, especially if adaptive management techniques are employed, as suggested by the authors. Humans always want to know the future, but uncertainty about future events seems to be the rule, not the exception.

The book is divided into the following chapters: Chapter 1: Mathematical Fishing, Chapter 2: Mathematical Models: Escaping from Reality, Chapter 3: Yucca Mountain: A Million Years of Uncertainty, Chapter 4: How Fast the Rising Sea?, Chapter 5: Following a Wayward Rule, Chapter 6: Beaches in an Expected Universe, Chapter 7: Giant Cups of Poison, Chapter 8: Invasive Plants: An Environmental Apocalypse, and Chapter 9: A Promise Unfulfilled. Among them, there are only three chapters that deal with coastal issues per se. Although they are only a part of the Pilkey attack on numerical modeling procedures and practices, the authors manage to zero in on subjects that lie close to the hearts of many coastal researchers. Make no mistake about this book. It is an attack on many of the sacred cows and holy grails of coastal management. Although the other chapters are of interest, and I read them all, my comments focus on topics that are germane to our subject area. It is also perhaps worth noting that one aspect of coastal management, beach restoration (replenishment, renourishment), is held up as one of the worst examples of quantitative modeling efforts in science and engineering specialties. Though couched within the purview of life spans (durability) of renourished beaches, this is nonetheless an attack on the very heart of beach nourishment, its principles and practices. Having recognized the attack, it is difficult to argue with the following passage on page 187.

…The modeling of beaches for coastal engineering takes the cake for the worst of the bunch. There is not the slightest public recognition of problems with the models, and the models are so tightly bound up with politics that the answers are essentially useless, except to those whose purposes are served by inaccurate answers. Fudge factors are routinely used, looking back to learn from the past is simply not done, and most practitioners remain blissfully unaware of, or at least uncaring about, model weaknesses. The problem is amplified because engineers who use the models are rarely specialists in sedimentary geology, and they fail to appreciate the model's detachment from reality…

Clearly inflated and overstated to make the point, a thread of truth runs through the paragraph regarding political overtones, calibration, verification, validation, sensitivity, and caveats about model weaknesses. The perceived gap between engineers and sedimentary processes in some cases may be real, but in other situations, the contrary holds sway where some engineers make very good sedimentologists. There are always examples of inappropriate model application, running of models by those who may not be fully qualified in aspects of the discipline that they are trying to quantitatively approximate, and use of models by users who are not fully aware of assumptions and limitations of the model. This will most likely always be the case, and the situation essentially boils down to one of degree of excess, an aspect of the relative pervasiveness of the problem.

Close reading of the offending pages indicates a degree of unfamiliarity with some of the more recent numerical modeling packages, but these quibbles are minor. Contrary to the kind of thing that the authors suggest is the rule, the recent process-based morphodynamic area models are exemplary because they actually take into account changing morphologies (including sediment transport and biogeomorphological interactions) with time (time-integrated effects), and subsequent iterations reset the morphological stage prior to each sequential analysis. These kinds of morphodynamic models thus give more realistic computed two- and three-dimensional development when compared to measured development of coastal features, both natural (e.g., eddies, deltas) and human-induced (e.g., dredge pits, trenches). For the most part, however, father and daughter correctly home in on weaknesses inherent to numerical models. Although all numerical models retain weaknesses in the sense that they are approximations (simplifications) of nature, it is not expected that models should be perfect representations. This is an impossibility, and part of the idea is to develop a tool that can be used to acquire a reasonable or usable answer to a complex problem. Approaches to solutions of the problems in coastal science that are attacked in this book (e.g., predictions of longevity of renourished beaches, impacts of sea-level rise, application and veracity of the Bruun Rule) are indeed complex, and so are the models. The essential point that I read into the Pilkey complaints about model use is that some (most, many) users often neglect to present or discuss the limitations of the models they use and most often do not explain the simplifying assumptions upon which the models are based. Entrained in this oversight is also the tendency to project the notion that model results are "the answer" and that everything else (including common sense) is to be ignored at one's peril. The problematic notion here is that model results can be unquestionably relied upon, which explicitly suggests that the results of mentation by other means are unacceptable in the decision-making process. The realization that the thrust of the Pilkey intrusion into hallowed realms was altruistic and not vindictive decreased my propensity toward dualistic approaches to problem solving and increased my appreciation of nonlinear thinking as it freed me from the confines of the Aristotelian paradigm.

Having arrived at the foot of the Pilkeyian platform, I was able to better appreciate the thought process that begins to unravel the conundrum within which many of us get enmeshed. As a case in point, I recall attending a meeting in New Orleans a couple of years ago when a scientist-engineer type got up to explain the results of some coastal modeling effort. There were a couple of innocent questions that clearly niggled the presenter to which he replied, "Well, I could explain this procedure to you but I am sure you would not understand it." Sadly, he was speaking to colleagues and guests at a professional meeting, the atmosphere of which rang with hubris, to the chagrin of those in the audience. Arrogance on the part of modelers does not help their case nor inculcate a respect for what models can do. Humility goes a long way toward acceptance of less than perfect results.

To me, the Pilkey gripe about modelers' arrogance lies on solid ground, as does their concern for users not detailing model assumptions, limitations, and simplifications when presenting results to the public. Let's face it: numerical models are not useless arithmetic because they can serve useful purposes (e.g., hurricane storm-surge models, storm tracks). The problem is that results of numerical models can be made useless by injudicious application and presentation. In many coastal engineering projects, for example, scientists and engineers regard model results as guidance, a window into changing conditions, not the final answer to a problem. Depending on the nature and quality of input data, model results might be spot on or they may widely miss the mark, a situation that is often intuitively grasped. For many coastal experts, model results can be appreciated in terms of perceived correctness, and it is somehow known that the model must be run again in an attempt to achieve better results. Determination of parameter sensitivity, acquisition of real-time data (rather than use of hindcasted wave data, for example), calibration of the model, utilization of more iterations under different environmental conditions, and so on, represent attempts to get results that are more realistic and closer to what an experienced researcher knows to be a likely scenario. On the other hand, models under the control of inexperienced users can yield results that are not realistic, which can be an exercise in useless arithmetic, as the Pilkey team refers to it. Other situations, still more serious, occur when inexperienced users have no idea whether model results are realistic. These latter situations are what the book warns about—times when models are inappropriately applied by people who may have no idea when the results should be accepted or rejected. One would like to think that these cases hardly ever occur, but apparently this is wishful thinking.

So, what do we get out of this book? To me, it is a good book that should be read by coastal researchers and managers who have occasion to use numerical models or interpret model results from a planning or managerial point of view. The most valuable part of the book is the discussion of model limitations and the need for everyone to be aware of the simplifying assumptions that go into making the models work. Initial assumptions are not to be ignored, and they are often glibly passed over as being irrelevant, nonessential things one is not required to know. Such is far from the truth. It is also important to maintain a system of checks and balances, which is so crucial to science and engineering. We sometimes get so wrapped up in our work that we neglect warning signs or red flags that alert us to potential problems that we would much rather dismiss than resolve, mainly because it is a momentary perception of the path of least resistance.

The Pilkey effort is to be lauded because it forces us to see what we want to ignore, our own limitations and shortsight-edness. We are forced to consider why we run models. For many, they run models because everyone else is running models, and they can't be left behind. There is the notion that those who run models make more money and are somehow part of the intelligencia, which makes them better than those who do not run models. In many cases, the opposite is true. The human brain is still one of the best computers around for solving everyday problems. Intuition serves us well, even in the most complex situations, if we listen from within. It is when we become arrogant, anxious to feed our egos, gnawing to be superior, that we fall from our lofty peaks to the low valleys of reality. The Pilkey duo helps us to see our folly and encourages us to sit still, listen, think, and evaluate what we do. If we can do that, as we continue to develop better numerical models, they will eventually become useful arithmetic that serves us well. In the interim, we need to be as vigilant as the Pilkey team encourages us to be.

Buy the book, read it, embrace it, enjoy it. In my opinion, it would be a mistake to eschew this book. Rather, it should be embraced to advantage by considering pitfalls that are inherent to the modeling process. To a large extent, the recently developed process-based morphodynamic models go a long way to meet the challenges of predicting morphological change. Instead of regarding this book as an annoyance, it should be viewed as an aid in which researchers play devil's advocate. The watch dog's bark is a warning to take note, be cognizant, present a more integrated front to the public (by admitting limitations, describing assumptions, providing probabilities of accuracy), and adjust predictions of numerical models to the level of guidance, another line of converging evidence, and informative possibilities. Last but not least, it should be appreciated that conceptual models continue to serve the research community and that numerical models are part of the overall process where human minds attempt to appreciate nature in a comprehendible manner.

Dr. Charles W. Finkl
West Palm Beach, Florida

Author Instructions

downloadDownload and consult the JCR Instructions For Authors before submission. Contributions must follow these specifications or will be returned for correction.

Submit Your Article

uploadElectronic submission of all contributions is mandatory.  Please use this online link to submit your article for peer review and possible publication in the JCR.

Manage Membership

go-rightClick here to join or renew your CERF-JCR membership.