THE HUMAN BIAS IN EARTH SURFACE PROCESS MODELING: LESSONS FROM THE SEARCH FOR WMD

Important failures of applied quantitative modeling of earth surface processes have occurred recently in fields as diverse as marine fisheries (Grand Banks cod), ground water flow (Yucca Mountain), and nearshore processes (nourished beach lifespans). The recent highly visible and much analyzed predictive exercise of the search for weapons of mass destruction (WMD) was apparently far ahead of the actual field evidence. Very similar human biases, presumed by analysts to be partly responsible for the WMD failure, are factors in the many predictive failures of quantitative mathematical models of earth surface processes. These problems in common (beyond the political factor) include: l. Treating the worst/best case scenario as fact. A modeling analogue is the choice of a very low ground water percolation flux at the Yucca Mountain repository which prevailed until actual field work indicated it was wrong.2. Glossing over ambiguities. Model predictions of pit lake composition after abandonment of an open pit mine (For EIS purposes) have traditionally been highly unrealistic through failure to question obvious contradictions and inconsistencies of modeled sulfide geochemistry.3. Fudging mistakes. When nourished beaches disappear faster than predicted by models (the usual case), the problem is attributed to unexpected storms rather than to model failure. Those who apply models or are affected by model based decisions should be aware of these and other human biases and should judge model results in that context.