Decisions, decisions : uncertainty, variability, modelling and psychology

Abstract

There is strong evidence that many major oil and gas (O&G) projects are plagued by large cost and Schedule overruns, or significant underperformance in productivity. Similar evidence exists for Large mining projects. This paper claims that uncertainty is the underlying reason why projects fail to deliver the metrics used to justify their investment, or to deliver the economic value possible if alternative decisions had been made. Most technical and economic analyses are performed to reduce uncertainty to help make decisions. Decision-making is about ranking alternatives and choosing the best; rarely are precise predictions required, if this is even possible. Better uncertainty assessment is one of the keys to better decisions, thus better business outcomes. To get better decisions it is suggested that the focus or reward/penalty incentives be shifted away from the quality of outcomes of the quality of decisions. high-quality decisions are those that are generated from the skilled use of the tolls, methodologies and processes from the discipline of decision analysis. Uncertainty is quantified of probability, which is a degree of belief in the truth/falsity of some proposition about a quantity relevant to a decision, such as volume of in-place resource, future price or action of a competitor. The probability a person ascribes to a proposition depends upon their information set,; hence probability is subjective – it is not an inherent attribute of the quantity of interest or the physical system. The above simple observations have profound implications for modelling approaches, how they deal with uncertainty, the use of data to help assess uncertainty and the reliability of information Provided by subject-matter experts. The limited circumstances under which frequency data (from ppatial or temporal variable) can be used to assign probabilities is also reviewed. People, in general, are subject to a range of biases and cognitive errors. Research with O & G subject-matter experts shows that their expertise offers little or no protection from making errors in probability assessments and other judgements.