Adopted from microeconomics, linear programming became popular beginning in the 1970’s in response to the critique that Optimal Foraging models which only use energy as the currency are too limited. This critique charges that other nutrients (e.g. vitamins, protein, fats, etc.) and/or the nonfood benefits of a particular resource (e.g. as a source of raw materials) may outweigh the importance of calories. In linear programming, each currency variable is assigned its own linear equation based on the set of constraints, and the point at the equations intercept is the most optimal solution for satisfying all dietary and non-dietary requirements. (For clarification, including a hypothetical example of linear programming, see Kelly 1995:73-78.)
Unfortunately, linear programming is not without its shortfalls. Primarily, models require accurate information regarding resource attributes (including the length of time necessary for manufacturing tools, processing, etc.) and human nutrition (including daily requirements, digestion rate, etc.). Even slight shifts in a model’s parameters can considerably alter its predictions (Belovsky 1987).
Thus, linear programming has limited application for archaeology. Although information regarding such parameters can be collected ethnographically, it is nearly impossible to make accurate assessments for prehistoric foragers based on the archaeological record. One cannot assume that environmental conditions or speed in processing food resources or in tool manufacture have remained equal through time. Although, Sheehan (2004) is optimistic in regards to future applications of linear programming. Paleoethnobotanical studies have the potential to reveal prehistoric environmental and climatic conditions. As such studies continue to become more common and refined, the potential to reconstruct prehistoric landscapes and more accurately apply linear programming models will expand as well.