Inspecting for Weapons Production a Modest Computer Simulation

We have tried to present an experiment in the detection of illegal weapons production, using a rather modest and simplified computer simulation. Three inspection schemes are defined by their ability to detect evasion and to avoid false alarms, and their effects are simulated using Monte Carlo methods. Because the model is not a fully dynamic one, it employs delayed rather than more or less instantaneous feedback; the full intensity of the inspection-evasion interaction is moderated. Furthermore, the model makes a number of assumptions which are justified only in so preliminary an analysis. For simu lation techniques such as that presented here to play a really useful role in the design of inspection systems and exercise a practical effect on the negotiations for disarma ment, this pilot project must be considerably surpassed in complexity and sophistication. Proceeding on the basis of data gathered during the computer simulation, we out lined a sequence of procedures whereby the potential signatories to a weapons produc tion ban agreement might be able to sort out and compare the relative costs of alterna tive inspection systems outlined. We have quite consciously sought to combine such disparate material and strategic-political factors as installation and operating cost on the one hand and successful evasion or false detection cost on the other. Two caveats are thus in order. First, such 'combining of utilities' may be a perfectly reasonable procedure, but we cannot afford to forget that certain of our inputs are based on political judgements of a most crucial sort. Second, and equally important, the calcula tions suggested here incorporate only a small portion of all the variables that must be considered in selecting a man-machine inspection system upon which a nation's security may well depend.