Why are data-testers afraid of adjustment?

The goal of cross-section evaluators is to produce a complete cross-section library the use of which should result in reliable calculations of any neutronic system. However, a major obstacle on the way to realizing this goal is the limitations of current cross-section measurement techniques: their inadequate accuracies, and their even more serious `systematic' errors brought to light by the not uncommon inconsistency of different measurements of the same cross sections. In any case, even the newest cross-section evaluations are just not good enough. The calculated nominal values of reactor (or other device) performance parameters are inconsistent with the corresponding measured values; and the calculated uncertainties (the propagated errors) are much too large. Reactor designers, of course, are only too aware of this predicament and invest great efforts in the construction of elaborate mock-ups in order to check on their calculations. Quite naturally, evaluators are interested to test the libraries they produce. To this end they make use of available benchmark integral experiments to compare calculations based on their libraries with measured values of diverse integral parameters. On the other hand, those involved in integral experiments analyze their measurements and are just as interested to demonstrate the agreement of their own results with `objective' calculations. Stretching the point a bit, the authors allege that while the cross-section evaluators refer to integral measurements as their absolute reference points, the producers of these very measurements consider the results of calculations as ideal values which they would like their own results to approach.