In particular, it would seem appropriate for the Bureau of Engraving and Printing (BEP) to engage in a series of large-scale field-like experiments in order to understand the “detectability” of various existing and proposed features.
The committee also observes that the model makes it obvious that machine authentication affects counterfeit flows only via culling in the banking system (given by the variable β). While counterfeit notes may pass through (and be rejected by) machine readers a number of times during circulation, these are almost always returned to the holder and remain in circulation until removed by a human or culled in the banking system. Therefore, machine readers only weakly affect r or h.
Finally, it should be clear that the hypothetical example presented in this appendix should not be interpreted to mean that the committee is suggesting that a single, dominant feature by itself would deter counterfeiting. The whole purpose of presenting a modeling approach is to provide a framework within which the effectiveness of combinations of features can be assessed in a consistent manner.