The analysis described for version 2 of cut optimization should be considered a work in progress. Several new variables will be investigated in the near future including cascade and iterative reconstruction. However, even at this stage, a few observations are warranted. The optimization procedure relies on the assumption that the MC predicts the true response of the detector to signal events. Distributions of the background MC at various stages of the analysis were compared to the 1997 data. In general, the simulated distributions roughly agreed with the data for the variables used in the analysis. However, the inability of the MC to predict the absolute values of the distributions may create a problem in evaluating the absolute efficiency for signal detection. The danger is that the selection criteria may remove signal if the absolute values of the distributions are incorrectly modeled. To investigate this potential problem, the final values of the cut variables were compared to the predicted distributions. The signal efficiency for most of the parameters is not sensitive to the precise value of the parameter. However, this may be an issue with the ldirc(2) cut as seen from figure 3. In the next version of cut optimization, further care will be taken to avoid a sensitive dependence on the precision of MC for signal. The discrimination power in the variable jkrchi(2) appears to decrease as the selection criteria become more refined. In the future, this cut will be applied after all other variables, but only if it improves the signal to noise. The zenith(2)-zenith(1) cut preferentially improves the signal to noise in the vertical direction. Care is required to insure that acceptance at large zenith angles is retained.