Much more thought needs to be put into how this will be done. The first line of defense against bugs will of course be the extraordinary care which all of us will take to insure that the code which we write performs as designed. When this fails dismally the next line of defense is the referee system. This will only work if the referee's take their job seriously and really work to find problems in the routines they are assigned. Given the SNO collaboration's normal enthusiasm for finding flaws in each other's work this should not be difficult to arrange. Since a man with two watches never knows what time it is, but at least he knows he doesn't know, it is hoped that Peter will continue to develop and use the Queen's code as an alternative to SNOMAN. The ultimate test will be how well the code agrees with the calibration data. A slightly different question is the variation of results between platforms. For this we should build on the earlier work of Bill and try to come up with a set of standard runs which will be executed on every platform during the 2 week bug-fixing trials of each new release.