We have a long history in the US of well-intentioned bots which end up causing problems[1]. My question is "Have you run your changes across the data to see what it actually does and then reviewed the results?" - Serge [1] Including my own.