Another persistent myth is that data mining for counter-terrorism requires additional data collection. As a recent report by the Markle Foundation notes, the real challenge is to make better use of the large amount of data that are already collected by government agencies or that are available publicly.

Some techniques developed by data mining researchers could actually help reduce data collection. First, some techniques for active learning have been developed that indicate which new data points would be most useful to collect. Second, a frequent finding of data mining analyses is that some sorts of data are not useful for prediction. For example, in my Alzheimer's diagnosis work, it was clear that some interview questions did not help distinguish between patients who were healthy and those suffering from Alzheimer's dementia. This sort of finding can reduce future data collection.