Sunday, October 12, 2008

More evidence against data-mining as an anti-terrorist tool

There have been a number of studies saying the same thing (search Bruce Schneier's blog for plenty of other examples)

It's kind of like trying to determine if someone is going to rob a bank by looking at their phone calls, etc.  How do you know _beforehand_ what kinds of "patterns" and "data" are indicative of a bank robbery?  And even if you see one bank robbery, not all are created equally, so why would you be able to infer that you could predict the next one based on previous data?

Scientists question terrorist-hunting techniques - CNN.com
The 352-page report by the National Research Council, an arm of the National Academy of Sciences, does not evaluate the TSA or any other specific government program. Instead, it explores issues related to data mining and behavior detection techniques and attempts to advise lawmakers how to appropriately balance security with privacy.

The report recommends the government be required to systematically evaluate the effectiveness and lawfulness of data-mining and behavior-detection programs before implementing them, and at regular intervals thereafter.

The programs also should be subjected to robust, independent oversight, the group recommends.


No comments:

Post a Comment