3 ms·
> But you are going to get an error rate, and probably a high one in the noisy real world. ... The reasonable thing to do with that information, would be to sur
by pwg 11y ago
> But you are going to get an error rate, and probably a high one in the noisy real world. ... The reasonable thing to do with that information, would be to surveil them further
It is even worse than that. Due to the base rate fallacy (https://en.wikipedia.org/wiki/Base_rate_fallacy https://en.wikipedia.org/wiki/Base_rate_fallacy) the 'machine' is going to have a huge false positive rate, so without that additional surveillance, most will not be guilty of what they are accused. But that surveillance takes money and resources, just the thing the 'machine' system is designed to save...
- moyix 11y agoYes. I feel like anyone considering such a system should be forced to read "The Base-Rate Fallacy and its Implications for the Difficulty of Intrusion Detection": http://www.raid-symposium.org/raid99/PAPERS/Axelsson.pdf http://www.raid-symposium.org/raid99/PAPERS/Axelsson.pdf
- ehudla 11y agoPeople interested in these issues should read the position paper written by Jeff Jonas and Jim Harper in 2006, "Effective Counterterrorism and the Limited Role of Predictive Data Mining". http://www.cato.org/publications/policy-analysis/effective-counterterrorism-limited-role-predictive-data-mining http://www.cato.org/publications/policy-analysis/effective-c...