Training and experience... Sounds like the police are misusing bayesian techniques, and the judge called them on it. In computer science terms: When training a bayesian filter for spam identification a sufficient number of items identified as 'not spam' (ham) needs to be included, along with items accurately identified as spam, or results will not be valid. If it isn't, all tested input will be identified as spam, whether it is or isn't.
Techdirt has not posted any stories submitted by Marksbcss.
Could be worse...
Years ago, someone alleged:
Reproduction without the express permission of MS Magazine is strictly prohibited.
Re:Judge Tears Apart Law Enforcement's Ridiculous Assertions About 'Suspicious' Behavior
Training and experience... Sounds like the police are misusing bayesian techniques, and the judge called them on it.
In computer science terms:
When training a bayesian filter for spam identification a sufficient number of items identified as 'not spam' (ham) needs to be included, along with items accurately identified as spam, or results will not be valid. If it isn't, all tested input will be identified as spam, whether it is or isn't.