Slashdot Mirror
Police Agencies Using Software To Generate "Threat Scores" of Suspects (washingtonpost.com)
Koreantoast writes: It's no secret that governments across the globe have been taking advantage of new technologies to create stronger surveillance systems on citizens. While many have focused on the actions of intelligence agencies, local police departments continue to create more sophisticated systems as well. A recent article highlights one new system deployed by the Fresno, California police department, Intrado's Beware. The system scours police data, public records, social media, and public Internet data to provide a "threat level" of a potential suspect or residency. The software is part of a broader trend of military counterinsurgency tools and algorithms being repurposed for civil use. While these tools can help police manage actively dangerous situations, providing valuable intel when responding to calls, the analysis also raises serious civil liberties questions both in privacy (where the data comes from) and accuracy (is the data valid, was the analysis done correctly). Also worrying are the long term ramifications to such technologies: there has already been some speculation about "citizen scores," could a criminal threat score be something similar? At very least, as Matt Cagle of the ACLU noted, "there needs to be a meaningful debate... there needs to be safeguards and oversight."
and even I think this stuff is bad ju-ju. The Extra Creditz youtube blog did a good piece on the Chinese version of this called Sesame Credit.
For those of you wondering how I can stay pro central gov't, I don't see how you can have a world without one. We're going to have a big military to protect us from other countries with big militaries. If you're going to have a big military then you better have a big, strong civilian gov't to counter balance it or you're just asking for a coup de eta. Besides, what else besides a strong central gov't can possibly stand up to a large multi-national corporation?
Think of it this way: It's like there's a box of loader firearms out in the open and somebody picked up a bunch of them and starts waving them around demanding things. Are you gonna sit there and do what they say because you might shoot your eye out or are you gonna pick up a gun and defend yourself? Yeah, you might shoot yourself (heck, it's statistically likely) but it's either that or spend the rest of eternity doing what they guy with the gun says. Gov't is that gun. It's a dangerous tool we're all stuck with...
Hi! I make Firefox Plug-ins. Check 'em out @ https://addons.mozilla.org/en-US/firefox/addon/youtube-mp3-podcaster/
if you are using military counterinsurgency tools and algorithms on the general population, you are just preventing any change that might upset the status quo. the military industrial complex is going to cannibalize the country if shit like this continues.
Anons need not reply. Questions end with a question mark.
That really isn't true. The average pitbull owner is just like you or me. It's just that 500,000 pitbulls didn't attack anyone today, but instead greeted their "owners" with love and affection doesn't sell. When all you hear about is the shady motherfuckers it is easy to get the impression that they are in the majority.
Guns don't kill people; Physics kills people! - John Lithgow as Dick Solomon on Third Rock From The Sun
I came across two pit bulls tearing a German shepherd apart today... ugly mess. Pit bull owner was a 70 year old woman who was helpless. Shepherd finally got free and headed for the hills.
I don't read your sig. Why are you reading mine?
There is one aspect of this that's just begging to be misunderstood by authorities that I'd like to preemptively explain to everybody. Once such a scoring system is in place, there will be some value X for which "an innocent person only has 0.1% chance of scoring higher than X" and that X will become a threshold for suspicion, a threshold that authorities think warrants special treatment and going on various lists as a likely terrorist.
This is analogous to a p-value in statistics. In this case the p-value of an innocent person scoring higher than X by chance is 0.1%.
This will instantly be translated into the completely wrong statement that "there's a 99.9% chance that a person is guilty if they score over X". Those two statements sound really really similar, but they're seriously not the same thing, and in statistics it's an error known as the base rate fallacy.
It's most easily seen by example: suppose only 5 people in a million are actual terrorists, and suppose we run one million people through our test. There are only about 5 terrorists then in our group. Using X as our threshold with its p-value of 0.1%, our scoring system identifies 1000 people as a likely terrorist. So what are the odds that one of those 1000 suspicious people is actually a terrorist? In this example 1000 people were flagged, only 5 were actually terrorists, so despite the fact that "there's only a 0.1% chance that an innocent person will score over X" there's only 0.5% that one of our people who scored over X is actually a terrorist.
Fwiw, this is largely info from http://www.statisticsdonewrong.com/ which gives a great discussion of the p-value and what it does and doesn't mean.