85% accurate is doing a lot of hiding LOL. Searching a multi-million-face gallery and even high per-comparison accuray turns into mostly false positive. THese systems are only ever defensible as an investigative lead, neve as probable cause.
Which is stupid though because obviously it's just going to be someone that looks like the person they are after. The idea with a lineup is that you have some other sort of evidence, not based on how people look, and then have them identify the person. If you used a tool to find people that look a certain way and then put that person in a line up with other "non-85%" matches it's reasonable the respondent would pick that person.
The idea with a lineup is that you have some other sort of evidence, not based on how people look, and then have them identify the person
No, that's not the idea with lineups; if that was it, you could just show one photo of a single suspect and ask "Is that them?" Which, as you know, has tremendous problems with accuracy of identifications.
Yes, but that's why you use the line up. You don't just show them random people that look like what they described.
It's like, the suspect was a white guy with brown hair in a blue car, and then you get all the white guys with brown hair and blue cars that were in the area and present a line up of them with similar looking people but not having blue cars and being in the area.
To the average person it means: No matter what task you apply the tool to, that you will be right 85% of the time, and 85% is a solid B, a passing grade, so let's use it.
So in other words, if you show it 100 random people and ask who did it, it will tell you to arrest 15 people, with a 15% chance not a single one of them is guilty?
(assuming it will operate independently for every individual you show it)
That's great.
Of course, putting such a system on public cameras, scanning large amounts of people, is about as stupid as some of the biggest disasters in medicine (using a 98% effective drug on random people, regardless of whether it would help them or not. You REALLY want to check what happens if you're wrong before applying it to a large number of people)
Comments
85% accurate is doing a lot of hiding LOL. Searching a multi-million-face gallery and even high per-comparison accuray turns into mostly false positive. THese systems are only ever defensible as an investigative lead, neve as probable cause.
That's why the cops followed up with a photo lineup shown to the victim before applying for a warrant.
Which is stupid though because obviously it's just going to be someone that looks like the person they are after. The idea with a lineup is that you have some other sort of evidence, not based on how people look, and then have them identify the person. If you used a tool to find people that look a certain way and then put that person in a line up with other "non-85%" matches it's reasonable the respondent would pick that person.
No, that's not the idea with lineups; if that was it, you could just show one photo of a single suspect and ask "Is that them?" Which, as you know, has tremendous problems with accuracy of identifications.
Yes, but that's why you use the line up. You don't just show them random people that look like what they described.
It's like, the suspect was a white guy with brown hair in a blue car, and then you get all the white guys with brown hair and blue cars that were in the area and present a line up of them with similar looking people but not having blue cars and being in the area.
"followed up"...
he was extradited from 400 miles away in a different state, had never been to florida, and had timesheets from working at his job at the time.
how did that craziness even pull him into the lineup?
Honestly, at some point this kind of tool is going to find LOTS of similar people from a pool of 350,000,000
We need a new term for this, maybe likeness-fishing.
What does 85% accurate even mean?
To the average person it means: No matter what task you apply the tool to, that you will be right 85% of the time, and 85% is a solid B, a passing grade, so let's use it.
So in other words, if you show it 100 random people and ask who did it, it will tell you to arrest 15 people, with a 15% chance not a single one of them is guilty?
(assuming it will operate independently for every individual you show it)
That's great.
Of course, putting such a system on public cameras, scanning large amounts of people, is about as stupid as some of the biggest disasters in medicine (using a 98% effective drug on random people, regardless of whether it would help them or not. You REALLY want to check what happens if you're wrong before applying it to a large number of people)