4 ms·
There's an in between situation where someone is looking for correlations with a crime, but they are simply searching so much data, that they inevitably get a h
by perl4ever 5y ago
There's an in between situation where someone is looking for correlations with a crime, but they are simply searching so much data, that they inevitably get a huge number of very close, but invalid matches.
This tends to be a problem more and more with ever increasing databases that are convenient to access.
People have intuitions for coincidences that are not appropriate for vast amounts of data.
- framecowbird 5y agoThis is a really great insight, and demonstrates where the "library book" example is lacking. Since Google is so vast, I can find a page of search hits to support any particular belief I might have, no matter how wild.
- 542354234235 5y agoThings like this make me think of the movie Gattaca. Mild Spoilers After the murder, they just go around vacuuming up any and all organic material to test for DNA. They get a hit from someone who “isn’t supposed to be there” and immediately focus the entire investigation on that person. It fits the very real pattern of using science to just confirm the human biases held by the people using it. If police can sweep up and search through a mountain of data, it is going to be the Muslim immigrant hit they focus on over the white middle manager. It will be the person struggling off and on with homelessness that gets arrested because they matched a pattern and didn’t have an alibi, over the suburban mom. That if you give police a list of random correlations, they will focus on “those people” that are assumed to be more criminal and usually have less means and ability to fight back.