|
|
|
|
|
|
by MrMan
1949 days ago
|
|
|
Some of our applied social scientists are being allowed to work with very large datasets at places like Facebook and I believe they are increasingly in a position to make quantifiable statements on what racists or any other cluster-able groups of humans like. If by spectrum you mean “distance from a centroid” it might be more precise. Marketing is essentially the working, reproducible arm of the social sciences, and marketers know a lot about human preferences and how to link them to motivation. |
|
|
Just because you have more data, it doesn't mean that you can identify constructs like racism from this loads of data. You'd need some kind of ground truth mechanism (like an index of behaviour towards different races) which neither Facebook (nor anyone else) has. It's just wildly implausible.
Maybe, in ten years, NLP will be good enough to identify this, but I don't think these constructs are easily identifiable from text, certainly not in a public space such as Facebook.
> Marketing is essentially the working, reproducible arm of the social sciences, and marketers know a lot about human preferences and how to link them to motivation.
I get what you're trying to say here, and maybe that works in a small number of places, but having worked with marketers in analytics for the past decade or so, suffice it to say that I rarely praise the standards of experimention and (lack of) rigour employed by them.