DataTaunew | comments | leaders | submitlogin
3 points by TheKyleAmbert 2549 days ago | link | parent

Should data science deliver a fake news detector?


3 points by ddandur 2549 days ago | link

As long as there isn't an incentive to know what's true, nobody will agree on what constitutes "fake news." That's why political fake news faces such a problem: people post these stories as a form of tribal signaling, not as a source of information. It's not to hard to make a rough "ideology detector", but a true fake news detector is much harder.

A real fake news detector (where "fake news" is defined to be stories that are demonstrably false in their major point) is more likely to come from a betting market (see Robin Hanson). Look at areas where people are rewarded monetarily for being correct, that is where the action is. I would guess some of the best fake news detectors in the world right now are being developed at banks and hedge funds.

-----

1 point by flohad 2549 days ago | link

That's exactly what the post is about. And you're right about the betting market - though my money is more on the entertainment industry. Think of the usual early adopters of emerging technology.

-----

1 point by flohad 2549 days ago | link

That's obviously an entirely different question. Either way, it's happening.

-----

1 point by TheKyleAmbert 2548 days ago | link

Indeed. It's a cool idea, but it would be nice to also address the lack of critical thinking that probably lead to the rise of fake news in the first place. Granted, that's less of a machine learning problem.

-----




RSS | Announcements