Profile cover photo
Profile photo
Yew Jin Lim
771 followers -
Professional cannon fodder; amateur stream-entry seeker
Professional cannon fodder; amateur stream-entry seeker

771 followers
About
Yew Jin's posts

Post has attachment

Post has attachment
Debunking myths is problematic. Unless great care is taken, any effort to debunk misinformation can inadvertently reinforce the very myths one seeks to correct. To avoid these “backfire effects”, an effective debunking requires three major elements. First, the refutation must focus on core facts rather than the myth to avoid the misinformation becoming more familiar. Second, any mention of a myth should be preceded by explicit warnings to notify the reader that the upcoming information is false. Finally, the refutation should include an alternative explanation that accounts for important qualities in the original misinformation.

https://www.skepticalscience.com/docs/Debunking_Handbook.pdf

Post has attachment
"To test this, Redlawsk and his team created a mock presidential election in which people would gradually learn more and more terrible things about their preferred virtual candidates from a virtual news media. Unbeknownst to the subjects, the news stories they read included a precise mix of negative information about their chosen candidates so the effect of those messages could be measured as the negativity increased in intensity.

The scientists thought that surely, at some point, after a person had chosen one candidate over another, a constant flow of negative information about that person would persuade them to reconsider their choices. They expected to see the backfire effect at first, of course, but they believed with enough persistence they might also discover its natural limit.

...

Also in this episode, psychologist Stephan Lewandowsky takes us step-by-step through The Debunking Handbook, a guide he and Peter Cook wrote for avoiding the backfire effect when confronting vaccine and climate change deniers. Originally meant to be an instruction manual for science communicators, it can be applied to just about any situation where the facts are on your side yet the people who need to hear them are dead set on keeping belief-threatening ideas out of their heads."

https://youarenotsosmart.com/2017/02/11/yanss-095-how-to-fight-back-against-the-backfire-effect/

Post has attachment
"First, stop believing that protests alone do much good. Protests galvanize groups and display strong opposition, but they’re not sufficient. Not only are they relatively ineffective at changing policy, they’re also falsely cathartic to those protesting. Protestors get all kinds of feel-good that they’re among fellow believers and standing up for what’s right, and they go home feeling like they’ve done their part. Even if protestors gain mild, symbolic concessions, the fact that their anger has an outlet is useful to the other side. Do protest, but be very wary of going home feeling like you’ve done your job. You haven’t."

https://medium.com/@jakefuentes/the-immigration-ban-is-a-headfake-and-were-falling-for-it-b8910e78f0c5#.qw2vq2kpm

Post has attachment
In 1983, the U.S. District Court in San Francisco formally vacated Korematsu's conviction. At the time, he told Judge Marilyn Patel that instead of a legal pardon, he wanted to be assured the U.S. government would never again take such an action.

"If anyone should do any pardoning," he said, "I should be the one pardoning the government for what they did to the Japanese-American people."

http://www.npr.org/sections/thetwo-way/2017/01/30/512488821/its-fred-korematsu-day-celebrating-a-foe-of-u-s-internment-camps

Post has attachment
"The backfire effect is a name for the finding that, given evidence against their beliefs, people can reject the evidence and believe even more strongly. The phrase was first coined by Brendan Nyhan and Jason Reifler."

https://en.m.wikipedia.org/wiki/Confirmation_bias

Post has attachment
Being out of power doesn't mean being in the minority.

https://www.youtube.com/watch?v=H76e8yIbXi4

Post has attachment
I spent the weekend coding up the supervised policy network potion of AlphaGo from scratch.

I think it's completely bonkers that the network, given only a bitmap of the stones (player, opponent) and a bitmap of 1s and 0s, could predict the next move in professional games 20+% of the time after a few minutes of training on my Macbook Air (i.e., on CPU).

MNIST used to be at least a small-ish project in an undergraduate AI class, so was coding up a checkers/chess playing program.

Pretty soon, coding up a professional-level Go program will be the norm.

https://storage.googleapis.com/deepmind-media/alphago/AlphaGoNaturePaper.pdf

Post has attachment
Maybe that's starting on Mythbusters a little too early in life?

Photo

Post has attachment
Good policy-making (and any large project) is about making tough choices. I really hope that somebody "solves" healthcare in US, with the caveat that it will make a good set of compromises that ensures the maximal public utility.

https://www.youtube.com/watch?v=aduTfKE5IOM
Wait while more posts are being loaded