Best of Twitter - Week of September 28, 2020
Such good, good news about a truly dreadful disease! Registry-based study from Sweden: half a million young women who had HPV vaccination, over a million who didn't. Substantial drop in invasive cervical cancer nejm.org/doi/full/10.10… My 2018 backgrounder absolutelymaybe.plos.org/2018/08/31/the…
“most people don’t respond to most economic incentives” is totally compatible with “economic incentives rule the world and control everything around us.”
- @slatestarcodex
Are you intermittent fasting? You should definitely read this. Actually, everybody should read it. Ethan intermittent fasted for 7 years, realized there wasn’t good data in humans, and then ran a clinical trial. He changed his own mind. Exceptional.
Ethan Weiss @ethanjweiss
—>
22/ This is important. If this had been an uncontrolled study, we'd be saying there was a significant decrease in weight. But there was that pesky control arm. We know that people in clinical trials to lose weight no matter what you do. Including the control arm was important
29/ We were surprised by these results. At that time, I had been doing 16:8 TRE for ~7 years. I stopped. This study taught me a lot. More later but the take home lessons for me are: 1) it is possible to do RCTs in nutrition; 2) ignore all nutrition studies without a control group
is Harvard employing someone a strong signal of this researcher's competence? Probably not if the entire scientific field is bs
reddit.com/r/slatestarcod…
The NIPS Experiment:"In particular, about 57% of the papers accepted by the first committee were rejected by the second one and vice versa. In other words, most papers at NIPS would be rejected if one reran the conference review process (with a 95% confidence interval of 40-75%)"
Rose Yu @yuqirose
Found an incredible Reddit thread, "What's a rule that was implemented somewhere that massively backfired?"
Here are the best ones 🧵
1/ Alcohol bans at college football games led to increased intoxication problems b/c fans were getting really drunk before entering the stadium.
"There's no amount of genius that can invent fletching if you're a cave man, it's purely a product of dumb luck, and some cultures just didn't happen to have the right collection of flukes align for them."
Mind-blowing stuff:
Facebook's TransCoder is a deep learning algorithm that's capable of migrating code from one language to another.
(Just stop for a second, and think about this!)
Produced code even pass unit tests!
Quick read: ai.facebook.com/blog/deep-lear…
An example of a “visual circuit” using visual stimuli and illusions to trick your brain into carrying out arbitrary computations: changizi.com/uploads/8/3/4/…