In 2011, Bollen et al. published a paper with powerful claims entitled Twitter Mood Predicts the Stock Market. The paper has generated a whole academic industry. It has been cited 3000 times, lead to the creation of workshops… and so forth. Lachanski and Pav recently tried to reproduced the original claims:
Constructing multiple measures of Twitter mood using word-count methods and standard sentiment analysis tools, we are unable to reproduce the p-value pattern that they found. We find evidence of a statistically significant Twitter mood effect in their subsample, but not in the backwards extended sample, a result consistent with data snooping. We find no evidence that our measures of Twitter mood aid in predicting the stock market out of sample.
(…) in his field there were a number of examples of very serious work, including by him, that turned out to have important mistakes. This led him to think that the current practice of believing in a result if enough other experts believe in it was no longer sustainable.
What is most amazing, to me, is how academics are utterly convinced that their own work is beyond reproach. When you ask: “have you had a competing lab reproduce your experiments?” The answer, invariably, is that it is unnecessary. Yet even mathematicians recognize that they have a serious problem avoiding mistakes. It should be clear that of all of scholarship, mathematicians should have the least significant problems in this respect. Nobody “cheats” in mathematics, as far as the mathematical truth is concerned. Truth is not an ambiguous concept in mathematics, you are right or wrong. Yet leading mathematicians grow concerned that “truth” is hard to assess.
This week, the Nobel prizes for 2017 were awarded. No woman received a Nobel prize. At a glance, it looks like caucasian scholars dominate. No Japanese, no Chinese, no Korean researcher that I can see. On a related note, Japan seems to be losing its status as a research power. (So we are clear, I do not believe that caucasian men make better researchers as far as genetics is concerned.)
One of the winners of the medicine Nobel prize, Jeffrey Hall, is quite critical of the research establishment:
I can’t help feel that some of these stars have not really earned their status. I wonder whether certain such anointees are famous because they’re famous. So what? Here’s what: they receive massive amounts of support for their research, absorbing funds that might be better used by others. As an example, one would-be star boasted to me that he’d never send a paper from his lab to anywhere but Nature, Cell, or Science. These submissions always get a foot in the door, at least. And they are nearly always published in one of those magazines where, when you see something you know about, you realize that it’s not always so great.
Authorea has published a list of eight research papers that were initially rejected, but which ended up being rewarded by a Nobel prize. This is an illustration of the fact that it is very difficult for even the best experts to recognize the significance of some work as it happens.
It appears that under the Trump presidency, the Food and Drug Administration (FDA) has been approving new drugs at twice the “normal” rate.
Google will commercialize headphones that should allow you to understand (through just-in-time translation) over 40 different languages. The pixel buds will sell for under $200.
My wife asked me, the other day, whether people in China used hyperlinks made of Chinese characters. I assumed so. It turns out that the answer involves something called punycode which is a way to encode arbitrary characters as ASCII (English-only) characters.
From 1989 to 2015, breast cancer death rates decreased by 39%, which translates to 322,600 averted breast cancer deaths in the United States.
To be clear, we are still very far from having a breast-cancer cure, let alone a cancer cure.
On a related note, over half of the new cancer drugs approved in recent years do not improve your odds of surviving cancer. What happens, apparently, is that drugs are approved on the basis of narrow measures that may or may not translate into concrete health benefits.
How likely is a medical professional to prescribe unnecessary therapies? More so than we’d like:
We find that the patient receives an overtreatment recommendation in more than every fourth visit.
One of the effect of aging is that our telomeres get shorter. With every cell division, this non-coding piece of your DNA gets shorter and shorter. At the margin, this may affect the ability of your tissues to regenerate. TRF1 protects your telomeres, and it appears that having more of it could be helpful:
We further show that increasing TRF1 expression in both adult (1-year-old) and old (2-year-old) mice using gene therapy can delay age-associated pathologies.
A common theme in politics these days is “inequality”: some people are getting richer while others are not. In turn, this inequality can be related to a falling share of income that goes toward salaries. That is, a society where most of the wealth is distributed as salaries tends to be more “equal”. Right now, it appears that a lot of wealth goes into property values in a highly desirable areas, for example. Since the poor do not own buildings in Manhattan, they do not benefit from this kind of wealth distribution. So why aren’t we getting larger salaries? Researchers for Harvard, Princeton and the London School of Economics believe that this fall could be explained by our low productivity. Of course, even if they are correct, this just leads us to another question: why aren’t we getting more productive?
In related news, some scholars from Stanford believe that research productivity is way down. Apparently, new good ideas are getting harder to find. From my corner of the world (software), this looks like an incredible claim. I cannot even superficially keep up with even a narrow subset of the software industry. There are significant advances left and right… too much for my little brain. Speaking for myself, I certainly have no shortage of what appears to me to be good ideas. I am mostly limited by my ability to find the energy to pursue them… and by the fact that I want to spend quality time with my family. I cannot believe that all the researchers, many much smarter than I’ll ever be, are finding new ideas harder to find.
It has recently been shown that Guinea baboons can be trained to discriminate between four-letter words (e.g., TORE, WEND, BOOR, TARE, KRIS) and nonwords (e.g., EFTD, ULKH, ULNX, IMMF) simply by rewarding them for correct lexicality decisions. The number of words learned by the six baboons ranged from 81 to 307 (after some 60,000 trials), and they were reported to respond correctly to both novel words and nonwords with above chance performance.
It looks like the company that is reinventing the taxi industry, Uber, might pull out from Quebec, Canada. At issue is the requirement to having several days of training before one can act as a taxi driver.