Tue Oct 31 20:11:15 EDT 2017

Items of Interest

Various web links I found to be of interest recently.

  • We Libertarians Were Really Wrong About School Vouchers

    Some studies suggest that voucher programs do modest good; others suggest that they do very little; and a few suggest that the impacts are actually negative. My overall takeaway from the literature is that voucher programs probably do a little bit of good. But the emphasis is on the word "little"; they are not a cure-all, or even much of a cure for anything. It was reasonable to think, in 1997, that voucher programs could change the world. Now we have two decades of evidence.
    ...
    But the quality of the pedagogy isn't the only thing that shapes student outcomes in schools. The peer group matters a great deal; families with higher socioeconomic status are better able to navigate the educational system, and they value education very highly, traits they pass on to their children. Those parents also work hard to improve the quality of the schools their children attend.

    The socioeconomic status of the students in a school is somewhat easier for parents to observe than the quality of the pedagogy. It's not then, all that surprising that when researchers sat down to analyze parental decision-making in New York City public school, peer group seemed to be what parents were looking at. And peer group matters a great deal.

  • DeepMind's Go-playing AI doesn't need human help to beat us anymore

    The company's latest AlphaGo AI learned superhuman skills by playing itself over and over

    The original AlphaGo demonstrated superhuman Go-playing ability, but needed the expertise of human players to get there. Namely, it used a dataset of more than 100,000 Go games as a starting point for its own knowledge. AlphaGo Zero, by comparison, has only been programmed with the basic rules of Go. Everything else it learned from scratch. As described in a paper published in Nature today, Zero developed its Go skills by competing against itself. It started with random moves on the board, but every time it won, Zero updated its own system, and played itself again. And again. Millions of times over.
    ...
    After three days of self-play, Zero was strong enough to defeat the version of itself that beat 18-time world champion Lee Se-dol, winning handily -- 100 games to nil. After 40 days, it had a 90 percent win rate against the most advanced version of the original AlphaGo software. DeepMind says this makes it arguably the strongest Go player in history.

  • This year's economics Nobel winner invented a tool that's both brilliant and undemocratic

    "Nudges" aren't good for democracy.

    The problem -- as Carnegie Mellon's Cosma Shalizi and I have discussed elsewhere -- is that government-by-nudging amounts to a kind of technocracy, which assumes that experts will know which choices are in the interests of ordinary people better than those people know themselves. This may be true under some circumstances, but it will not be true all of the time, or even most of the time, if there are no good opportunities for those ordinary people to voice their preferences.

  • The New York Times: Promoting False Hope as Journalism

    Criticism of book The Other Side of Impossible promoting alternative medicine.

    We have already explored at length the problems with anecdotal evidence. Meadows' casual dismissal of these problems by stating that "an example of one" has meaning in terms of giving hope is itself a harmful deception. You could probably write a book filled with stories of people who played the lottery as the solution to their financial situation and won. You could cherry pick the winners, and then explore what they did to choose their winning numbers. a statistically negligible hope. Focus on the hope and the winners.
    ...
    Of course, only the most dramatic examples are going to make it into such a book. They are by definition not representative. There are also all the other problems with anecdotal evidence. We don't really know if the original diagnosis was accurate. We don't know which of the many treatments used were actually effective. We don't know that the outcome wasn't just the natural course of the illness or a spontaneous remission. There is also a lot of subjective judgement involved.

    The original New York Time article by Jane E. Brody is,
    Hitting a Medical Wall, and Turning to Unproven Treatments

    Also see this criticism of the original article from Respectful Insolence in Science Blogs,
    The New York Times publishes fake news false hope in the form of a credulous account of dubious alternative medicine testimonials

  • Researchers have ditched the autism-vaccine hypothesis. Here's what they think actually causes it.

    Genes and the microbiome are some of the most promising leads.

    Of all the issues doctors have explored in children's health, none has been more exhaustively researched than the question of whether vaccines are linked to autism. After hundreds and hundreds of studies in thousands of children, "We can say with almost as much certainty than anybody could ever say that vaccines don't cause autism," Mayo Clinic autism researcher Dr. Sunil Mehta told me.
    ...
    "The bottom line is that when you add up all of the genetic risks, it looks like genetics can account for 50 percent of the risk for autism, which is very high," said David Amaral, an autism specialist at the UC Davis MIND Institute. To put that into context, compared to other common health problems -- like heart attacks, or cancer -- autism is much more genetic, with well over 100 genes now implicated.
    ...
    Exposure to infections and certain medicines during pregnancy may be linked to autism.
    ...
    The disorder seems to affect boys about four times more than girls.

  • Meet the nocebo effect, the placebo effect's evil twin that makes you feel pain

    A fascinating new study finds patients report worse side effects when a drug costs more money.

    Doctors even see a placebo response in patients who are told they are on a placebo. And the more invasive, expensive, and drastic the placebo intervention, the greater the healing effect. Fake surgeries -- where doctors make some incisions but don't actually change anything -- make people feel better than placebo pills alone.
    ...
    But the placebo effect has an evil twin: the nocebo. It can kick in when negative expectations steer our experience of symptoms and create side effects where none should occur.

    This means, incredibly, that you can get side effects from a sugar pill. And sometimes these side effects are so severe that patients drop out of clinical trials, as a 2013 paper in Nature Reviews explains. A review of fibromyalgia drug trials revealed that 72 percent of people who left the trial did so because they felt severe side effects while on placebo.
    ...
    When patients are led to believe one drug is less expensive than another they're also less likely to report painful side effects.


Posted by mjm | Permanent link | Comments
comments powered by Disqus