Regulating Homeopathic Treatments, Thousands of Toenails, and A.I. in the Classroom

JOHN DANKOSKY: This is Science Friday. I’m John Dankosky. Ira Flatow is away.

When it comes to Twitter, you’ve got your run-of-the-mill online interactions, tweets, and hashtags. There’s even passive aggressive sub-tweeting. Some topics are trending. But what conditions are necessary to bring on a full fledged Twitter storm?

When it comes to discussing the field of science, it seems Twitter may be amplifying a big problem– sexism. And while it may be easy for some to dismiss these seemingly fleeting engagements, these conversations are bringing to light discrimination and harassment that consistently dog the field of science. Joining me now to discuss the nexus of science and sexism on Twitter and other select short subjects and sciences, Virginia Hughes. She’s science editor at Buzzfeed News in New York, and she joins me in the studio. Welcome back to the program, Virginia.

VIRGINIA HUGHES: Thank you very much, John. Good to be here.

JOHN DANKOSKY: So what are some instances of sexism in science that have gotten a lot of attention on Twitter?

VIRGINIA HUGHES: So this all came from a fascinating story in Nature this week. I’ll just rattle off a few things that you probably remember. Late last year, when the European Space Agency landed that spacecraft on a comet, one of the project’s leaders ended up wearing this shirt with scantily clad women all over it.

JOHN DANKOSKY: Yeah, not a great look.

VIRGINIA HUGHES: Yeah, doing a ton of interviews. So that led to #shirtstorm and several other hashtags, I think. So that was one of the first ones recently.

Then there was a woman– her name was Fiona Ingleby– who tweeted about a sexist peer review that one of her papers received at PLOS. Then there was a Tim Hunt saga, the Nobel Prize winner who ended up making some sexist remarks at a conference, and those remarks were first reported on Twitter. And then most recently, about a month ago, after Buzzfeed broke the story about astronomer Geoff Marcy’s sexual harassment, it promoted this huge discussion among astronomers on Twitter.

So these Twitter storms, as the Nature article gets into a little bit, they have this pro and con where it’s a really great way, maybe, to draw awareness and get things done. For instance, the woman who tweeted about her paper’s peer review ended up getting an apology from the journal, and she got a new review. Tim hunt stepped down from a few positions.

So these things can lead to change, but they can also lead to some really nasty reactions– death threats, rape threats. Twitter can be this stew of hate sometimes.

JOHN DANKOSKY: Yeah, Twitter can have its good its bad side. Let’s turn to a 200-year-old form of alternative medicine– homeopathy. And it’s facing some new scrutiny now.

VIRGINIA HUGHES: Yes. So this week, the FDA closed a public comment period, which is kind of one of these FDA terms that doesn’t really mean much. But basically the FDA is finally looking into homeopathy remedies, which have been pretty much unregulated so far.

In case you don’t know what it is, it’s a huge business. $6.4 billion. And these remedies– the basic idea is that if you take the thing that makes you sick and you dilute it by huge factors of many, many thousands of times, then that will cure you.

It doesn’t have any basis in scientific reality, but it’s very, very popular. And it has this fascinating history. It debuted 200 years ago, as you said, in Germany. And I’ll just give one very weird, popular example where, after the Spanish flu, a French physician claimed that this imaginary germ was responsible for the flu, and that this germ just happen to grow on duck livers.

So he took duck liver, diluted it many, many, many times over, and started selling it as something called oscillococcinum flu treatment. Which, you know, is nonsense. In my opinion.

JOHN DANKOSKY: In your opinion. But are these things just harmless placebos, are can they hurt you?

VIRGINIA HUGHES: So the vast majority are probably harmless placebos. They’re sugar pills, most of them. However, there are these examples of seeming harm. In 2009, the FDA received more than 130 accounts of people who lost their sense of smell after taking a Zicam homeopathic remedy, which is a nasal spray.

There was another one– teething tablets that were marketed as 100% percent natural, that actually apparently led to some seizures and slowed breathing in kids. So I think these safety things are pretty rare. Most the time these things are not going to hurt you, but you never know.

JOHN DANKOSKY: Let’s turn to another story. Almost 10 years after faking data for a number of cancer studies, a scientist is just now being reprimanded by the federal government. Tell us what the scientist and why this took so long.

VIRGINIA HUGHES: Yeah, took forever. So this guy’s name is Anil Potti, and he used to be a cancer researcher at Duke University. And he’s been known for years as one of the most notorious fraudsters in science, but just this Monday the federal government’s Office of Research Integrity sort of officially deemed him a fraud. So he faked data– officially now– in nine papers that were related to six federal grants, and his research was really– you could see why it got into all these big studies.

So he claimed to have found genetic signatures that could predict which cancer patients would respond to particular chemotherapies. And he did all this stuff in the lab, and then he also launched human clinical trials based on this. It turned out to be a fraudulent idea.

So the craziest thing to me is, as of 2012, Retraction Watch, the blog that often reports on misconduct, reported that he was still a practicing oncologist in North Dakota. And this week the Washington Post confirmed that he was still apparently there. He’s still on the voice answering machine at this place.

I think the bottom line is the federal process for formally sanctioning fraud is a really slow. Like, ridiculously slow.

JOHN DANKOSKY: We’re going to one last story here. You bring us one about a bunch of freezers in a Harvard basement that are chock-full something kind of gross. Toenail clippings? What can we learn from toenail clippings?

VIRGINIA HUGHES: Yeah, this is such a fun story. It came from a brand new publication about science and health news called Stat. And yeah, it’s about these Harvard freezers full of toenails. And it’s part of a project called the Nurses’ Health Study, which has been collecting health data and samples from nurses for several decades.

So they have 3 and 1/2 million samples of blood and urine, and then, as you say, 100,000 samples and toenails. So why toenails? When you look at a blood sample, that is representing a biological signature at a given moment in time. Your blood can change after you eat or exercise or whatever.

But toenails can apparently– they contain traces of chemical exposures like arsenic or mercury that can give information about sort of a month-to-month time scale of your body. And the weirdest thing is that your toenails grow at different speeds. So apparently a clipping from your little toe could get a snapshot of your exposures for about a month, whereas your big toe might give a snapshot of about a year’s worth of exposures.

JOHN DANKOSKY: So to really get anything, you’ve got to have a lot of toenails.

VIRGINIA HUGHES: You’ve got to clip and clip.

JOHN DANKOSKY: OK, well, that leaves me feeling a little funny. We’ve run out of time. I want to thank Virginia Hughes, science editor at Buzzfeed News in New York. Thanks so much, Virginia.

VIRGINIA HUGHES: No problem.

JOHN DANKOSKY: Now it’s time to play Good Thing, Bad Thing. Because every story has a flip side, imagine showing up to math class. You’re in your seat surrounded by classmates, some of whom have already mastered the material, others who continue to struggle.

Class starts, but instead of your teacher going to the front of the classroom and writing out equations, you start working by yourself on a computer program called ALEX. This adaptive software gives you problems tailor-made to your comprehension level, and it sounds like it could be pretty great, but it’s not quite that simple.

Joining me now to talk about this adaptive software increasingly pushed by textbook companies as they look to reinvent themselves is Will Oremus. He’s senior technology writer at Slate in New York. He joins me in the studio. Welcome back to the show, Will.

WILL OREMUS: Thanks for having me.

JOHN DANKOSKY: So what are some of the good things about introducing artificially intelligent software into the classroom?

WILL OREMUS: Well, the good thing is we may finally found a way to improve on a 500-year-old technology, which is the printed textbook. I’m not just talking about ebooks. This is software that uses machine learning algorithms and other sorts of algorithms to actually try to learn the strengths and weaknesses of each student as they work through the material. And in some cases, the way this is being implemented in classrooms is actually changing the whole classroom experience.

So in the old model, you have a teacher who stands up and lectures to, let’s say, 20 students at once. They go home at night, and they all work on the same pages of the same textbook, do the same practice problems. Now, they came in probably at very different skill levels.

Some of them are going to be ahead. They’re going to be bored by the material. Others are going to be falling behind, but the class will slow down for them, so they might just get lost and tune out. They might decide they’re bad at math. A lot of people think they’re really bad at math. Maybe it’s the system that’s failing them.

So in the implementation that I saw at a community college in Valhalla, New York, the students came into class and sat down at the computers, as you described. They all worked through the same material, but at a different pace, so the software would guide them through the material. It wouldn’t move them on to the next subject until they’d mastered one. They could set their own course, to some extent. There’s some flexibility involved there.

The advantage is they’re eventually all going to learn the same stuff. They’re just going to learn it at different rates. So the students who are falling behind don’t necessarily fail. They just may take longer than a semester to pass the class. The students who are ahead could pass the class halfway through the semester, go ahead and pass another class in the same semester.

JOHN DANKOSKY: So that all sounds really great. What could be the downside of that?

WILL OREMUS: Yeah, it does sound great, especially when you listen to the textbook companies that have invested their future on this type of technology. It’s actually a lot more profitable for them, potentially, because whereas you can borrow a friend’s textbook, or Xerox some pages, or buy it used, the technology can require a digital license for every single student, every year. So that’s a good thing for the textbook companies.

What could be the bad thing? You have to think a little bit harder about what the real goal of education is. So when you’re individualizing the material for each student, it’s nice, because each student is able to go at their own pace. But you also lose that thing where students are working on the same thing at the same time, and they can help each other out. One of them gets stuck, and the other one has to learn the skill of how do you teach a peer to get over a hump in a problem. They have to learn patience working with each other.

And those kinds of software skills maybe don’t get picked up as easily in a class where the computer is designed to give you exactly the material you’re ready for, exactly when you’re ready for it. And the other issue is that there are softer skills– sorry, there are softer skills involved in the math reasoning itself.

So there’s a famous problem that was posed in the 1986 research paper. It’s called “How old is the shepherd?” So bear with me for a second here. A bunch of great school students we’re asked, “There are 125 sheep and 5 dogs in a shepherd’s field. How old is the shepherd?” The majority of them gave some kind of numerical answer. They would say 125 divided by 5, the shepherd’s 25. And it’s because they’ve been trained to apply formulas by rote to problems rather than stepping back, asking what is this all about, why am I being asked this.

JOHN DANKOSKY: OK, some good things and some bad things there. Thanks for bringing us this story, Will. Will Oremus is senior technology writer at Slate in New York. He joined us here in studio. Thanks again, Will?

WILL OREMUS: Thanks.