Category: Science

*The AI Does Not Hate You*

That is the new book by Tom Chivers, and the subtitle is Superintelligence, Rationality and the Race to Save the World.  Here is one excerpt:

Overall, they have sparked a remarkable change.  They’ve made the idea of AI as an existential risk mainstream; sensible, grown-up people are talking about it, not just fringe nerds on an email list.  From my point of view, that’s a good thing.  I don’t think AI is definitely going to destroy humanity.  But nor do I think that it’s so unlikely we can ignore it.  There is a small but non-negligible probability that, when we look back on this era in the future, we’ll think that Eliezer Yudkowsky and Nick Bostrom  — and the SL4 email list, and — have saved the world.  If Paul Crowley is right and my children don’t die of old age, but in a good way — if they and humanity reach the stars, with the help of a friendly superintelligence — that might, just plausibly, be because of the Rationalists.

There is also material covering Scott Alexander and Robin Hanson, among others.  Due out in the UK in June.

Gender and the confidence gap

That is the topic of my latest Bloomberg column, here is one bit:

The first new study focuses on performance in high school, and the startling result is this: Girls with more exposure to high-achieving boys (as proxied by parental education) have a smaller chance of receiving a bachelor’s degree. Furthermore, they do worse in math and science, are less likely to join the labor force, and more likely to have more children, which in turn may limit their later career prospects.

Those are disturbing results.  Exposure to high-achieving peers is normally expected to be a plus, not a minus. It is what parents are trying to do when they place their children into better schools, or when school systems work hard to attract better students.


A second new study finds that even blind review does not avoid gender bias in the processing of grant proposal applications, drawn from data from the Gates Foundation. It turns out that women and men have different communications styles, with the women more likely to use narrow words, and the men more likely to use broader ones. And reviewers, it turns out, favor broad words, which are more commonly associated with more sweeping claims, and disfavor the use of too many narrow words.

The net result is that “even in an anonymous review process, there is a robust negative relationship between female applicants and the scores assigned by reviewers.” This discrepancy persists even after controlling for subject matter and other variables. Notably, however, it disappears when controlling for different rhetorical styles.

These two studies probably are connected to each other. While the two sets of researchers do not address each other’s claims, it is not a huge leap to think of broader, more sweeping language as reflecting a kind of confidence, whether merited or not. Narrow words, on the other hand, may reflect a lower level of confidence or a greater sense of rhetorical modesty. Not only might lower confidence hurt many women in life, but a greater unwillingness to signal confidence — regardless of whether it’s genuine — might hurt them too.

There is much more at the link, recommended.

Peter Thiel on medicine and longevity

Or is it that there’s something wrong with culture, with the funding?  Almost no grants go to younger scientists.  When it’s scientists under age 40 that make […] of the most big discoveries, 2% of NIH grants go to scientists under age 40.  That seems a little bit off.  You have a peer-review process where anything heterodox can’t get funded.  You have sort of a publish or perish dynamic where you have to do small, incremental things to publish lots of articles that don’t add up to anything ever…

And again, my sort of libertarian cut on what happened would be the history of was that we had a healthy, scientific world that was non-governmental.  It was decentralized.  It was idiosyncratic.  Different people were doing different kinds of things.  And in the 1930s, 1940s, it got centralized accelerated.  The Manhattan Project…there was actually a way you could accelerate science temporarily by adding tons of money and centralizing…

So the centralization worked.  But to use an ecological metaphor, it worked by creating a monoculture.  And we’re now two generations in to where that monoculture has been just catastrophic.

That is from this taped dialogue between Peter and Bill Hurlbut, previously linked on MR.

Is Dentistry Safe and Effective?

The FDA may be too conservative but it does subject new pharmaceuticals to real scientific tests for efficacy. In contrasts, many medical and surgical procedures have not been tested in randomized controlled trials. Moreover, dental care is far behind medical care in demanding scientific evidence of efficacy. A long-read in The Atlantic spends far too much time on a single case of egregious dental fraud but it’s larger point is correct:

Common dental procedures are not always as safe, effective, or durable as we are meant to believe. As a profession, dentistry has not yet applied the same level of self-scrutiny as medicine, or embraced as sweeping an emphasis on scientific evidence.

…Consider the maxim that everyone should visit the dentist twice a year for cleanings. We hear it so often, and from such a young age, that we’ve internalized it as truth. But this supposed commandment of oral health has no scientific grounding. Scholars have traced its origins to a few potential sources, including a toothpaste advertisement from the 1930s and an illustrated pamphlet from 1849 that follows the travails of a man with a severe toothache. Today, an increasing number of dentists acknowledge that adults with good oral hygiene need to see a dentist only once every 12 to 16 months.

The joke, of course, is that there’s no evidence for the 12 to 16 month rule either. Still give credit to Ferris Jabr for mentioning that the case for fluoridation is also weak by modern standards–questioning fluoridation has been a taboo in American society since anti-fluoridation activists were branded as far-right conspiracy theorists in the 1950s.

The Cochrane organization, a highly respected arbiter of evidence-based medicine, has conducted systematic reviews of oral-health studies since 1999….most of the Cochrane reviews reach one of two disheartening conclusions: Either the available evidence fails to confirm the purported benefits of a given dental intervention, or there is simply not enough research to say anything substantive one way or another.

Fluoridation of drinking water seems to help reduce tooth decay in children, but there is insufficient evidence that it does the same for adults. Some data suggest that regular flossing, in addition to brushing, mitigates gum disease, but there is only “weak, very unreliable” evidence that it combats plaque. As for common but invasive dental procedures, an increasing number of dentists question the tradition of prophylactic wisdom-teeth removal; often, the safer choice is to monitor unproblematic teeth for any worrying developments. Little medical evidence justifies the substitution of tooth-colored resins for typical metal amalgams to fill cavities. And what limited data we have don’t clearly indicate whether it’s better to repair a root-canaled tooth with a crown or a filling. When Cochrane researchers tried to determine whether faulty metal fillings should be repaired or replaced, they could not find a single study that met their standards.

Enraging dinosaur bone markets in everything

For sale on eBay: what’s claimed to be “maybe the only” young Tyrannosaurus rex ever discovered for $2.95 million. Paleontologists decried the sale, saying that the specimen’s cost was artificially inflating the cost of other valuable fossils. “Only casts and other replicas of vertebrate fossils should be traded, not the fossils themselves,” an open letter from the Society of Vertebrate Paleontology in Bethesda, Maryland. read. “Scientifically important fossils like the juvenile tyrannosaur are clues to our collective natural heritage and deserve to be held in public trust.”

…“The asking price is just absurd,” one researcher said.

Here is the full story, via Ze’ev.  Might this increase the incentive to find such fossils?

Early-career setback and future career impact

Setbacks are an integral part of a scientific career, yet little is known about whether an early-career setback may augment or hamper an individual’s future career impact. Here we examine junior scientists applying for U.S. National Institutes of Health (NIH) R01 grants. By focusing on grant proposals that fell just below and just above the funding threshold, we compare “near-miss” with “near-win” individuals to examine longer-term career outcomes. Our analyses reveal that an early-career near miss has powerful, opposing effects. On one hand, it significantly increases attrition, with one near miss predicting more than a 10% chance of disappearing permanently from the NIH system. Yet, despite an early setback, individuals with near misses systematically outperformed those with near wins in the longer run, as their publications in the next ten years garnered substantially higher impact. We further find that this performance advantage seems to go beyond a screening mechanism, whereby a more selected fraction of near-miss applicants remained than the near winners, suggesting that early-career setback appears to cause a performance improvement among those who persevere. Overall, the findings are consistent with the concept that “what doesn’t kill me makes me stronger.” Whereas science is often viewed as a setting where early success begets future success, our findings unveil an intimate yet previously unknown relationship where early-career setback can become a marker for future achievement, which may have broad implications for identifying, training and nurturing junior scientists whose career will have lasting impact.

That is the abstract of a new paper by Yang Wang, Benjamin F. Jones, and Dashun Wang.

Samir Varma on the forthcoming tech breakthroughs

From my email:

We are now starting to get a hint of the future transformative technologies that you guessed were on their way in “The Great Stagnation”. You had not speculated on what they might be, but there are faint hints on what is likely to happen.

I believe this article is one leg: extremely fast air travel. The second leg is the Hyperloop and similar: extremely fast ground travel. The third leg is synthetic biology (e.g: The fourth leg is quantum computing, which is finally starting to show that it might work. And the fifth, and final leg, is fusion energy, which looks eerily like it will actually come to fruition this time.
Put those 5 together and you have the makings of a new economy, with a huge burst of growth to come for many decades. These are just faint hints, of course, but they’re starting to get increasingly clear.

My Conversation with Ed Boyden, MIT neuroscientist

Here is the transcript and audio, highly recommended, and for background here is Ed’s Wikipedia page.  Here is one relevant bit for context:

COWEN: You’ve trained in chemistry, physics, electrical engineering, and neuroscience, correct?

BOYDEN: Yeah, I started college at 14, and I focused on chemistry for two years, and then I transferred to MIT, where then I switched into physics and electrical engineering, and that’s when I worked on quantum computing.

COWEN: Five areas, actually. Maybe more.

BOYDEN: Guess so.

COWEN: Should more people do that? Not the median student, but more people?

BOYDEN: It’s a good question.


COWEN: Are we less creative if all the parts of our mind become allies? Maybe I’m afraid this will happen to me, that I have rebellious parts of my mind, and they force me to do more interesting things, or they introduce randomness or variety into my life.

BOYDEN: This is a question that I think is going to become more and more urgent as neurotechnology advances. Already there are questions about attention-focusing drugs like Ritalin or Adderall. Maybe they make people more focused, but are you sacrificing some of the wandering and creativity that might exist in the brain and be very important for not only personal productivity but the future of humanity?

I think what we’re realizing is that when you intervene with the brain, even with brain stimulation, you can cause unpredictable side effects. For example, there’s a part of the brain called the dorsolateral prefrontal cortex. That’s actually an FDA-approved site for stimulation with noninvasive magnetic pulses to treat depression. But patients, when they’re stimulated here . . . People have done studies. It can also change things like trust. It can change things like driving ability.

There’s only so many brain regions, but there’s millions of things we do. Of course, intervening with one region might change many things.


COWEN: What kind of students are you likely to hire that your peers would not hire?

BOYDEN: Well, I really try to get to know people at a deep level over a long period of time, and then to see how their unique background and interests might change the field for the better.

I have people in my group who are professional neurosurgeons, and then, as I mentioned, I have college dropouts, and I have people who . . . We recently published a paper where we ran the brain expansion process in reverse. So take the baby diaper polymer, add water to expand it, and then you can basically laser-print stuff inside of it, and then collapse it down, and you get a piece of nanotechnology.

The co–first author of that paper doesn’t have a scientific laboratory background. He was a professional photographer before he joined my group. But we started talking, and it turns out, if you’re a professional photographer, you know a lot of very practical chemistry. It turns out that our big demo — and why the paper got so much attention — was we made metal nanowires, and the way we did it was using a chemistry not unlike what you do in photography, which is a silver chemistry.

And this:

COWEN: Let’s say you had $10 billion or $20 billion a year, and you would control your own agency, and you were starting all over again, but current institutions stay in place. What would you do with it? How would you structure your grants? You’re in charge. You’re the board. You do it.


COWEN: If you’re designing architecture for science, what do you do? What do you change? What would you improve? Because presumably most of it is not designed for science. Maybe none of it is.

BOYDEN: I’ve been thinking about this a lot, actually, lately. There are different philosophies, like “We should have open offices so everybody can see and talk to each other.” Or “That’s wrong. You should have closed spaces so people can think and have quiet time.” What I think is actually quite interesting is this concept that maybe neither is the right approach. You might want to think about having sort of an ecosystem of environments.

My group — we’re partly over at the Media Lab, which has a lot of very open environments, and our other part of the group is in a classical sort of neuroscience laboratory with offices and small rooms where we park microscopes and stuff like that. I actually get a lot of productivity out of switching environments in a deliberate way.

There is much more of interest at the link.

Sentences to ponder

…modest genetic selection/concentration was evident for teen pregnancy and poor educational outcomes, suggesting that neighbourhood effects for these outcomes should be interpreted with care.

Note however:

Findings argue against genetic selection/concentration as an explanation for neighbourhood gradients in obesity and mental health problems.

Here is the full piece, via K.

Genes, income, and happiness

Significant differences between genetic correlations indicated that, the genetic variants associated with income are related to better mental health than those linked to educational attainment (another commonly-used marker of SEP). Finally, we were able to predict 2.5% of income differences using genetic data alone in an independent sample. These results are important for understanding the observed socioeconomic inequalities in Great Britain today.

That is from a new paper by W. David Hill,  And from Abdel Abdellouai’s summary:

Educational attainment shows a larger genetic overlap with subjective wellbeing than IQ does (rgs = .11 & .03, respectively), while income shows a larger genetic overlap with subjective wellbeing than both education or IQ (rg = .32).

All via Richard Harper.

Using Nature to Understand Nurture

An excellent new working paper uses genetic markers for educational attainment to track students through the high school math curriculum to better understand the role of nature, nurture and their interaction in math attainment. The paper begins with an earlier genome wide association study (GWAS) of 1.1 million people that found that a polygenic score could be used to (modestly) predict college completion rates. Panel (a) in the figure at right shows how college completion is five times higher in individuals with an education polygenic score (ed-PGS) in the highest quintile compared to individuals with scores in the lowest quintile; panel b shows that ed-PGS is at least as good as household income at predicting college attainment but not quite as good as knowing the educational level of the parents.

Of the million plus individuals with ed-PGS, some 3,635 came from European-heritage individuals who were entering US high school students in 1994-1995 (the Add Health sample). Harden, Domingue et al. take the ed-PGS of these individuals and match them up with data from their high school curricula and their student transcripts.

What they find is math attainment is a combination of nature and nurture. First, students with higher ed-PGS are more likely to be tracked into advanced math classes beginning in grade 9. (Higher ed-PGS scores are also associated with higher socio-economic status families and schools but these differences persist even after controlling for family and school SES or looking only at variation within schools.) Higher ed-PGS also predicts math persistence in the following years. The following diagram tracks high ed-PGS (blue) with lower ed-PGS (brown) over high school curricula/years and post high-school. Note that by grade 9 there is substantial tracking and some cross-over but mostly (it appears to me) in high-PGS students who fall off-track (note in particular the big drop off of blue students from Pre-Calculus to None in Grade 12).

Nature, however, is modified by nurture. “Students had higher returns to their genetic propensities for educational attainment in higher-status schools.” Higher ed-PGS students in lower SES schools were less likely to be tracked into higher-math classes and lower-SES students were less likely to persist in such classes.

It would be a mistake, however, to conclude that higher-SES schools are uniformly better without understanding the tradoffs. Lower SES schools have fewer high-ability students which makes it difficult to run advanced math classes. Perhaps the lesson here is that bigger schools are better, particularly bigger schools in poorer SES districts. A big school in a low SES district can still afford an advanced math curriculum.

The authors also suggest that more students could take advanced math classes. Even among the top 2% of students as measured by ed-PGS only 31% took Calculus in the high-SES schools and only 24% in the low SES schools.It’s not clear to me, however, that high-PGS necessitates high math achievement. Notice that many high-PGS students take pre-calc in Grade 11 but then no math in Grade 12 but they still go on to college and masters degrees. Lots of highly educated people are not highly-educated in math. Still it wouldn’t be a surprise if there were more math talent in the pool.

There is plenty to criticize in the paper. The measure of SES status by school (average mother’s educational attainment) leaves something to be desired. Moreover, there are indirect genetic effects, which the authors understand and discuss but don’t have the data to test. An indirect genetic effect occurs when a gene shared by parent and child has no direct effect on educational capacity (i.e. it’s not a gene for say neuronal development) but has an indirect “effect” because it is correlated with something that parent’s with that gene do to modify the environment of their children. Nevertheless, genes do have direct effects and this paper forces us to acknowledge that behavioral genetics has implications for policy.

Should every student be genotyped and tracked? On the one hand, that sounds horrible. On the other hand, it would identify more students of high ability, especially from low SES backgrounds. Genetics tells us something about a student’s potential and shouldn’t we try to maximize potential?

For homework, work out the equilibrium for inequality, rewatch the criminally underrated GATTACA and for an even more horrifying picture of the future, pay careful attention to the Mirrlees model of optimal income taxation.

U.S.A. fact of the day, *Jump-Starting America*

The United States, as of 2014, spends 160 times as much exploring space as it does exploring the oceans.

That is from the new and interesting Jump-Starting America: How Breakthrough Science Can Revive Economic Growth and the American Dream, by Jonathan Gruber and Simon Johnson, two very eminent economists.  And if you are wondering, I believe those numbers are referring to government efforts, not the private sector.  I am myself much more optimistic about the economic prospects for the oceans than for outer space.

Most of all this book is a plea for radically expanded government research and development, and a return to “big science” projects.

Overall, books on this topic tend to be cliche-ridden paperweights, but I found enough substance in this one to keep me interested.  I do, however, have two complaints.  First, the book promotes a “side tune” of a naive regionalism: “here are all the areas that could be brought back by science subsidies.”  Well, maybe, but it isn’t demonstrated that such areas could be brought back in general, as opposed to reshuffling funds and resources, and besides isn’t that a separate book topic anyway?  Second, too often the book accepts the conventional wisdom about too many topics.  Was the decline of science funding really just a matter of will?  Is it not at least possible that federal funding of science fell because the return to science fell?  Curing cancer seems to be really hard.  Furthermore, some of the underlying problems are institutional: how do we undo the bureaucratization of society so that the social returns to science can rise higher again?  Will a big government money-throwing program achieve that end?  Maybe, but the answers on that one are far from obvious.  This is too much a book of levers — money levers at that — rather than a book on complex systems.  I would prefer a real discussion of how today science has somehow become culturally weird, compared say to Mr. Spock and The Professor on Gilligan’s Island.  The grants keep on going to older and older people, and we are throwing more and more inputs at problems to get at best diminishing returns.  Help!

Still, I read the whole thing through with great interest, and it covers some of the very most important topics.

Preemptive Overfishing

The Endangered Species Act endangered some species and announcing that a fishing area will be protected in the future increases fishing now.

PNAS: Most large-scale conservation policies are anticipated or announced in advance. This risks the possibility of preemptive resource extraction before the conservation intervention goes into force. We use a high-resolution dataset of satellite-based fishing activity to show that anticipation of an impending no-take marine reserve undermines the policy by triggering an unintended race-to-fish. We study one of the world’s largest marine reserves, the Phoenix Islands Protected Area (PIPA), and find that fishers more than doubled their fishing effort once this area was earmarked for eventual protected status. The additional fishing effort resulted in an impoverished starting point for PIPA equivalent to 1.5 y of banned fishing. Extrapolating this behavior globally, we estimate that if other marine reserve announcements were to trigger similar preemptive fishing, this could temporarily increase the share of overextracted fisheries from 65% to 72%. Our findings have implications for general conservation efforts as well as the methods that scientists use to monitor and evaluate policy efficacy.

One puzzle is why there should be an increase in over-fishing? Shouldn’t a commons already be overfished to the point of zero return? One possibility is that previous steps to limit overfishing were working.

The possibility of preemptive overfishing suggests the utility of surprise protections, but that’s not always possible and the authors don’t suggest that preemptive overfishing makes protection unwise only that it has a short term cost.

Hat tip: Paul Kedrosky.

The Destruction from an Asteroid Strike

The Day the Dinosaurs Died is an amazing tale of scientific discovery. You should read the whole thing. One sub-point, however, is a vivid description of the asteroid that wiped out the dinosaurs.

The asteroid was vaporized on impact. Its substance, mingling with vaporized Earth rock, formed a fiery plume, which reached halfway to the moon before collapsing in a pillar of incandescent dust. Computer models suggest that the atmosphere within fifteen hundred miles of ground zero became red hot from the debris storm, triggering gigantic forest fires. As the Earth rotated, the airborne material converged at the opposite side of the planet, where it fell and set fire to the entire Indian subcontinent. Measurements of the layer of ash and soot that eventually coated the Earth indicate that fires consumed about seventy per cent of the world’s forests. Meanwhile, giant tsunamis resulting from the impact churned across the Gulf of Mexico, tearing up coastlines, sometimes peeling up hundreds of feet of rock, pushing debris inland and then sucking it back out into deep water, leaving jumbled deposits that oilmen sometimes encounter in the course of deep-sea drilling.

…The dust and soot from the impact and the conflagrations prevented all sunlight from reaching the planet’s surface for months. Photosynthesis all but stopped, killing most of the plant life, extinguishing the phytoplankton in the oceans, and causing the amount of oxygen in the atmosphere to plummet. After the fires died down, Earth plunged into a period of cold, perhaps even a deep freeze. Earth’s two essential food chains, in the sea and on land, collapsed. About seventy-five per cent of all species went extinct. More than 99.9999 per cent of all living organisms on Earth died, and the carbon cycle came to a halt.

…One of the authors of the 1991 paper, David Kring, was so frightened by what he learned of the impact’s destructive nature that he became a leading voice in calling for a system to identify and neutralize threatening asteroids. “There’s no uncertainty to this statement: the Earth will be hit by a Chicxulub-size asteroid again, unless we deflect it,” he told me. “Even a three-hundred-metre rock would end world agriculture.”

When the asteroid hit it unleashed the energy of a billion Hiroshimas, that’s one reason I support foundations like the B612 Foundation who are working to map asteroids and develop systems to protect our world. As Tyler and I point out in textbook, protection from asteroids is a true public good which is one reason why we aren’t spending enough on this project.

Hat tip: Kevin Lewis.