There will ever be more Norman Borlaugs

What’s that I hear?  Is it scientists humiliating the sad crisis-mongering neo-Malthusians?  Must be a weekday, then.

• Turning grass and trees into food: Neo-Malthusians mocked by human ingenuity again (Reason Hit & Run)

Good show!

Self-reporting shenanigans

The careful reader will know that I’m no fan of epidemiological studies.  I expect that they have their uses in some arenas, but the context in which I tend to see them is personal health — and they’re generally worse than useless.  But even worse than a run-of-the-mill epidemiological study is one based entirely upon self-reported data.

“Hi, I’m calling from the Center for Disease Control and Prevention.  I’d like to ask you some questions about your loathsome, immoral, un-American fatassedness.  To begin, what is your current height and weight?”

Does that sound like a reliable method for obtaining high-quality data to you?

Hence, I find this report utterly hilarious:

• People in the South are not so fat after all — and they lie less (Mike Oliver)

Looking at the numbers shows the wide discrepancy between what people say on the telephone and the physical evidence of actually getting weighed. When weighed in the REGARDS study, all of the regions’ obesity’s numbers went up — it’s just that the southern region numbers went up less.

“Everybody underreports their weight but women do it more,” Howard said.

Men, on the other hand, do something else that affects the Body Mass Index, which is weight divided by height squared and is used to define obesity.

“They overreport their height, which makes them seem less obese.”

It’s difficult to tell from the abstract whether the authors somehow corrected for the BMI’s manifold and obvious deficiencies, but it hardly matters.  The real punchline here is that many published obesity researchers — and the policy wonks who craft legislation from their abstracts and maybe a few of the more colourful figures — are astonishingly naive.

Anti-Wikipedia bias

David Henderson complains:

When I first heard about Wikipedia, I thought, “this can’t work.” My reason: there was no assurance that letting huge numbers of people fill in entries and update things would lead to correct information. That said, it works much better than I had expected.

But in my only two cases where I have paid close attention to how information gets screened, Wikipedia has worked badly.

It’s a fair point.  My forays into Wikipedia have generally been on the hard-science side, where I’m either reading Wikipedia and Mathworld in parallel to try to teach myself something mathy or looking for an interested dilettante’s introduction to some endocrinological system or other.  I haven’t noticed any errors in the math stuff, but I have noticed a strong status-quo bias in the physiology stuff — particularly as one gets closer to lipid transport and metabolism.

However, we have to be careful to avoid slamming Wikipedia for failing to adhere to an implicit best-case standard when deciding whether or not to use it.  It’s fine (and helpful!) to point out its flaws when discussing how to fix them, but when picking a first point of reference the question isn’t “is Wikipedia any good?” but “is Wikipedia any better than the alternative?”  If I want a quick overview of AMPK-alpha2 in order to get some context for interpreting a SuppVersity post, is Wikipedia’s status-quo bias really going to be all that big an issue?  Would I really be any better off trekking an hour to the nearest research library and digging up a textbook on biochemistry which is probably less peer-reviewed than that Wikipedia article?  Judged by the same standards people — usually people in gatekeeper-of-knowledge roles, like teachers, profs, and other credentialed experts — tend to apply to Wikipedia, most textbooks are shit.

What’s the BATWA — Best Alternative To a Wikipedia Article — for any given topic?  Since those alternatives naturally include “whatever pops up first in a Google search” and “staying ignorant/common wisdom”, I submit that checking Wikipedia first is a much better strategy than it’s trendy to admit.

“I don’t understand it; it must be magic”

By way of Andrew Sullivan we discover that the folks at Slate have been reading popularized science again.  Yeah, go ahead and start facepalming right now.

• Will GATTACA come true?

Just from the title.

Noting that fetal DNA sequencing is becoming simultaneously more popular and less expensive — funny how that works; capitalism strikes again! — our Intrepid Reporter Maria Hvistendahl does Underpants Gnomes genetics:

What fetal genes might one day suggest about a baby’s eye color, appearance, and intellectual ability will be useful to parents, not insurers. But with costs coming down and insurers interested in other aspects of the fetal genome, a Gattaca–like two-tiered society, in which parents with good access to health care produce flawless, carefully selected offspring and the rest of us spawn naturals, seems increasingly plausible.

SWEET SHIVERING FUCK, YOU IGNORANT PRICKDRIZZLE, THAT IS NOT HOW GENES WORK.

The process Hvistendahl has in mind looks like this:

1. Sequence a baby’s DNA into genomic sequences that look like this: […] GAC ACC GTC ATT TTA CTA CTT […]
2. ???
3. Gattaca!  Look it has nucleotide initials in it!  Sciencey!

As you might expect, step 2. is doing all the heavy lifting.

If I’m lucky, LabRat will pick up this story and have the time to fisk it so hard NEST will be sifting through her dogs’ poop for the next decade and a half.  But even so, let’s simplify the problem outrageously and turn it into a programming analogy.

Suppose someone gives you a shiny disc, heavy, about 3″ in diameter.  You can carefully install it in a box you barely understand and it’ll make the box do magical, wonderful, insanely complex stuff like display porn from the other side of the world.  After decades of research you might come up with the hypothesis that instructions for making the box do stuff are somehow encoded on the disc.  You spend another few decades carefully examining the disc, and after poking around with an electron microscope you discover regular patterns on the disc, which might be translated into a base-two numeric system like so:

[…] 10110100 00101001 11000100 11010101 0000000 […]

You announce your discovery to the world, and people like Hvistendahl start wailing: “OMG ONOZ now we can control the magical wonderful box, all the magic and wonder is gone, soon we’ll be telling the box it can only display straight porn, O the horror!”  But really you’ve just found a sequence of numbers, and guessed (not entirely unreasonably) that they mean something.

Fast forward through a few more decades of heroic effort.  You’ve discovered that those ones and zeros correspond, more or less, to eight-bit segments, which combine to form integers, or offsets along something you expect might be a tape, or instructions in a programming language of some sort, or some absolutely horrifying combinations which eventually turned out to be floating-point numbers, and were met at every step of the way with OMG ONOT IT IZ TEH GATTACA.  Finally you’re able to more or less build a model of the programming language, and you discover to your resigned exasperation that most of the programs are in fact used to generate statements in other programming languages, which appear to be vastly more complex than those you’ve discovered to date:

infixl 1  >>, >>=
class  Monad m  where
(>>=)            :: m a -> (a -> m b) -> m b
(>>)             :: m a -> m b -> m b
return           :: a -> m a
fail             :: String -> m a

m >> k           =  m >>= \_ -> k

Distressingly, there seem to be millions — at the least — of these programs floating around the magic wonder box at any given point, all of which interact with each other in various forbiddingly complex ways, and any of which may (or may not) be vitally important at any given time.

And the jackasses at Slate are still going on about how you’re destroying the mystery and gravitas of the magic wonder box, as if it’s as easy to command as an LED on a breadboard!

Without even getting into the issues surrounding abusus non tollit usum, the flagrant ignorance behind the notion that “if we can sequence a baby’s DNA, we can engineer super-babies!” is sufficient to peg my stupid-meter.

Malthus is ever with us

Here’s Lester Brown, founder of the Earth Policy Institute and obviously an impartial observer with no incentive to stoke ecological panic:

• Food, not war, is the biggest threat to world security (Scientific American)

Grain yields are beginning to hit a “glass ceiling” in many countries, Brown said, where farmers have already taken advantage of what science has to offer for improving yield. As more and more countries hit an upper limit on productivity, the world grain harvest will begin to plateau, even as demand for food continues to rise, causing a rise in prices.

That’s right: We’re all tapped out for scientific discovery in agriculture.  Science has nothing more to offer farmers who want to improve their crop yields; there will be no more Norman Borlaugs.  Brown has a Master’s degree — in Public Administration, but focus on the “Master’s degree” part — so you know he knows whereof he speaks.  He even appropriates the term “glass ceiling” for what we math nerds would call a ceiling (or, if we’re feeling especially math-nerdy, an asymptote) because, you know, Social Justice and stuff.

Oh yeah, and grain is apparently all that matters when it comes to feeding people.  Consumers couldn’t possibly substitute into other staple foods; that would imply change in response to incentives, which is too complex a thought for even someone with a Master’s in Public Administration to hold in his knowledge-swollen head.

Someone had better tell these bioscientists that they were focusing on the wrong plant.  Or not; after all, we’ve been informed that Science is all out of ideas.  By a very smart bureaucrat.  Maybe they should seek Master’s degrees in Public Administration instead.

Special message to Lester Brown:

It’s high in protein!

All linky, no thinky

How about some good news for a change?

• Suu Kyi hopes victory is dawn of new era (Time)

Aung San Suu Kyi claimed victory Monday in Burma’s historic by-election, saying she hoped it would mark the beginning of a new era for the long-repressed country.

Suu Kyi spoke to thousands of cheering supporters who gathered outside her opposition party headquarters a day after her party declared she had won a parliamentary seat in the closely watched vote.

I think this is pretty clearly a step in the right direction.

——

Meanwhile, in the Anglosphere:

• FDA declines to ban BPA (Reason Hit & Run)

BPA has crossed this blog before (and also here), and while it’s known to be something you probably want to avoid — especially in the sorts of doses commonly given to lab rats — the actual level of risk of chronic BPA exposure in the Standard American Diet remains somewhat obscure.  Here’s what the FDA discovered from their own work:

The NCTR researchers have been conducting in-depth studies of BPA since September 2008, when a report by the NIEHS and NTP called for more research into the potential toxic effects of BPA on fetuses, infants and children.

NCTR’s findings include:

• The level of BPA from food that could be passed from pregnant mothers to the fetus is so low that it could not be measured. Researchers fed pregnant rodents 100 to 1,000 times more BPA than people are exposed to through food, and could not detect the active form of BPA in the fetus eight hours after the mother’s exposure.
• Exposure to BPA in human infants is from 84 to 92 percent less than previously estimated.

NCTR researchers report that they were able to build mathematical models of what happens to BPA once it’s in the human body. These models showed that BPA is rapidly metabolized and eliminated through feces and urine. They found that BPA is “exactly the opposite” from some other toxins, like dioxin, that can stay in the body’s tissues for months or even years.

The center’s toxicology research has not found evidence of BPA toxicity at low doses in rodent studies, including doses that are still above human exposure levels.

Bearing in mind that mathematical models of complex phenomena are often dangerously skewed — see also Black-Scholes — and that the FDA is in a perfect position to be captured by corporate interests, we should take these data with a grain or three of salt.  I certainly wouldn’t want to bet my endocrine system on a mathematical model of BPA metabolism, although the rodent studies increase confidence.  What we need here is a solid hypothesis for the mechanism of BPA metabolism and some targeted studies to back it up.

On the other hand, the typical BPA tox study — or at least its reporting — has devolved into hysterical fearmongering, and the molecule’s obesogenic effects are adequately explained by other phenomena.  And since (on the gripping hand) most research in the life sciences is utterly unreproducible, we should be somewhat skeptical of both results.

Given these levels of uncertainty, I’m cautiously happy about the FDA’s decision to do nothing for the moment.

——

Speaking of science, apparently some social scientist (yes, I typed that with a straight face) did a survey and found out that trust in scientists has been declining among Republicans over the past thirty-odd years.  This has naturally led to a lot of electrons being twitted about to make text appear on the Big Truck, and Andrew Sullivan has collected some responses (and a graph!):

• The evolution of the anti-science party

The neat thing about these data is that they don’t quite fit the narrative — Republicans started off trusting “science” just as much as Democrats, and have since declined to levels of (mis)trust shared by moderates since the study’s time series began. Furthermore, the decline in trust has been driven by the best-educated conservatives, not the easily-stereotyped toothless hicks.  This leads Kevin Drum to write hilarious things (citing Hollywood in the same breath as academia as an “information-disseminating institution”, for example).

Having made this particular brand of sausage, I can see both sides of the story.  Research institutions are fundamentally human institutions, and are just as vulnerable to the usual human foibles and weaknesses as any other.  On the other hand, when stretched out over a sufficient number of mutually competitive participants, the scientific method makes it really hard to ignore your mistakes, which makes it a pretty great tool for explaining and predicting observable phenomena.

Seems to me that one of the root problems here is that people generally expect science to be the sort of thing that you’d find in a Saturday-morning cartoon or an episode of CSI — lab coats, whirring machinery with blinking lights, and fantastic inventions or certain results by the end of the episode.  Only people who get a few years into a STEM major are ever forced to discard that mental model.

——

Speaking of mental models, Frances Woolley does an outstanding job of making

• The case for taxing basic groceries (Worthwhile Canadian Initiative)

More uniform consumption taxes are not only more efficient, but when properly implemented can be more progressive as well.  Click through and RTWT; you’ll be smarter when you get out the other side.

Epic comment thread watch

Derek Lowe has a post up which ought to pique your interest real quick:

• When reagents attack! (In The Pipeline)

He solicits exciting reaction-gone-bad stories from his commentariat, with this as an example:

My most vivid reagent-gone-bad story is probably this one; that’s a time I literally came down counting fingers. What other things have you had turn on you?

32 comments so far, with some real pucker-inducing stories sprinkled liberally therein.  Go have a look.

Science: Usually not spectacular

The thing about science is that it only gives you reliable data about what you actually did in your experiment, rather than what you wanted to do.  Most of the time our reach exceeds our grasp, so while we’d like to be able to detect faster-than-light neutrinos what we actually do is measure the time it takes for a neutrino beam to go from point A to point B in what we hope is a fairly constant relativistic frame of reference.  “Holy shit, this is amazing!” results are often attributable to unexpected correlations in the experimental setup rather than holy-shit-amazing new properties of the Standard Model.

At least, that’s how physics works, when we’re doing it right.

Dr. Andro finds a new study purporting to show that:

[A] high-protein diet with a group-based dietician-led support and education programme easily deliverable in a real-world setting does not promote greater weight loss than the prescription of a high-carbohydrate diet

We can expect this to be trumpeted to the heavens by the hearthealthywholegrains evangelists, of course.  But, y’see, there’s a problem — and if your eyebrows rose to the heavens at the word “dietician” in that excerpt you’re not far wrong.  See, the “diet” under consideration asked participants to cut 500 kcal/day from their intake, completely unsupervised except for an initial meeting with that dietician and what Dr. Andro scare-quotes as “culturally-appropriate recipes”.  The instrumental variable the researchers thought they were studying was macronutrient composition, but a look at the data (click through to Dr. Andro’s writeup) shows that what they were actually studying was variance in instruction to the dieters.

“High-carb” dieters reduced their daily intake by about 250kcal, largely by cutting fat.  “High-protein” dieters reduced their daily intake by about 150kcal, largely by, er… that’s a good question.  They don’t seem to have eaten any fewer carbs than the “high-carb” dieters, so while this study is likely to be used as a bludgeon against low-carb advocates it is in fact nothing of the sort.  Furthermore, adherence to the “high-protein” dictum fell off drastically over the study’s two year span, which you’d have thought a qualified dietician — maybe even one certified in North Carolina — could have prevented or at least mitigated.

Note that as usual the researchers evaluated the success of the diet based on weight loss.  Was that lost weight adipose tissue?  Skeletal muscle?  Brain matter?  Meh, who cares; we got the result we were looking for, time to publish!

More and more I’m coming to suspect that the “secret” to fat loss is being enough of a nerd to (a) form a plan based on a hypothesis (this could be “fat will make you fat” or “insulin/leptin control is critical”), (b) implement that plan for long enough to see results, (c) evaluate whether the plan worked, (d) re-evaluate the hypothesis based on the results, and (e) iterate.

Next we have an admirable dissection by Paul Ingraham of a study that purports to show that post-exercise massage promotes mitochondrial growth:

• Massage reduces inflammation and promotes mitochondria? (SaveYourself.ca)

The problem isn’t so much the research as the breathless conclusions being drawn therefrom.  The post is difficult to excerpt, but here’s a hint:

These conclusions reach a fair ways past what the evidence can support. It’s a red flag that “clinically beneficial” is in abstracts, headlines, press releases: the physiological effects are too complex to assume that they are all therapeutic, or even benign. The research is interesting and worthwhile, but it badly needs a critical perspective that I haven’t seen anywhere else yet.

Click through and RTWT.

The moral of these stories is that science is more often a slow, arduous process than it is a spectacular one.