This is officially an award-winning blog
HNN, Best group blog: "Witty and insightful, the Edge of the American West puts the group in group blog, with frequent contributions from an irreverent band.... Always entertaining, often enlightening, the blog features snazzy visuals—graphs, photos, videos—and zippy writing...."
53 comments
February 18, 2009 at 10:30 am
James Davies
While you may not think Nature, ostensibly a science journal, is your natural habitat, think about the similarities of the methods and practices of historians and many scientists who regularly publish in that journal.
For instance, astronomers don’t sit in a lab. We read the light from distant objects in the heavens, and that light tells us about the history of our universe. Paleontologists reads fossils to understand the narrative of evolution of life on earth. Archaeologists and anthropologists do much the same with artifacts. These are historical sciences through and through.
So science is not so different from reading primary sources from the 1930s and putting together a narrative on the Great Depression.
Nice article BTW. I didn’t realize the WPA was that involved in basic research. I probably should just read your book.
February 18, 2009 at 10:32 am
dana
Nice angle, eric.
February 18, 2009 at 10:48 am
Josh
When did your hair go completely white?
February 18, 2009 at 10:52 am
Ahistoricality
Nicely done, yeah. Though publishing that in a science journal may be more preaching to the converted than anything else.
February 18, 2009 at 11:18 am
Bitchphd
Omg that illustrator made you look like Buckley. I’m so sorry.
February 18, 2009 at 11:39 am
Buster
I had the same thought as BithPhD, but I didn’t have the guts to say it. I was just quietly hoping that nobody else noticed.
Eric, there now seems to be a plot to make you look evil!
February 18, 2009 at 11:42 am
eric
If you click through to the other pieces in the feature, you’ll see everyone has white hair. Just the artist’s choice, I guess.
And James Davies, I wholeheartedly agree with your summary of why many if not most sciences are in fact historical. But many scientists don’t like the notion, I find.
February 18, 2009 at 12:30 pm
kid bitzer
well, they don’t like it, because they’re embarrassed.
i mean, i remember complaining to an astronomer once about the old news he was peddling. i mean, they pretend that they are telling us new stuff, when it turns out to be billions of years out of date. i don’t care what the damn star did 6 billion years ago–i want to know what it’s doing now!
and instead of a real answer, or even a decent apology, he gave me some cock and bull story about “speed of light” blah blah “relativistic event horizon” hoo hah and a raft of other lame evasions. the fact is, their findings are completely antiquated–it’s deader than a dead letter posted in a shroud.
and meanwhile, they pretend to be on the “cutting edge”. hah. charlatans, all of them.
February 18, 2009 at 1:06 pm
LizardBreath
I had the same shallow reaction to the illustration as everyone else — not a flattering (or, for that matter, particularly recognizable) portrait.
Nice article, though. I had no idea commercially available penicillin was could be credited to the New Deal.
February 18, 2009 at 2:42 pm
Colin
Bring back the swimsuit.
February 18, 2009 at 2:57 pm
LizardBreath
But on no account allow the artist responsible for that illustration to paint you in the swimsuit. Guaranteed to turn out badly.
February 18, 2009 at 3:47 pm
Brad
So science is not so different from reading primary sources from the 1930s and putting together a narrative on the Great Depression
Except, unlike in history, one can go out and get the equivalent of more primary sources if one needs to answer a specific question.
Not to mention the narratives are based on a scientific measurements done in laboratories. If one is interested in the lifecycle of stars, many of the measurements that control that lifecycle are made in labs. The arguments over the origin and solution to the Great Depression rest on macroeconomics, a field where the practitioners seem to have a some disagreement over some pretty basic relations.
February 18, 2009 at 5:03 pm
beamish
You’re now eligible for the worship me science badge.
February 18, 2009 at 5:12 pm
Cosma
I assumed the drawing was of some New Deal worthy.
February 18, 2009 at 5:22 pm
Cosma
Except, unlike in history, one can go out and get the equivalent of more primary sources if one needs to answer a specific question.
Pity the poor supernova astronomers, doomed to lack scientific status. (“From the beginning of time, astrophysicists have yearned to blow up suns.”)
February 18, 2009 at 5:58 pm
Sifu Tweety
”From the beginning of time, astrophysicists have yearned to blow up suns.”
I knew it.
February 18, 2009 at 6:35 pm
Brad
Pity the poor supernova astronomers, doomed to lack scientific status. (”From the beginning of time, astrophysicists have yearned to blow up suns.”)
Hunh?
If one wants more supernova, go take more data. The rate of supernova discovery goes up every year.
February 18, 2009 at 6:37 pm
Brad
I knew it.
Those poor people, they are forced to write enormously complicated computer programs instead.
February 18, 2009 at 6:51 pm
Sifu Tweety
The historian: first Nature, then Stanford. Rousseau would have seen this.
February 18, 2009 at 6:57 pm
eric
I can haz Stanford professorship?
February 18, 2009 at 7:25 pm
dana
I think if they don’t print a correction, you are permitted to challenge Stanford’s chair in a duel, and if you win, you can teach at Stanford.
February 18, 2009 at 7:55 pm
Cosma
Those poor people, they are forced to write enormously complicated computer programs instead.
Yes, I used to help debug some of those programs. (I had friends doing degrees in nuclear astrophysics while I was doing stat. mech.) Simulations are helpful, especially for seeing if your story really hangs together, but they’re not data.
February 18, 2009 at 8:53 pm
essear
Yes, I used to help debug some of those programs. (I had friends doing degrees in nuclear astrophysics while I was doing stat. mech.)
You helped debug your friends’ code? You, sir, are a saint. (Want to take a look at some problematic Fortran in your spare time?)
February 18, 2009 at 8:56 pm
Brad
Cosma, I am going to just disagree with you. Now, if you write a yet another interesting discourse on your weblog about scientific reasoning and model testing, I maybe forced rethink my position.
February 18, 2009 at 8:56 pm
Brad
(Want to take a look at some problematic Fortran in your spare time?)
Well, I can think of at least one problem straight off….
February 18, 2009 at 9:07 pm
Brad
I forgot to add
Simulations are helpful, especially for seeing if your story really hangs together, but they’re not data.
Yea, which is why people work so hard to findlots of new ones.
February 18, 2009 at 9:21 pm
Josh
if you win, you
canhave to teach at StanfordFebruary 18, 2009 at 9:23 pm
Sifu Tweety
Well, I can think of at least one problem straight off….
All FORTRAN is problematic?
February 18, 2009 at 9:39 pm
Sifu Tweety
Cosma’s link is interesting, but I wonder if it doesn’t undersell how much goodness you can get out of probabilistic inference, especially if you’re talking about sampling from arbitrarily large models?
February 18, 2009 at 9:54 pm
Michael Turner
Except, unlike in history, one can go out and get the equivalent of more primary sources if one needs to answer a specific question.
Historians can’t turn up new primary sources?
February 18, 2009 at 10:44 pm
teofilo
Speaking of astronomical data collection…
February 18, 2009 at 11:04 pm
Brad
Historians can’t turn up new primary sources?
Let us see if I can make myself clear. This is always a challenge.
To continue with the supernova analogy:
A supernova is when a star blows up. Astrophysicists cannot make a star blow in the lab, but they can study some parts of the process in the lab. As models and laboratory data become better, astrophysicists can go over the old data, and, at the same time, observe new supernova (there are, like, 100+ found a year). The cool thing about supernova is that the remnants hang around for a while, so you can keep poking your models with new results (the supernova 1987A keeps teaching us new things) but all because you are getting new data from new technologies from both new events and old events.
To drag this back on topic, there has been only one Great Depression. If you want, you can lump in some really bad panics in the late 1800’s. If you want to understand the causes of the Great Depression(s), you can look back when it happened and see what you can find. You might find some new viewpoints, but most of the time you have access to most of the major sources of what was written about what happened and when, especially the data about spending, borrowing, unemployment, etc. What you cannot do is wait for a 100+ new Great Depressions to come along and test your ideas. I mean, if you are really lucky, you might get another one to happen in your lifetime…..
Hopefully this makes some sense.
February 18, 2009 at 11:06 pm
Brad
All FORTRAN is problematic?
I guess I need to resort to emoticons.
February 18, 2009 at 11:08 pm
Brad
Speaking of astronomical data collection…
Purty….
February 18, 2009 at 11:15 pm
teofilo
What you cannot do is wait for a 100+ new Great Depressions to come along and test your ideas. I mean, if you are really lucky, you might get another one to happen in your lifetime…
This seems like a quantitative rather than a qualitative difference.
February 19, 2009 at 4:23 am
Michael Turner
There it is, the resort to “quantity has a quality all its own” (or its modern variant, “size does matter.”) It’s a great time-saver in handwaving arguments. Especially when you’re arguing with people who you’ve come to believe are not very bright, who need to be talked down to. You should always make sure you’re right in that assumption, however.
Brad leaves out a couple of significant points. Most of them relate to the following (rather obvious, I would have thought) concept in scientific discovery: if you have a good model of the mechanism, you don’t need a wealth of data about the results of many complete cycles of the mechanism as long as you can get a wealth of data from each cycle of the mechanism.
One would not, for example, argue that evolution through natural selection was a poorly supported theory of how our planet came to be populated with so many species, because, hey, it’s only one data point: planet Earth, and how can you infer anything from just one planet!
Brad reports that we keep discovering new supernovae at the rate of 100+ per year. Half-true. While supernovae remnants can have resolvable scale, the actual detonation is observable only very rarely, and only as a point source at the time.
I’ve attended very interesting presentations at conferences devoted to combustion and detonation where astrophysicists claimed to have modeled what must have happened inside certain supernovae. They even used techniques developed for modeling for chemical detonations, revealing cellular structures similar to those of ordinary explosions. However, unlike with chemical detonations (or even nuclear and thermonuclear explosives here on earth) astronomers mainly observe after-effects of supernovae, not the creation processes, when they want to compare their models with reality. It was interesting to see these modeling results, not just because they seemed so convincing, but because that convincing case didn’t come along until about the beginning of the 21st century, long after supernovae had been discovered.
In contrast, econometricians and economists go back to the historical record all the time, looking for data in primary sources, data that in some cases never figured into any economists’ calculations before, because its relevance wasn’t hitherto suspected. They can “climb inside the economic supernovae” (if the records exist, and often a fair enough sampling does exist). If, for example, an economic model turned out to have overrelied on counting only warehoused goods as inventory where there’s reason to suspect that goods in transit could, for the purposes of accuracy, count as “inventory”, it’s possible to go back to primary sources that might never have been consulted before on that issue, such as shipping records.
You can prove out aspects of supernova models in the lab? Economics has been a laboratory science for almost 60 years now. At least one economist (Vernon Smith) has gotten a Nobel Prize largely for his work in experimental economics.
Now, if you want to talk about how certain points in macroeconomics only seem highly debatable because of certain political or career agendas (Greg Mankiw, very likely), or perhaps because certain major economists are talking outside their areas of specialization (lookin’ at you, Eugene Fama — a renowned financial economist, certainly, but does he have any macro chops to speak of? I don’t think so) . . . well, then Brad might be able to get somewhere here. But it would help if he actually knew what he was talking about in the first place, and in this case I don’t see much evidence that he knows how economics is actually practiced.
Would anyone say Galileo wasn’t really doing science because he happened to be operating in a politicized arena? No, you’d say he was trying to do science in an area where politics made it hard. Econ is highly politicized. That doesn’t mean there aren’t some results in economics about as solid as the heliocentric theory.
February 19, 2009 at 5:05 am
Michael Turner
To drag this back on topic, there has been only one Great Depression. If you want, you can lump in some really bad panics in the late 1800’s.
“Really bad panic” makes it sound like some transient event. Some of these events were far from transient. Paul Krugman has noted the Long Depression (starting from the Panic of 1873) as resembling the present day more than the Great Depression in its root causes, and also to point out how long a depression can go on if not properly treated (6 years or more).
And he was hardly the first to notice the resemblance.
And this is just America we’ve talked about so far. When you start looking at other countries . . . well, we think of Pinochet as Chicago School, straight up, but in the early 1980s he rapidly converted to very Keynesian remedies when they became necessary. There are a lot more supernovae out there than you seem to think, Brad. Get a telescope.
February 19, 2009 at 5:33 am
Cosma
Thing 1: my friends helped me debug my code too. And I can’t recommend too highly the trying-to-explain-the-problem-to-someone-else method of debugging. It often makes it obvious why your approach could never have worked in the first place.
Thing 2: Most of what Michael Turner said @ 4:23.
February 19, 2009 at 8:09 am
Jonathan Dresner
And I can’t recommend too highly the trying-to-explain-the-problem-to-someone-else method of debugging.
My father called that “talking to a wall”: He’d use my mother — smart lady, but not a programmer — or, once in a while, his children. Really did work better than you’d think.
I have … memories of FORTRAN: back in the late ’80s I helped program some scientific modelling software, and the mathematical modules could only be linked from FORTRAN. At the time I was mostly fluent in Pascal, but my boss had an IBM language package that allowed crosslinked modules between Pascal and FORTRAN (and C, but I didn’t do C), so I did most of my organizational work in Pascal, and the computational work in FORTRAN. Mostly I remember the problem I had with positioning lines, and the “fun” of figuring out how to define variables and pointers that would pass cleanly back and forth.
I liked COBOL better, because you could use cool variable names.
February 19, 2009 at 9:00 am
Sifu Tweety
And I can’t recommend too highly the trying-to-explain-the-problem-to-someone-else method of debugging.
Now codified, in somewhat different form, as pair programming.
February 19, 2009 at 10:13 am
Brad
There it is, the resort to “quantity has a quality all its own” (or its modern variant, “size does matter.”) It’s a great time-saver in handwaving arguments. Especially when you’re arguing with people who you’ve come to believe are not very bright, who need to be talked down to. You should always make sure you’re right in that assumption, however.
Excuse me, please state where I made some statements about who is “bright” and who is not? Or where I made some statements about the relative value of different inquiries? I think you are make some assumptions that are not warranted.
February 19, 2009 at 10:30 am
Brad
This seems like a quantitative rather than a qualitative difference.
Yea, and I would almost say that is all it is. Multiple independent measurements, large numbers of data (though of often of suspect quality) and a foundation that is built on experiments.
I wanted to be historian, but I decided that history was too hard. The ability to make new insights by looking at the same events seems beyond me. Instead, I let nature do the hard part of making lots of independent examples for me.
February 19, 2009 at 11:27 am
John Emerson
Unfortunately, Brad, economics is an experimental science. That’s the very problem:
February 19, 2009 at 3:05 pm
James Davies
Except, unlike in history, one can go out and get the equivalent of more primary sources if one needs to answer a specific question.
Speaking as an astronomer, I would say this is certainly the case in some respects where datasets are large. So people studying galaxy evolution have lots of galaxies with which to test their theories. In other areas, not so much. For instance, right now I’m studying planetary nebulae in globular clusters, which theoretically shouldn’t exist (but do in very low numbers), and finding the data to flesh out our understanding of the mechanism and theory is very, very difficult. I would say it’s similar to fossil hunting paleontologists looking for transitional forms over the course of the last 150 years. Needle in a haystack stuff.
So it’s not so much a qualitative difference, only quantitative. We are relying on a historical event in the past (a star exploding or a now-extinct animal getting fossilized) and we need to detect or find that historical event. We can’t reproduce it in the lab, like say the folks at Fermilab and LHC can when searching for the Higgs boson, another needle-in-a-haystack sort of search.
Economists and historians (the good ones at least) I think do something very similar. They may not have as robust theoretical models (though some economics models can be quite robust) or as much historical data with which to test those models, but the methods used to reach consensus in those fields of research relies on a similar sort of consensus building amongst researchers as it does in the historical sciences.
February 19, 2009 at 6:55 pm
Cosma
Emerson: OK, I’ll bite, how much of that blockquote is authentic?
February 19, 2009 at 8:02 pm
John Emerson
Not a word, except the Henry Ford story, which may well be apocryphal, but is old.
February 19, 2009 at 11:36 pm
teofilo
What Henry Ford story?
February 20, 2009 at 2:06 am
Michael Turner
The one where he said “History is bunk”, maybe?
February 20, 2009 at 2:42 am
Michael Turner
. . . the methods used to reach consensus in those fields of research relies on a similar sort of consensus building amongst researchers as it does in the historical sciences.
Naomi Oreskes is interesting on how the notion of scientific consensus itself has evolved. She claims that you could take all of what geologists now know about plate tectonics back to the 19th century and geologists then might accept all the theory and data as valid but still wouldn’t adopt plate tectonics as their consensus view — because they were uncomfortable with the very idea of a consensus view. Pluralism in theory was considered a good thing — socially hygienic, for geologists, at least. (Or so she claims.)
Geologists can while away afternoons in courtly discussion while the continents drift. With economics, you don’t always have the luxury of collecting data, honing theory and crafting policy at a leisurely rate. Obama might have put it best: you can’t let the perfect be the enemy of the absolutely essential. I don’t know of any reasonable economist who says that spending stimulus is absolutely certain to be better, in every way, than tax-relief-only stimulus, or than doing nothing at all. What seems to be certain, however, is that spending stimulus is the best thing to try, and is quite likely to work, and even Milton Friedman had good things to say about the New Deal and plenty of bad things to say about the “Treasury View”. I don’t see Greg Mankiw offering objections much more substantial than “we-don’t-know-for-sure”, and Mankiw is about the best you can do among the stallers — it’s pretty much all downhill from him to Larry Kudlow. You could call such spending-stimulus policy moves “experimental economics” if you want, but there’s already a well-established practice of laboratory economics. I’d call it courageous and informed policy action, with possibilities for interesting scientific measurement thrown in only as a bonus.
February 20, 2009 at 5:12 am
John Emerson
Thomas Edison, I mean. Those guys all look alike to me.
February 20, 2009 at 11:36 am
John Emerson
Back on-topic, I’m not objecting to the stimulus experiment but to the Greenspan experiment.
February 20, 2009 at 11:40 am
mealworm
I’m an astronomer. So, Brad, what about cosmology? Got any more Big Bangs in your pocket?
February 20, 2009 at 11:55 am
mealworm
Cool photos, Teo. The coolest name for a telescope around: Very Large Array. Because that’s what it is!