Living through a moment in history, and why social science really is different from the physical sciences
This post is prompted by a survey I just took, sent by one of my colleagues from grad school. The survey covered a lot of ground, but the question that prompts this post was about how cognizant political scientists are of the potential for effects to vary over time.
Gee... why might that question have any resonance for me, or anyone else, right now?
In fact, I had been pondering that one before COVID the Birdie started circling our collective heads, as several posts here indicated.
In light of that, here's a lecture I've been giving students on the first day of Introductory classes, year after year, since I started this whole gig. It goes a little something like this. I start by asking students how the dinosaurs died. Inevitably, there are a few kids who went through a dinosaur phase, who know immediately about the meteor. I then tell them that that wasn't what I read when I was a kid. And I ain't that old. When I was a kid, that was idle speculation. Common hypotheses were climate change and disease. What happened? They found the iridium layer at the 65 million year stratum, iridium being very rare on Earth, but very common in meteors, and satellite imagery found the crater, much of which was under water, off the Yucatan, but it was so big that you had to have aerial imagery to see it. Point? Textbooks get stuff wrong, and the sciences-- the physical sciences-- are about advancement, not a list of facts.
The next phase of my mini lecture goes as follows: "science" is a method, not a subject. It is the process of testing falsifiable hypotheses through a combination of observation and experimentation. That process can be applied to questions about chemical compounds, with an emphasis on experimental approaches, geologic and astronomical processes through observation (have fun trying to experiment with plate tectonics or creating black holes), and it can be applied to questions of politics. Science is not only the best method for answering questions about how the world works, it is the only method specifically designed to converge towards the right answer. Note my wording. "Converge." A scientific study will not always give you the right answer. Science is a process, and it is a collective process between a bunch of people engaged in scientific studies, interacting with each other. Scientists replicate each others' work, build on each others' work, and through that interaction, the scientific process is the process of improvement. That's the short version. Give me more time and I'll give you a lecture on Thomas Kuhn, paradigms, paradigm-shifts, and other stuff, but that's basically how I introduce the concept of science.
How I introduce Political Science.
And I push back pretty hard on the phrase, "hard science." Note my phrasing, "physical science." What makes "hard science?" Math? I do math. Plenty of it. Statistics, game theory, social choice theory... Try reading Kenneth Arrow's Social Choice & Individual Values sometime. Try reading some Richard McKelvey. I've had some badass math students in my office try to work their way through the chaos theorem. Not chaos theory, but theorem. You want to delve into the world of multi-level modeling? Simultaneous equation models? Let's do it.
People like me do more math than your average biologist.
Experimentation? Um... excuse me, but Albert Einstein didn't do experiments. He did thought experiments. Train traveling near the speed of light? You do know that he didn't actually accelerate a train to near-c, right? In fact, he developed the concept of quantum entanglement as what he thought was a disproof of the Heisenberg uncertainty principle. You can't simultaneously measure the location and velocity of a particle, Einstein scoffed?! Well, what if two particles interact, and I then measure the location of one, and the velocity of the other, huh, wise guy?! I can calculate what I didn't measure! The only way Heisenberg can be right is if measuring the location of one particle affects the location of the other! And that would be so totally insane that it can't possibly be true!
Reductio ad absurdum, Heisenberg is wrong, my work here is done.
And then some wise guys came along and did the experiments, and quantum entanglement turned out to be real. Crazy, right? But you know who didn't do the experiment? Einstein. 'Cuz he didn't do that kind of thing. He did thought experiments.
Which is actually what a lot of economists do, and what I do.
And like I said, nobody is out there doing experiments on black holes. I suppose eventually, someone might try, and you can recreate "the big mistake of '08" from Dan Simmons's Hyperion novels, and move the Earth through a farcaster portal, but right now, that's not experimentation. It is observation. Deal with that, Mr. Keats.
You know who does experiments? Lots of social scientists. Granted, some of it is fraud. Cough, cough... Zimbardo... We even have experiments in political science. Some of that is fraud too, but we try to catch it. However, are we ruling out observation as science? Have fun with that one. I guess astronomy isn't science, nor is geology, nor... you see my point.
No, that ain't it. This "hard sciences" thing... this has bugged me for a while.
But right now, I'm coming back to that survey question, and for some time now, I have been struggling with the basic question of how much existing social science research ceases to be applicable as social structures change.
So here's something real. People react to their environments. By that, I mean not just the physical conditions in which they live, but their social, political and economic circumstances. Put you in a different time and place, and you act differently.
And this is where social science is different. The laws of chemistry and physics and biology get to be called, "laws," because they are invariant over time. There aren't a lot of things we call, "laws," in the social sciences. Part of that is a higher level of uncertainty around many of our findings, but another part is that many of our findings are contingent. If it's contingent, it's not a law. It's a thing we observe in a given time and place, that we can't be confident would occur in another time and place.
Here's what I mean. I am extremely confident that party identification is the most important determinant of vote choice, and most other factors in public opinion and political behavior. How confident can I be that party identification would retain that primacy in other times and places? Not very. In the extreme, totalitarian countries ban opposition parties. Orban is now a full-on dictator in Hungary. Party ID doesn't matter anymore there, and even prior to his recent moves to consolidate power, elections were basically a sham. See what I mean?
What would it take to reduce the primacy of party ID in the US?
I don't care what the answer is for the purposes of my post today. My point is that if there is an answer, then we don't have a law, in scientific terms. We have a context-dependent finding, that has been tested and replicated through the scientific method in its appropriate context. However, if it is context-dependent, we are in a different realm from chemistry and physics.
And here's where things get messier. Economics. I'm not sure if you have noticed, but the economy is kind of a disaster right now. This isn't just a conventional recession. Governors have ordered businesses closed, making this recession different. Does lifting the order return the economy to "normal?" We don't know. Probably not. Can distributing money, in something like a Keynesian way, have the same effect? When businesses are still shuttered? Maybe not. You see the point. And if you re-open businesses too soon, coronavirus has a second wave, and we're back into another recession, another lock-down, and... this is not 2008. This is not 1929.
Amid this, we have electoral chaos of every kind. What's the precedent for this? No precedent, no social scientific model. No understanding.
No... law, in scientific terminology.
I had written, in a variety of ways, about how politics had been changing in ways that make existing scholarship questionably applicable prior to the coronavirus crisis. Right now, this is a scientific crisis as well. We are living through a moment in history. Most of us, anyway. And as the epidemiological numbers come in, coronavirus does appear less deadly than those early 2% estimates. That's very good! We are also flattening the curve through lock-downs and competent planning, because many of our state and local governments are handling things well, as opposed to those who just want to go on tv and yell at journalists for insufficient obsequiousness.
Mike DeWine.
Yet, right now, April 19, 2020, we are living in a turning point in history. On September 11, 2001, it looked like the world was in the process of transforming dramatically. In retrospect, it didn't change as much as it could have. Mostly, it was a day of uncertainty. We didn't know whether it was the first wave of coming attacks, or if al Qaeda had just shot its wad. It turned out to be the latter. But, we didn't know. This is so much bigger, and it is hard to see either how we return to the status quo ante, or what happens next.
And social science really is different, as loathe as I am to admit it, because human behavior is contingent. Contextual. How much of existing scholarship, across multiple disciplines, will become obsolete? I don't know, but that doesn't happen in physics. Physics can be debunked, but not shown to be obsolete.
What am I doing? What are the social sciences doing? Right now, we're adrift. Anyone who says otherwise is putting up a front.
Living through history. Weird. I suppose it's better than dying through history, though, right? Then again, the one thing guaranteed to kill you eventually is history.
Gee... why might that question have any resonance for me, or anyone else, right now?
In fact, I had been pondering that one before COVID the Birdie started circling our collective heads, as several posts here indicated.
In light of that, here's a lecture I've been giving students on the first day of Introductory classes, year after year, since I started this whole gig. It goes a little something like this. I start by asking students how the dinosaurs died. Inevitably, there are a few kids who went through a dinosaur phase, who know immediately about the meteor. I then tell them that that wasn't what I read when I was a kid. And I ain't that old. When I was a kid, that was idle speculation. Common hypotheses were climate change and disease. What happened? They found the iridium layer at the 65 million year stratum, iridium being very rare on Earth, but very common in meteors, and satellite imagery found the crater, much of which was under water, off the Yucatan, but it was so big that you had to have aerial imagery to see it. Point? Textbooks get stuff wrong, and the sciences-- the physical sciences-- are about advancement, not a list of facts.
The next phase of my mini lecture goes as follows: "science" is a method, not a subject. It is the process of testing falsifiable hypotheses through a combination of observation and experimentation. That process can be applied to questions about chemical compounds, with an emphasis on experimental approaches, geologic and astronomical processes through observation (have fun trying to experiment with plate tectonics or creating black holes), and it can be applied to questions of politics. Science is not only the best method for answering questions about how the world works, it is the only method specifically designed to converge towards the right answer. Note my wording. "Converge." A scientific study will not always give you the right answer. Science is a process, and it is a collective process between a bunch of people engaged in scientific studies, interacting with each other. Scientists replicate each others' work, build on each others' work, and through that interaction, the scientific process is the process of improvement. That's the short version. Give me more time and I'll give you a lecture on Thomas Kuhn, paradigms, paradigm-shifts, and other stuff, but that's basically how I introduce the concept of science.
How I introduce Political Science.
And I push back pretty hard on the phrase, "hard science." Note my phrasing, "physical science." What makes "hard science?" Math? I do math. Plenty of it. Statistics, game theory, social choice theory... Try reading Kenneth Arrow's Social Choice & Individual Values sometime. Try reading some Richard McKelvey. I've had some badass math students in my office try to work their way through the chaos theorem. Not chaos theory, but theorem. You want to delve into the world of multi-level modeling? Simultaneous equation models? Let's do it.
People like me do more math than your average biologist.
Experimentation? Um... excuse me, but Albert Einstein didn't do experiments. He did thought experiments. Train traveling near the speed of light? You do know that he didn't actually accelerate a train to near-c, right? In fact, he developed the concept of quantum entanglement as what he thought was a disproof of the Heisenberg uncertainty principle. You can't simultaneously measure the location and velocity of a particle, Einstein scoffed?! Well, what if two particles interact, and I then measure the location of one, and the velocity of the other, huh, wise guy?! I can calculate what I didn't measure! The only way Heisenberg can be right is if measuring the location of one particle affects the location of the other! And that would be so totally insane that it can't possibly be true!
Reductio ad absurdum, Heisenberg is wrong, my work here is done.
And then some wise guys came along and did the experiments, and quantum entanglement turned out to be real. Crazy, right? But you know who didn't do the experiment? Einstein. 'Cuz he didn't do that kind of thing. He did thought experiments.
Which is actually what a lot of economists do, and what I do.
And like I said, nobody is out there doing experiments on black holes. I suppose eventually, someone might try, and you can recreate "the big mistake of '08" from Dan Simmons's Hyperion novels, and move the Earth through a farcaster portal, but right now, that's not experimentation. It is observation. Deal with that, Mr. Keats.
You know who does experiments? Lots of social scientists. Granted, some of it is fraud. Cough, cough... Zimbardo... We even have experiments in political science. Some of that is fraud too, but we try to catch it. However, are we ruling out observation as science? Have fun with that one. I guess astronomy isn't science, nor is geology, nor... you see my point.
No, that ain't it. This "hard sciences" thing... this has bugged me for a while.
But right now, I'm coming back to that survey question, and for some time now, I have been struggling with the basic question of how much existing social science research ceases to be applicable as social structures change.
So here's something real. People react to their environments. By that, I mean not just the physical conditions in which they live, but their social, political and economic circumstances. Put you in a different time and place, and you act differently.
And this is where social science is different. The laws of chemistry and physics and biology get to be called, "laws," because they are invariant over time. There aren't a lot of things we call, "laws," in the social sciences. Part of that is a higher level of uncertainty around many of our findings, but another part is that many of our findings are contingent. If it's contingent, it's not a law. It's a thing we observe in a given time and place, that we can't be confident would occur in another time and place.
Here's what I mean. I am extremely confident that party identification is the most important determinant of vote choice, and most other factors in public opinion and political behavior. How confident can I be that party identification would retain that primacy in other times and places? Not very. In the extreme, totalitarian countries ban opposition parties. Orban is now a full-on dictator in Hungary. Party ID doesn't matter anymore there, and even prior to his recent moves to consolidate power, elections were basically a sham. See what I mean?
What would it take to reduce the primacy of party ID in the US?
I don't care what the answer is for the purposes of my post today. My point is that if there is an answer, then we don't have a law, in scientific terms. We have a context-dependent finding, that has been tested and replicated through the scientific method in its appropriate context. However, if it is context-dependent, we are in a different realm from chemistry and physics.
And here's where things get messier. Economics. I'm not sure if you have noticed, but the economy is kind of a disaster right now. This isn't just a conventional recession. Governors have ordered businesses closed, making this recession different. Does lifting the order return the economy to "normal?" We don't know. Probably not. Can distributing money, in something like a Keynesian way, have the same effect? When businesses are still shuttered? Maybe not. You see the point. And if you re-open businesses too soon, coronavirus has a second wave, and we're back into another recession, another lock-down, and... this is not 2008. This is not 1929.
Amid this, we have electoral chaos of every kind. What's the precedent for this? No precedent, no social scientific model. No understanding.
No... law, in scientific terminology.
I had written, in a variety of ways, about how politics had been changing in ways that make existing scholarship questionably applicable prior to the coronavirus crisis. Right now, this is a scientific crisis as well. We are living through a moment in history. Most of us, anyway. And as the epidemiological numbers come in, coronavirus does appear less deadly than those early 2% estimates. That's very good! We are also flattening the curve through lock-downs and competent planning, because many of our state and local governments are handling things well, as opposed to those who just want to go on tv and yell at journalists for insufficient obsequiousness.
Mike DeWine.
Yet, right now, April 19, 2020, we are living in a turning point in history. On September 11, 2001, it looked like the world was in the process of transforming dramatically. In retrospect, it didn't change as much as it could have. Mostly, it was a day of uncertainty. We didn't know whether it was the first wave of coming attacks, or if al Qaeda had just shot its wad. It turned out to be the latter. But, we didn't know. This is so much bigger, and it is hard to see either how we return to the status quo ante, or what happens next.
And social science really is different, as loathe as I am to admit it, because human behavior is contingent. Contextual. How much of existing scholarship, across multiple disciplines, will become obsolete? I don't know, but that doesn't happen in physics. Physics can be debunked, but not shown to be obsolete.
What am I doing? What are the social sciences doing? Right now, we're adrift. Anyone who says otherwise is putting up a front.
Living through history. Weird. I suppose it's better than dying through history, though, right? Then again, the one thing guaranteed to kill you eventually is history.
I don't mind the 'hard' label, because I think that one of the things that implies is a lack of context. A benzene molecule is a benzene molecule is a benzene molecule. Not so for us soft scientists. Our subject is soft. It is squishy. It distorts, depending on context.
ReplyDeleteSo, I don't mind the label, because I understand that there isn't a subfield of physics called "solid dynamics," because solids are pretty easily explained with classical Newtonian physics. There is a field called "fluid dynamics" becauase, well, soft subjects are MUCH harder to study.
My brother got a degree in astrophysics. He studied gamma rays. One type of radiation. He studied it by looking at their interactions with the atmosphere. Essentially, interacting with about 6 different types of molecules.
We study people. There are 7.6 billion people now, and maybe 13 billion ever. Since history can affect current events, that's a lot of potential interactions. I tried to calculate 13 B choose 2, and the calculator literally said "infinity."
So, every time people ooh and ahh over my brother's degree, I say that my subject is infinitely more complicated.
The central claim at the heart of "hard" science, though, is the claim that it is more rigorous and more difficult, with that latter descriptor running up against your characterization that social science is more complex. If you take social science as a combinatorics problem, and reason from there that it is more complex than physics, then it follows that it is harder to do social science well, not because it is squishy, but because the problems are just harder to manage without resorting to cheats. I'm fine with that argument, by the way.
Delete