Another interesting essay I just read, by way of The Browser . . .
I usually say something about the whole 'dark matter' controversy in my Research Methodology (RM) class. I am no theoretical or astrophysicist, and any of you who know more on this (or not!!) are welcome to correct me or comment. (Here's the Wikipedia link on dark matter). But in simple terms, there is way more gravitational pull in space than is accounted for by the amount of matter available. Since no one knows where the excess gravity is coming from, they simply call it 'dark matter' (. . . and they sneer at social 'science'!). The problem is that no instrument has so far detected it. But scientists who question the theory are cast out into the netherworld. As Fry points out, "Astrophysicists who try to trifle with the fundamentals of dark matter tend to find themselves cut off from the mainstream." Remember Kuhn's 'normal science'? As Fry suggests, no one wants to say there is something wrong with the theory, because it will be too "drastic". "Physicists could take non-detection as a hint to give up, but there is always the teasing possibility that we just need a better experiment." I wonder how they would do in the social sciences!!
Here's the link:
Hope you enjoy it.
Updated on May 2, 2013:
After I emailed some of my graduate students this essay, I had an exchange with one of them, Kasturi Moitra. I am pasting the relevant part of our email discussion (with her permission) because it carries the discussion forward.
KM: The first thing that came to my mind was something about the Periodic Table. Apparently, when Mendeleev drew up the Periodic Table, not all the elements that are there today had been discovered. Yet, he had predicted their characteristics and had left ‘gaps’ to be filled up. So he knew that there was something, though he did not know what exactly it was. This somehow also reminds me of the time Waltz said that 'someday someone would fashion a unified theory of international and domestic politics'.
I was also wondering if one could maybe draw a parallelism between the ‘dark matter’ referred to in this article and the notion of the ‘black box’ of the state in IR parlance. As the author pointed out ‘Dark matter is the simplest solution to a complicated problem’. So basically, scientists came up with an anomaly or an instance that they could not explain with existing set of theories — ie: the extra gravitational pull. So they came up with an explanation by way of ‘dark matter’. Can we say that in a similar vein, when Waltz’s structural explanations could not account for why states similarly placed in the system behaved differently, the reserve category of the ‘black box of the state’ was created to keep the hard core of the realist scientific research programme intact? By doing so, anything anomalous that diverged from structural realist predictions could be said to be attributable to causes that lay within the ‘black box’. Again, just like the scientists in this article were not able or willing to measure or gauge much about ‘dark matter’, social scientists in IR were not willing to prise open the black box of the state. But at the same time they could not deny that it was there. Much later, when there was probably a Kuhnian ‘consensus’ generated in the IR community of scholars, was the black box opened by neoclassical realists (initially exposing them to the ridicule and wrath of their predecessors) and possibly signifying a shift from ‘normal science’ to a ‘revolutionary science’.
RR: Good point. My only thought is that in IR, no one disputes the importance of the state or decision-making processes, not even Waltz. And it was not that no one knew what was inside the black box but just that it was too complicated to explain it. Too many variables, not enough theory or time. The example I would use to explain this (I may have said this before too) is meteorology: everyone knows the laws that govern the various elements that make up weather, but it is so complicated with so many variables that no supercomputer can calculate how all these variables interact or their outcome. So we use simplified models, using fewer variables (bureaucratic politics? elite consensus? class interest?) but of course we know from daily weather predictions that they often fail to predict rain though they can almost always predict cyclones (well, . . . at least their path, if not when they form). On the other hand, the black box has also been used as convenience by Realists: India doesn't behave as expected? blame unpredictable domestic politics or bureaucratic interests. So, a few random continuing thoughts . .
KM: The met department analogy is a good one. Some of these outcomes and processes are highly over-determined. And as the authors of the article that you sent us also say, 'A good theory is just complex enough'. Occam and his ever-obliging razor would also agree that 'when you have two competing theories that make exactly the same predictions, the simpler one is the better'. But the question is where exactly to draw that delicate cut between parsimony and better explanatory power?
RR: I don't think parsimony vs. explanation is the real issue. What Waltz objects to (and I tend to agree with him) is not adding more variables for greater explanatory power but whether there really is any additional explanatory gain at all. In most cases, I don't find additional explanatory pay-off as a consequence of adding more variables. In such cases, why add more variables for no reason? A good example is Snyder & Christensen's "Chaingangs and Passed Bucks". They add additional variables (offense-defence balance) in the hope that it would lead to better explanation of alliance choices in multipolarity (why did chainganging alliance form before WW1 while buckpassing was the pattern before WW2 though both were multipolar periods), but conclude in the end that many more additional variables may be needed in other circumstances than just the offense-defence balance. Essentially, then, the offense-defence balance argument only explains the two cases they test (if one accepts their argument, which is a big 'if'). So what's the point? But where you think there is an additional explanatory pay-off as a consequence of adding more variables, obviously you should. Parsimony is no value by itself. At least, my sense.