This is a study group of the book Dynamical Processes on Complex Networks, by Alain Barrat, and a discussion group of the book's applications to Economic, Cultural and Social Anthropology as well as to other social sciences.
Latest Activity: Oct 31, 2014
This group does not have any discussions yet.
What I'm suggesting ("arguing" is much too strong) is that the combination of Greek philosophy and monotheistic religion resulted in a cosmology that makes the epistemological questions how do we know and when is belief justified particularly poignant ones in what we call the Western World. Call it a bit of super-sized ethnographic observation. But more about that tomorrow. Tonight it is 1:30 a.m. Time to dream a bit.
Very interesting exercise to imagine having graduated from a British University a hundred years ago. However, having gone to a university at that time and place would have made me a member of the elite from birth, something that it is in itself for me hard to imagine (as having been a homeless vagabond is so much part of my identity). To respond to your comment, do you then argue that modern Western epistemology is a direct product of Western mythology and religion, even if scientists today try hard to separate their various and multiple religious beliefs from their scientific practice? I might think that it is a reaction to religion, but not a direct outgrowth. Also, if you read books like Counterfactuals and Causal Inference: Methods and Principles for So... it seems as though we have come a long way from dogmatic faith in mono-causality! Nevertheless, there are some people who claim that such books are distinctly Western, in that they advocate Western ways of knowing. I must confess that, for a lack of precision and clarity of what this Westernness in such books is supposed to be, I'm not so convinced my self. But if you or anyone you know, know(s) something about what may constitute a Western bias in our universities today, please let me know! John, you might also find this interesting, as do I: http://faithandglobalization.yale.edu/
What is exclusively Western about Western epistemology? Take a look at Western theology. Combine Greek philosophy and monotheistic religion (the Judaeo-Christian-Muslim tradition) and what do you get? A search for total explanations that come down to positing a single, invisible Cause that is separate from the phenomenal world perceived by our senses. Then the question naturally arises: How you can ever truly know that Cause or even demonstrate its existence. Or, in other words, how can you possibly know what you absolutely, positively, have to believe in—because if you don't, it's Hell.
Consider the alternatives described in the intro to a chapter on traditional Chinese religion I wrote for Ray Scupin, ed., Religion and Culture: An Anthropological Focus. I've attached an RTF to this message.
From the look of it, fun is certainly in the cards, I agree! While we are at it, might I ask you for your opinion on a question that has been going through my head for a while and that has nothing particular to do with dynamics on complex networks? The question is: What exactly is it about our Western epistemology that is exclusively Western, or in other words, what might be biases that we as Westerners have to guard ourselves against? It has been my contention that causality in and of itself is not the problem. I've read plenty of Chinese and Indian texts, both, modern and ancient, that were filled with causal questions and claims (it seems as though the Vedics were even more obsessed with causation then we are currently and they had answers to every causal question imaginable). So far, I have come to think that it is the way in which we partition the world and the concepts that we find worthy of investigation that makes our ideas distinctly Western; when we ask about the Wealth of Nations it is our obsession with wealth and our focus on the nation as the unit of analysis that makes our science a Western one. Any thoughts?
Allow me to play the devil's advocate, with the caveat that the devil is pro-science. You have, it seems to me, embraced wholeheartedly the material conditions of knowledge production prior to the digital revolution and development of the computational tools that you and I both delight in. If you look at slides 5 and 6 in my Zhengzhi University presentation on SlideShare you will see where I am coming from.
Slide five illustrates the old regime. Given limited resources, researchers were forced to choose between ethnography/qualitative research, asking a lot of questions of a small number of subjects, to get a sense of the lay of the land, and hypothesis-testing, which required the radical simplification of predictive models to make them testable by asking a few, very precise, questions of a large number of subjects. In this context, the process you describe, moving from ethnography to analytic modeling made perfect sense. The other way around led to the familiar, and often justified, critique that even models that tested well omitted too much to account for what was really going on.
Nothing mysterious here. Consider, for example, the law of gravity in relation to an airplane crash. Gravity is real. The law is an established part of physics. It does not, by itself, explain how airplanes fly. Still less does it explain why any particular airplane crashed, due, for instance, to ice on its wings or engine failure when a bird is sucked into a turbine. If you are a crash investigator, denying the law of gravity is silly. But stating only that gravity caused the plane to fall to earth isn't dong your job.
No consider that article you like so much. I agree, it's a nice piece of work. As an Asianist, I find it particularly interesting because it presents an alternative to the hydraulic civilization model described in Karl Wittwogel's Oriental Despotism. I wonder, though, how generalizable it is. I am familiar with terraced rice paddies and, generically speaking, the problem of who controls the flow of irrigation water from uphill to downhill. Is the Balinese solution to this problem the same as the one we find in south China, Korea, Japan, Java or Vietnam? If not, why not? Does water hoarding always lead to insect infections, motivating the people uphill to cooperate with the people downhill? Is this particular problem and its Nash-equilibrium solution the same everywhere? The questions multiply.
This brings me round to the argument I make at the beginning of my presentation. The new technologies now make possible exploratory analysis of large sets of quantitative data and, thus, allow the use of quantitative techniques to enrich ethnographic and other forms of qualitative research. Some already well-established results, e.g., the tendency of large networks to have a single giant component, seem like the law of gravity. We certainly can't ignore them. But if our task is to understand history or guide political action we can't depend on them alone.
I think of what Nassim Nicholas Taleb calls the ice cube problem. We see a small puddle in the street. That someone dropped an ice cube that melted there is a plausible model that accounts for the observation. That it recently rained or a dog took a leak are also plausible models.
I also think of Miller and Page's observation in Complex Adaptive Systems that most current actor-based computational models assume actors that are either too simple-minded (a few simple heuristics) or too sophisticated (using game theory), neither of which is a plausible account of the muddle of heuristics and calculations involved in human decision making.
Please note that I am not denying the value of the ethnography->analytic modeling process. What I'm suggesting is that we now have the tools to move in the opposite direction as well: from exploratory data analysis to ethnography focused and sharped by what that analysis, and related analytic models, suggest.
Yes, we are going to have fun.
Here is a link to an online book that I'm going through right now, called The Programming Historian:
My recent cure for the problem is to use ethnography merely as a tool for data collection and hypotheses generation and once one has enough of an understanding through participant observation, use precise analytical modeling to explain the system one studies in the vein of this incredible paper by Lansing and Miller (2003). After such an analytical exercise, it should be easy to explain one's finding to business people or politicians. The bottom line is that no one in their right mind is interested in whether the researcher is a Marxist or a Foucauldian Scholar (no one cares about ideology, in fact it is likely to discredit a scholar), people care only about whether the researcher's explanation rests on solid epistemological ground and thus can be trusted; and yes, it helps if it is clear and as free from jargon and unnecessary metaphysics or personal ideology. These are the questions every serious scholar asks him/her self: 1) can my theory be tested and if so, 2) what would be the observable implications that would falsify the theory. If the answer to question 1) is no, every serious social scientist will disregard the question as it is a metaphysical one. If the answer to question 1) is yes, the scholar moves on to ask whether or not the question would be of interest to anyone (which is certainly true if it has policy implications, or if it would solve a long asked previously unsolved scientific puzzle). If the question is found to be interesting to someone, the serious scholar then moves on to think about what would be the observable facts that would falsify the hypotheses (the theory-motivated proposed answers to the question). He/she then collects data in an attempt to falsify the hypotheses; this works by trying as hard as possible to find observations that will falsify the proposed answers. For the seminal text of how social inquiry is to be conducted see King, Keohane, & Verba (1996) and for a refinement of the statement see Collier and Brady (2004).
Johannes, just an update to let you know that progress is being made on the R front. The Windows to OSX crossover issue has been solved thanks to some friendly advice from Andrew Mrvar, one of the creators of Pajek. He recommends Tools>Export to tab-delimited file. The good point here is that I can keep the .txt files in a Dropbox or SugarSync folder and access them from all my machines. The editor issue has been solved brilliantly by RStudio. This baby is beautiful. If you don't know about it, you should give it a try.
The best text editor that I know of is emacs, it is really a delight to work with, free to download and it has modes for all kinds of programming languages. It runs R from within it as well and gives it a great interface. Furthermore, it is platform independent, so it doesn't matter whether you are running Windows, Linux or DOS.
First, a bit of happy news. Starting with the Princeton R tutorial I have been able to export a vector from Pajek, read it into R, and plot an admittedly still very crude chart of the distribution.
Since, however, I was having to do a little manual fiddling to get from Pajek (running under Windows) to R (running under MacOSX), I downloaded and installed the Windows version of R. It runs OK and Pajek has no trouble finding Rgui to automate the file export/import. That's the good news.
The bad news (Mac fanboy speaking) is that the Rgui for Windows is crap compared to the one for Mac. The latter thoughtfully adds closing parentheses, quotes, brackets as needed and positions the cursor where the next input should go automatically. The former seems as primitive as the editors I was using on the Dec Tops-20 system back in 1979. I wonder, are Windows R hackers masochists or is there a better Rgui out there than the one in the standard Windows distribution?
Re Siena: I've avoided it because at the Sunbelts I've attended, I've heard it described as very slow and unable to handle networks bigger than 100 vertices or so. Does Rsiena remedy these issues?
In any case, the Princeton tutorial is lovely. I have a feeling that once I have worked through it and the Stanford labs, I will be getting somewhere. As always, thanks for the pointers and links.
Welcome toOpen Anthropology Cooperative
© 2019 Created by Keith Hart.
Report an Issue |
Terms of Service
Please check your browser settings or contact your system administrator.