Comments on: The problem with real-world problems http://langabi.name/blog/2006/02/19/the-problem-with-real-world-problems Paul Cook's blog Fri, 10 Feb 2012 20:58:00 +0000 http://wordpress.org/?v=2.6.2 By: Greg http://langabi.name/blog/2006/02/19/the-problem-with-real-world-problems#comment-1196 Greg Thu, 30 Mar 2006 03:18:03 +0000 http://langabi.name/blog/2006/02/19/the-problem-with-real-world-problems#comment-1196 Um, it's actually a bit worse than that, particularly in economics (which, as befits its imperial status, is dragging everyone else along with it). Even within social science, you can pick research topics, and some are more "clean" than others. As an economist, I can (and do) worry about the consistency of my quasi maximum likelihood estimator, and I can think through that problem, prove a few propositions and I'm done. Or, I can think about ROSCAs (Rotating Credit and Savings Institutions) in India, and see if they fit a particular auction model (a little less clean - what does a good "fit" look like?). Or I can worry about the political economy of Nigeria, and then I'm, well, not going to produce anything clean (indeed, I'm probably going to end up verbally arguing my point, without appeal to a strong quantitative methodology). Which one to pick? If you are of a purely scientific bent, you pick the former. If your inclination is towards academic social science, you go with the ROSCAs. If you want to set policy, you look at Nigeria. So - in addition to the problem that scientists are not devoting themselves to human problems - there's the potential problem that those people who ARE studying these questions are those that like to set policy, rather than those who are good at generating careful and scientific research. My impression is that this is certainly a problem in South Africa; it is mitigated in the US by the sheer number of tenured professors in the social sciences who have time to devote to both sorts of topics. Um, it’s actually a bit worse than that, particularly in economics (which, as befits its imperial status, is dragging everyone else along with it). Even within social science, you can pick research topics, and some are more “clean” than others. As an economist, I can (and do) worry about the consistency of my quasi maximum likelihood estimator, and I can think through that problem, prove a few propositions and I’m done. Or, I can think about ROSCAs (Rotating Credit and Savings Institutions) in India, and see if they fit a particular auction model (a little less clean - what does a good “fit” look like?). Or I can worry about the political economy of Nigeria, and then I’m, well, not going to produce anything clean (indeed, I’m probably going to end up verbally arguing my point, without appeal to a strong quantitative methodology).

Which one to pick? If you are of a purely scientific bent, you pick the former. If your inclination is towards academic social science, you go with the ROSCAs. If you want to set policy, you look at Nigeria.

So - in addition to the problem that scientists are not devoting themselves to human problems - there’s the potential problem that those people who ARE studying these questions are those that like to set policy, rather than those who are good at generating careful and scientific research. My impression is that this is certainly a problem in South Africa; it is mitigated in the US by the sheer number of tenured professors in the social sciences who have time to devote to both sorts of topics.

]]> By: Jonathan http://langabi.name/blog/2006/02/19/the-problem-with-real-world-problems#comment-976 Jonathan Tue, 21 Feb 2006 06:42:21 +0000 http://langabi.name/blog/2006/02/19/the-problem-with-real-world-problems#comment-976 That gets to the heart of the dilemma of social science - the attempt to force methodologies that assume a right answer onto material that is predictable only in large groups and in probablilities; and then with topics for which "right" in any case carries value baggage. This is illustrated historically, for example, by the research into how to "cure" gay people. It takes a certain resiience to persevere with research when you don't know if there is an answer or, if there is one, whether or not it should be the right one. In the absence of resilience, it takes a certain lack of insight or thought. Perhaps the latter is too true of many of us social scientists. Of course those who are of the persuasion that there are just two types of science, physics and stamp collecting, will no doubt find Rubik's cube much more fun. That gets to the heart of the dilemma of social science - the attempt to force methodologies that assume a right answer onto material that is predictable only in large groups and in probablilities; and then with topics for which “right” in any case carries value baggage. This is illustrated historically, for example, by the research into how to “cure” gay people.

It takes a certain resiience to persevere with research when you don’t know if there is an answer or, if there is one, whether or not it should be the right one. In the absence of resilience, it takes a certain lack of insight or thought. Perhaps the latter is too true of many of us social scientists.

Of course those who are of the persuasion that there are just two types of science, physics and stamp collecting, will no doubt find Rubik’s cube much more fun.

]]> By: Paul F http://langabi.name/blog/2006/02/19/the-problem-with-real-world-problems#comment-975 Paul F Mon, 20 Feb 2006 18:24:28 +0000 http://langabi.name/blog/2006/02/19/the-problem-with-real-world-problems#comment-975 I don't know if it is why scientists become scientists, but I think I do agree with your comment that we tend to like clean solutions. Certainly I would never want to publish work unless I could say: this is what happened and it will always happen like that! Following on from Adam's comments, while the science of modelling protein folding ab initio may not currently be "solvable", there clearly exists a solution or proteins could not fold correctly (in the time scales they do!) in the real world. I think the difference that arises is that of: in science it is a case that there is a possibly murky solution that may take several generations to find, but where the ultimate cleanest answer can eventually be found, compared to say art where there is no right or wrong as it is an expression of a feeling interpretted by the person viewing it and biased by their feelings. The value in this distinction comes when the clean answers are used successfully in society for its improvement. I don’t know if it is why scientists become scientists, but I think I do agree with your comment that we tend to like clean solutions. Certainly I would never want to publish work unless I could say: this is what happened and it will always happen like that!

Following on from Adam’s comments, while the science of modelling protein folding ab initio may not currently be “solvable”, there clearly exists a solution or proteins could not fold correctly (in the time scales they do!) in the real world.

I think the difference that arises is that of: in science it is a case that there is a possibly murky solution that may take several generations to find, but where the ultimate cleanest answer can eventually be found, compared to say art where there is no right or wrong as it is an expression of a feeling interpretted by the person viewing it and biased by their feelings.

The value in this distinction comes when the clean answers are used successfully in society for its improvement.

]]> By: Dixie http://langabi.name/blog/2006/02/19/the-problem-with-real-world-problems#comment-974 Dixie Mon, 20 Feb 2006 18:13:18 +0000 http://langabi.name/blog/2006/02/19/the-problem-with-real-world-problems#comment-974 This is why I knit and play computer games. There are problems, but they are designed to be solved. This is why I knit and play computer games. There are problems, but they are designed to be solved.

]]> By: Adam http://langabi.name/blog/2006/02/19/the-problem-with-real-world-problems#comment-969 Adam Mon, 20 Feb 2006 04:36:35 +0000 http://langabi.name/blog/2006/02/19/the-problem-with-real-world-problems#comment-969 I'm studying a decidedly unsolvable problem. Protein folding has infinite solutions. Our method for solving the problem is basically to use what knowledge we have to constrain the problem to a smaller set. There are still infinite possibilities, but we esablish a finite set of things that *aren't* right. Eventually we get something that almost definitely isn't right, but is close enough to being right that it's useful for real-world problems. It's not a clean solution, it's not a solvable problem, and in general our method isn't even all that elegant. But it's definitely a real-world problem. It *is* frustrating to not be able to say in the end, "This is Right." But if everything goes well, in the end you can say, "This can be (or has been) used to Help people." I’m studying a decidedly unsolvable problem. Protein folding has infinite solutions. Our method for solving the problem is basically to use what knowledge we have to constrain the problem to a smaller set. There are still infinite possibilities, but we esablish a finite set of things that *aren’t* right. Eventually we get something that almost definitely isn’t right, but is close enough to being right that it’s useful for real-world problems.

It’s not a clean solution, it’s not a solvable problem, and in general our method isn’t even all that elegant. But it’s definitely a real-world problem.

It *is* frustrating to not be able to say in the end, “This is Right.” But if everything goes well, in the end you can say, “This can be (or has been) used to Help people.”

]]>