Gavin Schmidt and Elisabeth Moyer wrote an essay for Nature:
A believer prays to the Earth - oh, you're so cool - in the land of Climatia (conventionally known as Alarmistan). His female colleague looks at the ocean because it shows that the Earth and its climate are so deliciously flat; at least she thinks so. The huge continent of Skeptia (whose tiny portion was shown on the picture, for propagandistic reasons) finally sent some dragons to the totalitarian country of Alarmistan, in order to restore freedom, democracy, integrity, and common sense. Thank God. An economist is just looking at Shell.
Even though this topic has no direct relationship with the alarmism of the authors, I happen to disagree with both of their points. My attitude to the problem of "interdisciplinary" research is the following:
"Interdisciplinary" buzzwords vs quality
I don't claim that it always has to be like that. But my experience is that people usually tend to use "interdisciplinary" buzzwords if they are not good enough in the normal disciplines. I won't tell you any names but be sure that I know dozens of good examples. It is very natural for people to do such things because one of the general features of the "interdisciplinary" research is that it is not monitored too well and it usually doesn't have well-defined standards. So it is attractive for many people.
On the other hand, I know several people who are doing amazing things that could surely be classified as "interdisciplinary" if you wanted except that these people never feel the necessity to talk about it in this way. They can simply do the right things in the context of any of the "old" well-defined disciplines. For example, quantitative biology could be viewed as an emerging discipline in between biology and mathematics.
But those people who do it well can do it both in maths departments as well as biology departments. And the really good ones could do other things in one of these departments, too. They may be "multidisciplinary" rather than "interdisciplinary". There are also people in this "interdisciplinary" field who are not that good.
Gaps are natural
There is only one Universe (or, to say the least, one multiverse) and one science. But it is divided to subdisciplines. While the precise way how the disciplines were separated depends on the human culture and the historical twists and turns - and in different places, there can exist different boundaries - all these socially influenced conventions depend on some objective underlying reality, too. If there are objectively existing natural quasi-boundaries between the topics studied by two disciplines, the people who study these topics will naturally split into cultures, too.
Whether or not they feel the need to be unified depends on the amount of interactions and the exchange of information between the disciplines. Sometimes, such a flow may be very useful. Sometimes, it may be less useful. It would be absurd to say that such a flow is always useful and should universally be encouraged.
Should people transgress the boundaries?
Now, should the people be transgressing the boundaries and switch their disciplines? Obviously, some people are doing so while others are not. Should you support one group or another? The answer is, once again, No. When people switch theif fields, such an event has both advantages and disadvantages.
Among the advantages, you may see that the people bring new fresh air and ideas to the other discipline. They may help to illuminate the previously obscure questions on the boundaries of the two disciplines under consideration. The disadvantages include the need to learn completely new things. Such a need often makes a large portion of their previous education (an investment made by themselves and the society) and their work useless - a waste of time and effort.
The advantages and disadvantages are compared and they help to determine the optimum balance - the number of people who actually do such things. Attempts to inhibit or expand the process of migration would be two equally counterproductive deviations from the balance dictated by the market of ideas.
Should new disciplines be born?
Also, the world needs no one who would be creating new disciplines. Whenever an interesting body of new insights is created, it can be studied by the representatives of old disciplines for a while. Eventually, there might be reasons to transform the new subfield into a new discipline - when the required specialized knowledge becomes extensive enough for these people to fail to be experts in the original, "bigger" disciplines. Disciplines should be as large as needed for their required amount of knowledge to match what a typical researcher can learn in a reasonable timeframe.
Once again, it is a bad idea to speed up or slow down the invention and creation of new separate disciplines. There may be many regions in between the disciplines that will become hot and important fields in the future but it is unlikely that these future hot disciplines are exactly at the places where random "interdisciplinary" researchers beg for funding today. ;-)
Different cultures?
Different disciplines tend to approach their problems - and the world in general - differently. It's because different classes of questions about the world make different approaches more optimal than others. Some fields may require a high degree of mathematical rigor. Others may emphasize the depth of their understanding. Another class of disciplines may find it unimportant for the deep explanations to be found and its members are satisfied as soon as they know the correct answers. One discipline may expect the numbers to be very accurate. Another discipline is satisfied with a lower accuracy.
One discipline may build on very complicated models and another discipline prefers robust models with a small number of wheels and gears. Whenever the scientists in the disciplines behave rationally, all these differences between the cultures reflect genuine and objective differences between the topics studied by these disciplines. Different topics make different strategies more promising - even though some narrow-minded simpletons are unable to realize this fact. But if there are significant differences that arise purely due to the "culture", not due to the objective causes, you can be sure that at least one of these "cultures" is behaving suboptimally.
Climate vs economics
Schmidt and Moyer argue that the climate models are much more complicated than the models in economics. The economics models look rudimentary to the climatologists while the climate models look like Rube Goldberg machines to the economists, they say.
Is it true? Is it justified? It is good? Is it bad? Well, I think it is true to some extent. But such differences can only partially be justified by the real differences between the discipilines which means that the differences reveal that at least one of the disciplines is doing something wrong. What do I mean?
Many climatologists believe that they know all the crucial physical phenomena that govern the climate and they can properly combine them into the models. Because the climate is a pretty complex system, the models will be pretty complicated. They may reflect many complexities of the 3+1-dimensional setup.
Economists tend to think that the relationships between many observables are badly understood. Their field is about the human behavior, after all. Human behavior is more complex and less predictable than physics, isn't it? So economists tend to use models with a limited number of observables.
The two previous paragraphs could explain why the climatologists tend to invent more convoluted models than the economists. But does this explanation of the difference also fully justify the different strategies? Well, not really.
When you try to decide about the optimum complexity of your model to learn something new about the system you study, what should matter are not your pre-determined, a priori opinions how much you want to include and how many effects you have heard about. Instead, you should look at the a posteriori results of the models with one degree of complexity or another. How many correct things are you actually learning by writing one model or another?
Are the models' complexities justified?
The thing that matters is the actual ability of the complicated models to reproduce the features of reality that a discipline wants to master.
In high-energy physics, you clearly need rather complex "models" - nothing easier than quantum field theories and/or string theory is enough to understand the phenomena that we can actually comprehend and predict today. They (quantum field theories etc.) are complex for the laymen but the experts know that in their essence, these models are actually extraordinarily robust and they don't have too many independent assumptions, parameters, or basic concepts, especially if you compare them e.g. with climate models (and string theory is even more robust, in this sense, even though this fundamental point is completely inpenetrable for 99.9% of the public). But what about climatology and economics?
I am convinced that if you look at the typical accuracy and predictability of phenomena in climatology and economics, you will see that they are fully comparable. Both of these disciplines work with a lot of chaotic functions of time, complex systems influenced by thousands of effects. And both of them end up with predictions that differ from reality by something comparable to 10-100% or more. ;-)
In my opinion, this fact - the actual ability to tell you useful and true things about the world - should decide about the optimal complexity of the models you use. And it is the economists who are approaching these problems more sensibly, by paying some attention to Occam's razor. If more detailed models with many new independent wheels and gears don't allow you to make more detailed and accurate predictions that can be verified than the older and simpler ones, there is no rational reasons to add these wheel and gears.
Economists don't add them unless they're forced to. Climatologists often add these features even though they are not tested and they don't play any demonstrably good role for their understanding of the actual physical system, the climate. Nevertheless, they seem to be proud about the complexity, whether or not this complexity teaches us something about the real world.
What is needed to fix this situation are not "interdisciplinary" researchers but rather climatologists who have sane ideas and a lot of expertise about the wheels and gears that have actually been validated. And those who know how to repair them or replace them whenever it is needed. To summarize, I think that what climate science and other sciences need are not researchers who are more "interdisciplinary" or a "new kind of scientists" but researchers who are more "good" in their job which should be the "old kind of science", not a new one.
And that's the memo.
Bonus: records of the day
The monthly average Sunspot number for June 2008 was 0.5, the lowest figure since June 1954. Also, Anthony Watts points out that the January-July increment of Mauna Loa CO2 reading was negative in 2008, for the first time in recorded history. Someone apparently began to suck CO2. ;-) (Update: it was of course an error in the data that was later corrected.)
A new kind of scientist (click)It almost sounds like Steve Wolfram. ;-) Schmidt talks about it at RealClimate, too. They promote "interdisciplinary" research because it is "very important" but they warn that it is "very difficult", too.
A believer prays to the Earth - oh, you're so cool - in the land of Climatia (conventionally known as Alarmistan). His female colleague looks at the ocean because it shows that the Earth and its climate are so deliciously flat; at least she thinks so. The huge continent of Skeptia (whose tiny portion was shown on the picture, for propagandistic reasons) finally sent some dragons to the totalitarian country of Alarmistan, in order to restore freedom, democracy, integrity, and common sense. Thank God. An economist is just looking at Shell.
Even though this topic has no direct relationship with the alarmism of the authors, I happen to disagree with both of their points. My attitude to the problem of "interdisciplinary" research is the following:
- the "interdisciplinary" nature of research is mostly a cover for low-quality research; the "interdisciplinary" researchers don't have to know anything well, and whenever they're caught by the true experts, they may pretend that they know something else, too (which is usually false)
- some gaps between the disciplines are natural and many pairs of disciplines are quasi-separated because there are not too many interesting things going on near their border
- it is not that hard to transgress the boundaries - and many people naturally do so - but it is sensible for most researchers not to do such things because their specialization prevents them from working in other disciplines equally well
- it would be counterproductive to artificially increase the ratio of "interdisciplinary" research; new disciplines and subdisciplines are emerging whenever some people realize that there is something interesting to investigate near the boundaries of the older disciplines
- different disciplines have different cultures and expectations because they have been shaped by the desire to understand a particular class of phenomena; however, there only exists one kind of rational thinking; the research in every discipline (and of every single scholar) is either good or bad (or something in between) and this quality doesn't depend on any "cultures"
"Interdisciplinary" buzzwords vs quality
I don't claim that it always has to be like that. But my experience is that people usually tend to use "interdisciplinary" buzzwords if they are not good enough in the normal disciplines. I won't tell you any names but be sure that I know dozens of good examples. It is very natural for people to do such things because one of the general features of the "interdisciplinary" research is that it is not monitored too well and it usually doesn't have well-defined standards. So it is attractive for many people.
On the other hand, I know several people who are doing amazing things that could surely be classified as "interdisciplinary" if you wanted except that these people never feel the necessity to talk about it in this way. They can simply do the right things in the context of any of the "old" well-defined disciplines. For example, quantitative biology could be viewed as an emerging discipline in between biology and mathematics.
But those people who do it well can do it both in maths departments as well as biology departments. And the really good ones could do other things in one of these departments, too. They may be "multidisciplinary" rather than "interdisciplinary". There are also people in this "interdisciplinary" field who are not that good.
Gaps are natural
There is only one Universe (or, to say the least, one multiverse) and one science. But it is divided to subdisciplines. While the precise way how the disciplines were separated depends on the human culture and the historical twists and turns - and in different places, there can exist different boundaries - all these socially influenced conventions depend on some objective underlying reality, too. If there are objectively existing natural quasi-boundaries between the topics studied by two disciplines, the people who study these topics will naturally split into cultures, too.
Whether or not they feel the need to be unified depends on the amount of interactions and the exchange of information between the disciplines. Sometimes, such a flow may be very useful. Sometimes, it may be less useful. It would be absurd to say that such a flow is always useful and should universally be encouraged.
Should people transgress the boundaries?
Now, should the people be transgressing the boundaries and switch their disciplines? Obviously, some people are doing so while others are not. Should you support one group or another? The answer is, once again, No. When people switch theif fields, such an event has both advantages and disadvantages.
Among the advantages, you may see that the people bring new fresh air and ideas to the other discipline. They may help to illuminate the previously obscure questions on the boundaries of the two disciplines under consideration. The disadvantages include the need to learn completely new things. Such a need often makes a large portion of their previous education (an investment made by themselves and the society) and their work useless - a waste of time and effort.
The advantages and disadvantages are compared and they help to determine the optimum balance - the number of people who actually do such things. Attempts to inhibit or expand the process of migration would be two equally counterproductive deviations from the balance dictated by the market of ideas.
Should new disciplines be born?
Also, the world needs no one who would be creating new disciplines. Whenever an interesting body of new insights is created, it can be studied by the representatives of old disciplines for a while. Eventually, there might be reasons to transform the new subfield into a new discipline - when the required specialized knowledge becomes extensive enough for these people to fail to be experts in the original, "bigger" disciplines. Disciplines should be as large as needed for their required amount of knowledge to match what a typical researcher can learn in a reasonable timeframe.
Once again, it is a bad idea to speed up or slow down the invention and creation of new separate disciplines. There may be many regions in between the disciplines that will become hot and important fields in the future but it is unlikely that these future hot disciplines are exactly at the places where random "interdisciplinary" researchers beg for funding today. ;-)
Different cultures?
Different disciplines tend to approach their problems - and the world in general - differently. It's because different classes of questions about the world make different approaches more optimal than others. Some fields may require a high degree of mathematical rigor. Others may emphasize the depth of their understanding. Another class of disciplines may find it unimportant for the deep explanations to be found and its members are satisfied as soon as they know the correct answers. One discipline may expect the numbers to be very accurate. Another discipline is satisfied with a lower accuracy.
One discipline may build on very complicated models and another discipline prefers robust models with a small number of wheels and gears. Whenever the scientists in the disciplines behave rationally, all these differences between the cultures reflect genuine and objective differences between the topics studied by these disciplines. Different topics make different strategies more promising - even though some narrow-minded simpletons are unable to realize this fact. But if there are significant differences that arise purely due to the "culture", not due to the objective causes, you can be sure that at least one of these "cultures" is behaving suboptimally.
Climate vs economics
Schmidt and Moyer argue that the climate models are much more complicated than the models in economics. The economics models look rudimentary to the climatologists while the climate models look like Rube Goldberg machines to the economists, they say.
Is it true? Is it justified? It is good? Is it bad? Well, I think it is true to some extent. But such differences can only partially be justified by the real differences between the discipilines which means that the differences reveal that at least one of the disciplines is doing something wrong. What do I mean?
Many climatologists believe that they know all the crucial physical phenomena that govern the climate and they can properly combine them into the models. Because the climate is a pretty complex system, the models will be pretty complicated. They may reflect many complexities of the 3+1-dimensional setup.
Economists tend to think that the relationships between many observables are badly understood. Their field is about the human behavior, after all. Human behavior is more complex and less predictable than physics, isn't it? So economists tend to use models with a limited number of observables.
The two previous paragraphs could explain why the climatologists tend to invent more convoluted models than the economists. But does this explanation of the difference also fully justify the different strategies? Well, not really.
When you try to decide about the optimum complexity of your model to learn something new about the system you study, what should matter are not your pre-determined, a priori opinions how much you want to include and how many effects you have heard about. Instead, you should look at the a posteriori results of the models with one degree of complexity or another. How many correct things are you actually learning by writing one model or another?
Are the models' complexities justified?
The thing that matters is the actual ability of the complicated models to reproduce the features of reality that a discipline wants to master.
In high-energy physics, you clearly need rather complex "models" - nothing easier than quantum field theories and/or string theory is enough to understand the phenomena that we can actually comprehend and predict today. They (quantum field theories etc.) are complex for the laymen but the experts know that in their essence, these models are actually extraordinarily robust and they don't have too many independent assumptions, parameters, or basic concepts, especially if you compare them e.g. with climate models (and string theory is even more robust, in this sense, even though this fundamental point is completely inpenetrable for 99.9% of the public). But what about climatology and economics?
I am convinced that if you look at the typical accuracy and predictability of phenomena in climatology and economics, you will see that they are fully comparable. Both of these disciplines work with a lot of chaotic functions of time, complex systems influenced by thousands of effects. And both of them end up with predictions that differ from reality by something comparable to 10-100% or more. ;-)
In my opinion, this fact - the actual ability to tell you useful and true things about the world - should decide about the optimal complexity of the models you use. And it is the economists who are approaching these problems more sensibly, by paying some attention to Occam's razor. If more detailed models with many new independent wheels and gears don't allow you to make more detailed and accurate predictions that can be verified than the older and simpler ones, there is no rational reasons to add these wheel and gears.
Economists don't add them unless they're forced to. Climatologists often add these features even though they are not tested and they don't play any demonstrably good role for their understanding of the actual physical system, the climate. Nevertheless, they seem to be proud about the complexity, whether or not this complexity teaches us something about the real world.
What is needed to fix this situation are not "interdisciplinary" researchers but rather climatologists who have sane ideas and a lot of expertise about the wheels and gears that have actually been validated. And those who know how to repair them or replace them whenever it is needed. To summarize, I think that what climate science and other sciences need are not researchers who are more "interdisciplinary" or a "new kind of scientists" but researchers who are more "good" in their job which should be the "old kind of science", not a new one.
And that's the memo.
Bonus: records of the day
The monthly average Sunspot number for June 2008 was 0.5, the lowest figure since June 1954. Also, Anthony Watts points out that the January-July increment of Mauna Loa CO2 reading was negative in 2008, for the first time in recorded history. Someone apparently began to suck CO2. ;-) (Update: it was of course an error in the data that was later corrected.)
Interdisciplinary trap
![Interdisciplinary trap]()
Reviewed by DAL
on
August 03, 2008
Rating:
No comments: