Causality and models
January 15th 2007 20:36
Hey all,
here we go for a little reflection on causality. It would be good to take a look at it if your're not used to the thing, so take your time. We won't be getting to deep on it, since i'm not an specialist either...
I have to talk about this subject because it's always present in any theory. People in sciences quite like it, and it's truth that for any model that simulates the world (from clima to markets...) people are often concearned if something can be the cause of the thing they are modeling or not....
The ideia can be resumed in one paragraph: since we don't have access to noumena (the "real"), we cannot get to the point if A is the cause of B, or if it's just another reflex of a deeper thing C, which is the real cause of A and B.
Giving an example to help you out. Plenty of people take the following conclusion: "third world countries do not develop because they're in a favourable clima." It's know that the intersection between "poor countries" and "favorable clima countries" is important. But do the second is the cause of the first? Or maybe in fact we have: "poor countries are the ones that were colonized for exploration of their resources" and "favourable clima countries often have good resources." So, we have a new fact that is the cause of the first, but in the same time the second fact is also present.
So, being pratical, we can say that when A and B come togheter, we have 2 possibilities: either A is the cause of B (or vice-versa), or they have a same "root cause".
I'm considering that there is only one cause for these facts. We can also take into account that a fact can have more than one cause, even an infinity of causes, and in a linear combination of them all we have the conclusion if this fact will happen or not. Like all mathematical models used nowadays to predict from wheather to the market, we have a combination of facts, which have each one a weight in the final result. These models try to get as much facts (that could be considered as cause of the thing they are trying to model) as possible, to then apply an 80/20 rule, to filter the ones that are not so important.
So how do we take our conclusions on a system like the "real"? Do we have to throw away all our models, since they do not represent "noumena"? When Einstein discovered that everything could change in function of it's velocity (time could last longuer and space could be shorter) we did not threw away all Newton's laws, and we keep using them in most mechanics we find nowadays. So this means that, even if it's not exactly true, it is close enought. And this applies to all laws in sciences.
In the end, we can consider a fact A in the analisis if B will happen or not even if A is not cause of B, but if A has a same "root cause" of B. For us, even if that's not truth, that helps. I'm not saying that there's no "root cause" for all things, but we will never know if we got to it or not (and so we get back to Nietzche's genealogy...).
This reflection on causes allow us to the talk about how our inteligence applies to the world. By comparing facts we find analogies and things in common, which we use for prediction of what will happen a next time (what people call inductive reasoning). This goes in the opposite way of Kant's "synthetic jugments a priori".
For Kant, we know by heart that 5 plus 7 = 12 after learning that 1 plus 1 = 2. But you know this because your teacher teached you that 1 plus 1 = 2, and told you all the numbers. So, when she putted her two fingers togheter to tell you that 2 is this, when you make 5 plus 7 you putted 5 and 7 fingers togheter and found 12. So, no "synthetic a priori" things, you're just making inductions based on analogies with "facts" you have in your mind (that's pretty much the way we think). And like this we get closer to Sartre. I remenber that this does not mean i agree with all his work, as i do get plenty of things from Kant...
This post on causality can help you create any model on anything you want to know: if you want to see if something can be described by a combination of variables, you compare them (like in a mathematical correlation), add it to a list of "variables that describe what i want to know", give a weight for each variable you want to use and then add them all. This may seem tuff to understand in a post, but there are no big mathematical trics on it, and you probably do this naturally all day.
Example: you wan't to know if a party is good inside or not, before paying the price to get in. So you'll phone some friends to see if they know about it, you'll see if security is letting everyone in (it's bad sign..), you'll check if there are no other good parties in other place in the same time...all of this information you'll balance to get to the simple answer if it's worth to waste your money to get inside or not.
Cheers! Next time on beings, and we'll pass to ethics!
Uula
here we go for a little reflection on causality. It would be good to take a look at it if your're not used to the thing, so take your time. We won't be getting to deep on it, since i'm not an specialist either...
I have to talk about this subject because it's always present in any theory. People in sciences quite like it, and it's truth that for any model that simulates the world (from clima to markets...) people are often concearned if something can be the cause of the thing they are modeling or not....
The ideia can be resumed in one paragraph: since we don't have access to noumena (the "real"), we cannot get to the point if A is the cause of B, or if it's just another reflex of a deeper thing C, which is the real cause of A and B.
Giving an example to help you out. Plenty of people take the following conclusion: "third world countries do not develop because they're in a favourable clima." It's know that the intersection between "poor countries" and "favorable clima countries" is important. But do the second is the cause of the first? Or maybe in fact we have: "poor countries are the ones that were colonized for exploration of their resources" and "favourable clima countries often have good resources." So, we have a new fact that is the cause of the first, but in the same time the second fact is also present.
So, being pratical, we can say that when A and B come togheter, we have 2 possibilities: either A is the cause of B (or vice-versa), or they have a same "root cause".
I'm considering that there is only one cause for these facts. We can also take into account that a fact can have more than one cause, even an infinity of causes, and in a linear combination of them all we have the conclusion if this fact will happen or not. Like all mathematical models used nowadays to predict from wheather to the market, we have a combination of facts, which have each one a weight in the final result. These models try to get as much facts (that could be considered as cause of the thing they are trying to model) as possible, to then apply an 80/20 rule, to filter the ones that are not so important.
So how do we take our conclusions on a system like the "real"? Do we have to throw away all our models, since they do not represent "noumena"? When Einstein discovered that everything could change in function of it's velocity (time could last longuer and space could be shorter) we did not threw away all Newton's laws, and we keep using them in most mechanics we find nowadays. So this means that, even if it's not exactly true, it is close enought. And this applies to all laws in sciences.
In the end, we can consider a fact A in the analisis if B will happen or not even if A is not cause of B, but if A has a same "root cause" of B. For us, even if that's not truth, that helps. I'm not saying that there's no "root cause" for all things, but we will never know if we got to it or not (and so we get back to Nietzche's genealogy...).
This reflection on causes allow us to the talk about how our inteligence applies to the world. By comparing facts we find analogies and things in common, which we use for prediction of what will happen a next time (what people call inductive reasoning). This goes in the opposite way of Kant's "synthetic jugments a priori".
For Kant, we know by heart that 5 plus 7 = 12 after learning that 1 plus 1 = 2. But you know this because your teacher teached you that 1 plus 1 = 2, and told you all the numbers. So, when she putted her two fingers togheter to tell you that 2 is this, when you make 5 plus 7 you putted 5 and 7 fingers togheter and found 12. So, no "synthetic a priori" things, you're just making inductions based on analogies with "facts" you have in your mind (that's pretty much the way we think). And like this we get closer to Sartre. I remenber that this does not mean i agree with all his work, as i do get plenty of things from Kant...
This post on causality can help you create any model on anything you want to know: if you want to see if something can be described by a combination of variables, you compare them (like in a mathematical correlation), add it to a list of "variables that describe what i want to know", give a weight for each variable you want to use and then add them all. This may seem tuff to understand in a post, but there are no big mathematical trics on it, and you probably do this naturally all day.
Example: you wan't to know if a party is good inside or not, before paying the price to get in. So you'll phone some friends to see if they know about it, you'll see if security is letting everyone in (it's bad sign..), you'll check if there are no other good parties in other place in the same time...all of this information you'll balance to get to the simple answer if it's worth to waste your money to get inside or not.
Cheers! Next time on beings, and we'll pass to ethics!
Uula
| 43 |
| Vote |
