 So, we will continue looking at causal loop diagrams, introduce what causal link is and identifying variables and started looking at some of the guidelines, we looked at saying it as we now know known phrases and unambiguous polarities as well as choosing variable names whose normal sense is positive. So, these are three guidelines we saw, there are a few more guidelines which will also be illustrated through examples like we did yesterday. So, today the newer guideline that we can look at to begin the class is make intermediate links explicit. So, seal these, again I am going to draw the incorrect or less correct one on the left side and the correct ones on the right side of the page. So, what you are trying to see is they make intermediate links explicit making intermediate links explicit. For example, we may have a feeling that for example, say sales can lead to larger sales can lead to reduced price as the product sells more and more then the cost of the product can actually fall down. But if you just model it that sales results in reduced unit cost, then we are missing some key element in the model. For example, we may want to actually say that sales larger the sale it can have larger production volume which can result in production and the unit cost as the production volume increases, the cost comes down. So, this option here is little more intuitive for us. So, the larger quantity we produce reduced cost rather than this one here are we trying to link yes there is larger amount of sales because of which I am able to give you discounts things like that, but there is an intermediate variable here which now makes a little more explanatory a little more intuitive to understand or for example, if you want to say production capacity can lead to increased inventory. So, we have large capacity. So, we end up having lot of inventory these are some of the phrases which the managers tend to use we have large capacity with us. So, we ended up making lot of products and we are storing lot of inventory, but instead of capturing like this when capacity is more we are having lot of inventory we can actually say production capacity as it increases it tends to increase our production rate which tend to increase our inventory. So, the larger capacity I can afford to have larger production rate which can lead to increased inventory accumulation within our facilities. So, this is what we mean by making intermediately in exploration. Next guideline capture causation not correlation, correlation any two variables you can take X and Y anything literally anything in there is so much data set will literally take and compute its correlation coefficient it will give you some number it does not mean anything unless there is actual causation between them we cannot make any sort of inferences. However, strong there make sure the relationship is causal no matter how strong the correlation is you can come up with correlation almost any two variables for example, since say since the 1950s or 60s the atmospheric carbon dioxide also increased the crime rate is also increased that does not mean that we identify variable called atmospheric carbon dioxide and link it to crime rate. So, that is what I mean there are a lot of variables which are showing positive trends or the increase in ownership of cars and crime rate they may not have direct correlation among them just because it is there just because a strong correlation does not mean there is a linkages or a common statements like for example, let us say if there is a lot of firemen fighting a fire that means it must be a big fire that does not mean we write we draw firemen and draw an arrow linking it to fire because that just is not true. When we got the direction drawn so, relation is causation let us say let us say take a some other scenario like for example, in the summer months there is a lot of there is an increase in the sale of ice cream and in summer months there is more deaths due to drowning ok. So, we do not go ahead and say things like ice cream sales what you want to say is in the months of summer we have increase in the sale of ice cream and in summer there is more deaths due to drowning you can forget the data set you can definitely get a good strong correlation among them, but that does not mean that we say these things. So, as I told this side is incorrect side, but here when you just listen to the narration that I am saying at all that as there is in the months of summer more sales in ice creams are clean observed and in summer there is deaths due to drowning, but if you just only take data set you can show a strong correlation and so, we do not want to do this. What we instead so, perhaps let us say the average temperature can drive up the ice cream sales and increase in temperature can result in more activities involving water which should have led to deaths due to this is pretty much what you wanted to cover it is not the correlation that you are interested in, but the causation that is actually driving different variables within the system. So, next guideline is make goals make goals explicit especially true for balancing loops or goal seeking systems or negative feedback loops all mean the same thing you want to make those goals explicit. Let me again come up with some examples let us say product quality and quality improvement programs. So, product quality is low then we will have large number of quality improvement programs right. So, product quality is low we will have more quality improvement programs relationship is opposite. Let us more quality improvement programs happen quality should improve that is idea of quality improvement program. So, we can have a plus sign so, it seems fine, but then how long do we keep doing it? There has to be some goals towards which we want to address the progress correct as I told this side is correct. So, all these will be defined by the desired product quality. Product quality does not exist in absolute sense, it exists in a relative sense all quality is defined I am sure we can whatever product you take there are some quality standards that is being mentioned and accordingly people work towards that and spend corresponding amount of efforts. So, we want to make those things explicit. So, let us just go ahead and do that quality let us have desired quality quality shortfall then we can have a quality improvement program this link is the same. So, if you have desired quality product quality and the product quality I really want to see how different they are with my desired product quality. So, pretty much I am going to take the difference between them the desired product quality as well as product quality is there and a shortfall is more then I am going to have a quality improvement more quality improvement program. So, higher the shortfall more quality improvement programs I am going to conduct. When you compare both are negative feedback loops as you can see there is only one negative feedback link here negative link here this is negative causal link only one odd number. So, that is again continues to be a balancing. So, this construct is kind of common since whenever we want to have a goal then we try to define a desired state of the system and then we try to measure the current state of the system and take their difference or take their ratio or take their relative comparison identify shortfall. So, this construct you can see it many models where there are some goals that is being defined. Assume the desired quality is here and actual quality is here. So, if increase is the desired quality the gap increases. So, instead of shortfall let us call it quality gap the gap in quality increases. Here who determines the goal? No, no. So, who who eventually who will decide for any product who decides the quality then this selects eventually management decides the goal is good determined by the management goal set by management. Pretty much they decide this is the kind of quality I am going to produce and let us define that. So, those are the that is an exogenous variable right here. So, they have defined it and based on that other activities happen and for different product ranges, different companies they have their own standards and where against which they try to perform or try to produce and work. It is not that all the goals have to be set by the management or these kind of systems we can even take a more simpler systems like buying a hot cup of coffee. So, let us take that up coffee temperature cooling rate assuming we are having a hot cup of coffee. Coffee temperature is high the cooling rate is going to be high as the rate is high the coffee temperature is going to fall down that becomes a negative feedback system. As I told the making goals explicit is really comes into play when we have a goal seeking or negative feedback system. It is all positive feedback there is no real goal we are working towards. But this is not going to hold true because coffee temperature is going to only cool down as much as your ambient temperature right it is not going to go beyond that that is a law of thermodynamics. So, we have this case we have coffee temperature then we have room temperature temperature difference then we can have the cooling rate. So, in this in this case if you just think about it in the hot cup of coffee we typically expect the coffee temperature to be higher than your room temperature right coffee temperature is higher than your room temperature. So, we can put a plus and minus here whereas, higher the temperature difference higher the cooling rate higher the cooling rate lower the coffee is going to cool down faster until it reaches the room temperature. So, here the goal is set by law of thermodynamic determines the goals as I am sure you would have seen the simple equations on this and cooling of surfaces they are all differential equation. You can visualize them as a causal loop you can even simulate it of course of this lecture in a more fun way than what you probably learn this is what you probably are thinking with perceived and actual conditions. See here as I told we are looking at more not just physical systems we are looking at kind of socio-economic as well as environmental systems and what we are trying to model sometimes is includes the behavior of people behavior of the community etcetera. So, many times our decisions are not based on the actual values it is based on the our perception of reality. So, we may need to explicitly model that to say to know to account for the lag between the actual condition which is what we are going to perceive and based on which decisions are being made so that we can get a actual state. If we ignore the perceptions and ignore those kind of informations and beliefs then what you are looking at is a very very highly rationalized scenario which is what is not working. So, if you are able to capture some of these perceptions and beliefs and which we can do in our model it will help in explaining and explaining certain phenomena are trying to understand system better. So, for example, classical example here could be weapons race as one country gets more weapons, its arrival country is going to also increase its weapons and so on and so forth. So, this phenomenon we classically call as the weapons race arms race or weapons race. So, we usually write it like weapons of say nation A, nation B as weapons as nation A gets more weapons, nation B is also going to get weapons, nation B gets more weapons, nation A is going to go to weapons, etcetera, etcetera and it keeps going. So, that is what we classically call as arms race. But a better way to explain the same thing could be weapons of nation A perceived by B such weapons is nation A's weapons increases the threat perceived by B increases. As their perceived threat increases they are going to invest more in weapons of more weapons which is going to increase the perceived threat levels by A which is going to increase the weapons of A. So, the advantage of making these kind of perceptions and beliefs explicit is it gives us more options to identify the kind of solution. For example, in the first case weapons increases and B increases, there is no real way we can break it. But once we have here we have no extra variables to play with. How do I alleviate the perceived threat of the nation? Is there some other activity that they can do to ensure that the threat perceived by B is also going to come down because we need to know how to break this loop. So, that is essential thing that what we are trying to do because we can see the positive feedback system. There is nothing stopping the system from going exponential which is what happens. So, something has to trigger to slow it down. So, only thing is can we somehow come up with some still it is a positive feedback system. Maybe there is some other pathways which can lead to reducing these overall threat levels in such a way that there is some sort of saturation or some sort of a limit to this kind of growth in the actual physical weapons or some top edge now. So, that is where distinguishing between perceived and actual condition helps. There you have is indicate delays in links. So, again as I told this is correct. Yeah. So, it will have an exponential collapse or accelerated decay can happen. So, again that depends on what is the threat perceived. Maybe the weapons reduced, but nobody with the threat perception continues to remain. There may be other factor. This is just a very illustrative example. So, unless we look at the full picture we can have dialogues on it. Yes, it will if you only look at the numerical values and simulation kind of it, then you can have either exponential growth or accelerated decay. Both ways the equations will work. Road construction leads to say highway capacity. This kind of make it sound like road construction instantaneously results in highway capacity, but it is not so. Road construction takes lot of time. After sometime only your capacity is going to increase. So, we can indicate it two ways road construction. We can put a D on the link or we can write the road constructs and we can kind of put like this double dash on that on each link on a link. So, this link means that there is a significant delay before the construction completes highway capacity increases. So, this delay we know this is a total duration to build it. Same thing goes for hostels, but when it is reported is as if IIT increases students capacity by so much IIT has added so much new capacity, but there is a lot of lag. So, until then the problem does not go. So, it is good to capture that the kind of significant delays within a system. So, there are seen some of the guidelines. Again these are some basic instructions given. So, once we practice we will internalize it as we go along, but whenever we make the links we should come up with some justification why we are making some of those links. So, this justification range from for example, conservation considerations, law of conservation must hold. For example, if you are modeling inventory, inventory is changing then we need to account for that production is affecting inventory. Sake is also affecting inventory, ordering and receiving goods affects inventory as well as consumption affects. So, we may need to model both. Inventory cannot just disappear that is a whole laws of conservation has to hold. We can follow accepted theory, can you follow economic theory? For example, in the common person an economic theory suggests that as price increases the customer demand for. So, in accepted theory we do not it may not be actually applicable everywhere, but if you are modeling generic system that is a thing we start with. There is an accepted theory I am going to say that as price increases mid demand is expected to fall. So, then accordingly I can make a negative link there. Or it could be just instructions for modeling a manufacturing firm and the desired product quality is defined by the management and so on. So, you cannot argue saying that no, no this is too low quality. You can there is no need to argue there that is the quality standard they have described. Or if company has come up with say based on performance they are going to give bonus. So, that is instruction the company has formed saying that if the performance reaches so and so level then appropriate amount of bonus and incentives will be given to the employees. So, we just have to capture that you know. So, it is direct observation I do not know many of you then vegetable shopping and other thing, but then you know when supply is more vegetable price immediately crashes down and supply is less price increases. So, we do not need theories and justification for it you are saying it. So, that is another way to do it. But direct observation also involves field studies and getting gathering the data and based on that taking a call whether it is a positive link or negative link. So, the justification we are looking at is whether the link whether relation exists between the two variables and if so is the link positive or negative for that is what we are starting with. Once you set the overall direction right then we will move into what kind of equation will come. So, that is direct observation. Last could be the hypothesis or just an assumption we may not have very fable data, but maybe popular hypothesis. We are all used to making lot of assumptions to make a nice simple example like you know friction does not exist and then start modeling the system. So, if that is required then we can go ahead and do that or for everything we may not be able to observe like for example, I can assume that as the as oil reserves fall down the cost of extracting oil is going to increase. It makes sense it is quite logical I am making that assumption and then going ahead and modeling that system. So, those kind of assumptions also fine. Last one is statistical evidence I just told that correlation we should not use, that does not mean that we should not, we should ignore statistics to observe data and if there is a correlation, if there is a causation and then we can use the other statistical methods to identify the direction of causation is in statistical evidence also. But anyway I think I got the gist of it right. So, they are not very, we already seen some of those examples earlier. So, I was just plan to write the similar again. So, we just use these to justify the links. It is not that they are without drawbacks, there are some drawbacks and it has been pointed out many times of these causal loop diagrams. It tends us to be little more relaxed in whether we have got a complete set and some of the scenario the model itself can become pretty complicated in short time. More such points are available in this link here which can go ahead and try to see which are the scenarios where we can apply these kind of causal loop diagram or not.