Episode 11

TOC Thinking Process – How to build logical reasoning without any additional investment (Part 4)

Category :  Thinking Process

In the past 3 episodes on the Thinking Processes series, we have discussed how the philosophy goes to the deepest level of a problem at hand, how it builds the logic around it and how it identifies the root cause to take a stab at it. Hence, it makes an extremely helpful tool for anyone, especially managers and leaders. However, it take s some amount of practice to become proficient in using the tool. It has some very easy-to-follow starting steps which can help one get started on the journey of using it.

This episode is dedicated to laying the foundation for starting on a journey to build simple logical constructs of cause and effect. It gets really interesting and extremely exciting from here on.

Transcript
Shubham Agarwal : Hello, and welcome to yet another episode on the Counterpoint podcast. We are here with Satya again in our fourth episode on thinking processes. So hi, Satya, welcome to the Counterpoint podcast once again. Hi, Shubham, how are you? I’m very good. How are you? Got vaccinated? Yeah. Great. That’s lovely. All right. So Satya the approach that you laid out, in the, you know, previous episodes is, is on how to use logic, or rational thinking to decipher new insights about an environment, right. But anyone can build a logical reasoning process, right, based on based on their own point of view, like we were discussing, briefly in a previous episode as well, now with different logical constructs, how does one reach a common conclusion?
Satyashri Mohanty : Okay. So, if you see the word causality, or the concept of causality, I believe is not very well understood.Now, let me ask you a question. Suppose you are having a headache. Okay. And you took a medicine, and the headache is now gone. Now, can you say that the medicine cured the headache?
Yeah.
Before the medicine, there was a headache, I took the medicine and after that, the headache was gone. So the headache actually was cured by the medicine. medicine was the causality is how a commoner would think. But if you ask a man of science, they would say no, this is not enough. Maybe the body cured itself. You just happen to take the medicine at that time.Okay.You know, the body has curative abilities, right?

That’s right. Yeah.
So how can you be so sure that the medicine caused the headache to go away?

You can only be sure if you know, a counterfactual scenario, which is what would have happened if I had not taken the medicine. Now, that’s a counterfactual that that doesn’t exist. You have already taken the medicine. Okay, yeah.

See, this is the problem with with causality. It’s not as easy as a layman builds causality. If you look at a rigor, we need to find out. What is the real reason and it’s not as easy right. For example, did the advertisement lead to increase in sales?

Shubham Agarwal : Yeah, that’s that’s a debate. I would agree, it’s a long debate, you can’t say for certain.
Satyashri Mohanty : You can’t say for sure. And it’s it’s a difficult, difficult one, as somebody would say, the sales would have gone up anyways. Right?
Shubham Agarwal : Yeah. Our product was too good, we were at the right place at the right time.
Satyashri Mohanty : Yeah, correct. So did the medicine cure did advertisement lead to increase in sales? Why is the reason for high employee attrition? Hmm, yeah, it could be all many reasons all over the place contradicting each other somebody saying that, you know, our employees are not getting good salaries. Somebody’s saying no, no, that’s not the real reason. It’s the stress in the work environment, right, I think could be all over the place. Right? And you, you’d agree that establishing causality is very important for managers, if you want to solve problems, right? Otherwise, we will just, you know, take a medicine without doing whether it really works or not. Okay, so. So we need to have a rigor, okay, in in thinking to establish causality.
Shubham Agarwal : Yeah. So can we look at some rules that we can follow here? Yeah, how to build that rigor?
Satyashri Mohanty : Yeah. Before that, I just want to highlight one important handicap in statistics, right? A lot of people say that you can’t we do number crunching and establish causality. That’s very important to understand that there is no mathematical operator, right? There’s no mathematical operator in statistics, which depicts causality, we can depict correlations right for example, we can say we know A is associated with B okay. Suppose if two variables so very high correlation, right. It does not mean that they are causally linked with each other.
Shubham Agarwal : I remember funny correlations like volume of chocolate eating done in a country is correlated with no of noble prize winners . I remember reading news items like universal health care breeds terrorists , living next free-ways causes autism. They are accidental associations falsely presented as causality.
I understand what you are saying – establishing causality directly from data of two sets of variables is impossible. We can only detect associations.
Answer
causality needs a rigor in thinking first, right? data crunching can come later, but rigor and thinking comes first. Okay, and that’s why there is a set of rules. Okay. Okay. Yeah, so let’s, let’s look at these set of rules, there are three levels of checks that we need to do right. And these are checks around a statement right our statement of causality is A is because of B, right? These are statements right? A is because of B these kind of things we say right? These sales are down why because quality of my products is low. Right. So, A is because of be now when these statements come in, we need to be very clear that we are communicating what is there in our mind first, whatever statement we are making, are we communicating what is there in our mind, it is a very simple level of check. The other level of checks are, have biases crept in, in our in our understanding of causality. Is this something that I’m saying? Without any basis? Is the is the logic presented sound or valid?
So the first check is a very basic check, which is about seeking clarity. Okay, okay. Let me give you an example. I make a statement. And a statement goes like this, we are not very customer oriented. Now tell me what what do you understand from this?
Shubham Agarwal : I don’t understand the customer. I don’t understand what the requirements are.
Satyashri Mohanty : Look the these entire word Yeah, the entire word could mean anything to anyone. Like somebody says we are not customer oriented. Because you know, the last time customer called up, somebody didn’t talk properly. Okay, and and he says, you know, customer orientation all about the way we verbally interact with our customers, somebody, you know, didn’t reply email properly. Now, for another person, customer orientation means you know, what our processes, you know, don’t deliver customer satisfaction. For example, you know, there are a lot of returns that are coming in is nothing to do with how I talk or anything it’s to do with returns that are coming in. Right and, and and the way I service those returns, right, it It shows that we are not customer oriented. The point here is that the word could mean anything to anyone. For example, I’ll I’ll I’ll take another one. Our employees are not very passionate. What does that really mean? What do you mean by passion?
Shubham Agarwal : Now, that’s subjective, obviously, what the company is trying to achieve from the employees and what the employees are doing themselves. There’s there’s not a coherence in that. Yeah. Yeah, there could be multiple things, you know, what do you mean by passion?
Satyashri Mohanty : Right? Or let me make it a little simple one, our inventory is high. What does that mean?
Shubham Agarwal : Yeah, we have more than what is needed at the point at our warehouses could be our inventories is high. Yeah, but But are you sure which inventory I’m talking about? No.
Satyashri Mohanty : So it’s very, very important that we seek clarity See, unlike science, where every word is precisely defined, for example, the word atomic weight is very precisely defined. A Russia would reach the same conclusion as an Indian would reach a conclusion, an American would reach the same conclusion and when they use the word our atomic weight or or what is an electron, right, it is so precisely defined people understand what it is and what it is not very clearly, but various terms of management, the way it is used in day to day language is not very well defined, which could mean I have a definition which is very broad, for example, you know, employees are not passionate, I could, I could have so many other things that is there in our mind and when I say that, you could have a different set of meanings for that word and we are not communicating with each other. Even a simple statement like, you know, our inventory is high, it’s very important to understand which inventory are you talking about? Is it the work in progress? Is it the WIP? Is it the RM? Are you excluding the inventory which is written off from the books? Or are you also including that, right, we need to first agree on what is the definition of the terms that you are using, so that you and I reached the same conclusion. Right? Before we start even evaluating causality, let’s understand that the the key words that you’re using in your sentence, I need to understand what is the definition that you are using? Right, I’m fine with that I just want to understand so that when I receive it as as a recipient, I also use the same definition not something else. Okay. So that’s, that’s what a clarity is all about.
Shubham Agarwal : In fact, I remember an interesting one, I had a discussion internally our market share is down. Now what is market share, there could be different definitions of market share for everyone in the company, for the same brand for the same product. So yeah, clarity is extremely important.
Satyashri Mohanty : The size of the market, in fact, somebody would be defining market saying that, you know, this is the total sales of the entire competition that I think is the competition, right? And out of that my sales is so much. So hence, my share is this much, right? And somebody would say, you know, what, why did you take out the unorganized segment? Right, that is also our market, right? So you define it in a different way. So it’s very, very important that when words are being spoken or written, we need to precisely define for that moment of conversation, right? Because we don’t have a standard definition like in science, which is said forever, we can agree with each other, what are the definitions that we are using the first level of check basic checks? And the second check is if there is any evaluation in that statement? Right. For example, I said, our inventory is high. Now, what do you mean by high? What are you comparing it with? Are you comparing with with with last year, last month? What are what are your What is your benchmark?

It’s very abstract.

Yeah, you need to tell me what is high. Okay. And the next check is, you know, I agree that it is high, it’s, you know, I understand what is inventory? But I also need to understand where have you collected the data from? Is it a sample? Is it from our ERP records? Is it from the Excel sheets, right? Many times, we know the data is from two three different places, and you need to find out where it has come from, right? These are the very basic checks before you even look at the causality and on a deeper level of checks. Right?

Shubham Agarwal : My black is different than your black.
Satyashri Mohanty : Yeah. So that that’s very important that we reach a common understanding. The next check is okay, I understand what you’re saying I understand if there is the evaluation criteria, I also understand that we reach a common understanding. But I seriously doubt whether what you said really exists. I’ll tell you a very interesting case. There was a statement made by one of the personnel our client organization, it’s a manager in the client’s organization. And he was saying, you know what, we have shortage of workers in the factory. Okay, no, somebody can ask what is shortage? What is workers? What is factory? You don’t have to go to that detail. But yes, there is a shortage of workers in the factory, we understand what he’s talking about. But the immediate thing that came in from the CEO, no way it is there, it is impossible that you have shortage of workers in the factory that means what he is doubting the existence of the entity. The entity is what the entity is that there is shortage of workers in the fact of the statement, that’s what I’m calling is entity and he has a serious doubt on it and that can happen. For example, somebody says my own time delivery is low, who said our on time delivery is low, our on time delivery is high. Okay, so next check after clarity check is reach a common understanding about the existence of the entity. Now how do you go about doing it?

Yeah, that right, that’s what I was about to ask

There. There is one check that you can do is you can actually get data in some cases, for example, in this case, what we understood that the CEO was always looking at a holistic data of all workers across all factories, he never had an intuition of the workers, which are there in the specific factory. And the manager out there had an intuition that it is a shortage. Why? Because he was facing the crunch almost every day, you could, you could hear those complaints, right? So here, if you find out okay, there are the your working norms tell us what is your working norms, and let’s compare where we stand against the working norm. So that kind of data can clearly bring about clarity on the entity existence, right? That Yes, this exists. This is the second check, right? But many times the data is not available. For example, in this case, let’s let’s assume that there is no such well defined norms. right for us to compare and announce that there is a shortage of workers. So, can you tell me what do we do now? He’s saying that the shortage of workers in the factory, there are no well defined norms or let’s say the norms were defined, you know, five, six years back, the CEO says that these norms are you know, as for the automation that I have done, these norms are okay. Right, they need not be revised. Right? On the contrary, the norms are to be reduced. So I believe the workers that you have is okay, and the poor guy saying no, the workers, I have a shortage of workers, the CEO saying no, no, you have enough workers? How do we sorted out now the data is not there. And norms are not there, I have the number of workers but I don’t have a benchmark to compare it to.

Shubham Agarwal : Probably I go back as in I reverse the tables, and I say, let’s consider one where we do have a shortage, and then we’ll see what happens due to shortage of workers correct. And then we also see
Satyashri Mohanty : yeah, so what do we do is we apply predicted effects. So if if the entity

We have discussed that in our previous episodes for all our listeners

Yeah here the predicted effect could be you know, guys, can we check the overtime? Can we check the overtime that this plant is doing right? Is it overtime as a percentage of total number of workers right if we find out what 70% of the workers are doing overtime, more than eight hours, it does talk about you know, that there is a problem there okay. So, entity existence can be checked with the with the data Okay, with the intuition of managers affected managers, if you believe them, right, and with some kind of a predicted effect checks, okay. So, these are these are basic checks. First is what we said is clarity checks. Let’s understand what we are talking about. The second check is does the entity exist? So this is what I call is level one checks, right? By the way, nobody does this kind of checks in meetings, you’ll see people immediately react reach conclusions, reach conclusion, fight. Provide your Counterpoint. Many times it makes sense just to you know, try to say boss, let me understand what you’re talking about. And I’ll say seek clarity and if you doubt the entity existence, ask those deeper questions predicted effect or ask for for data before you start getting into the next level of checks. The next level of checks are the difficult ones were now the logical construct of A causing B is being presented and now I want to check that okay. Now, what I want to check is what we call it the causality existence. Here I agree with you that A exists I agree with you B exists, okay. Okay. But I disagree that A is causing B which is called I doubt the existence of the cause itself. For example, you say that our sales is down because sales team is demotivated. Okay. I agree the sales team is demotivated, because the last employee surveys said that employee sales team is demotivated, I agree to that. I also agree to the fact.Yeah.

And we obviously know the sales is down.

But I I refuse to believe that this demotivation this is leading to the sales being down they might be correlated by the way they might be correlated in the sense that if I collect the data of you know, sales team demotivation was Since when did it kick in and I look at my sales they might be correlated, okay. But they may not be cause and effect, I doubt that the I have a serious doubt that the demotivation of the sales team is the cause for the sales being down.

Shubham Agarwal : So important to note here is that correlation and one leading to another is also different.
Satyashri Mohanty : correlation is not causation. You can you can hit it down the, you know, down your head, that’s something that we need to keep on saying to ourselves, that you know, just because the the rooster coows and then the sunrises. Right does not mean that the you know, the rooster causes the sun to rise, obviously so so, you know, this is a very naive example where we know it’s obvious but in many cases it’s not obvious right if somebody is presenting to you a data of you know, look at the correlation the sales team do motivation it kicked in, you know, around this time and I show the sales number going down this time and you say, wow, I have proven to you that sales is down why because the sales team is demotivated, no motivation, there is just correlations, it could be correlations, okay. It could be the fact that actually the common cause for both is quality of the product, because the quality is bad, the sales is down and the sales team is demotivated, why? Because when they try to go to the customer to sell the customer back to them, yeah. So the so they don’t like those conversation with the customer. So, it is not that one led to the other it is the common cause called quality problem, which is causing the sales demotivation and which is causing the sales to be down. Okay,
Shubham Agarwal : but there could be problems very, very subtle, right? as well, which we can’t we can’t find you know, a quality of a problem is a quality of a product is very, very easily perceivable understandable, but there could be something which is not so easy to understand and perceive
Satyashri Mohanty : Yes, it’s actually you know, many cases there are it is not that people do not know, they have a point of view, which is varying. In an organizational situation would never find a manager saying, you know, I don’t know everybody has a point of view would have a have a point of view, which is, by the way, is a is a good ingredient for further analysis. So, I am saying no, it’s the quality, right? And you are saying No boss is not the quality, the sales is down because the sales team is demotivated, and I’m saying know, what quality is the common variable, it is so called the confounding variable, the the hidden variable, which is impacting the dependent and the independent variable. Right. And that’s, that’s the main one. Now what we can do is we can we can look at, you know, control groups in data sets and trying to find out, understand, you know, let’s let’s take a case where quality problem is not there, right, let’s see what is happening right, we can we can do some kind of isolation kind of simulation or or data crunching and look at a lot of observational data and and, and look at comparisons to arrive at what is the what is the is this the causality? Right. So, so that’s that’s very, very important. And
Shubham Agarwal : So making those causality trees like we said in the last point also turning it around looking at that might really help at every point
Satyashri Mohanty : yeah. So, the So, the argument is like, if two people are are you make a causality tree right you say that A causes B in terms of a graphical diagram, you make it right and somebody else makes something else right. And there are there are two ways of of checking it right one is you try to say that okay, I agree A is there I agree B is there, but I doubt that you know that arrow mark that you have led I doubt that causality I doubt that because you are making the the phrase because is what I’m I’m debating against right. Now, if you have such a thing, right you can do these kind of control group checks, right. For example, if somebody says you know, our our sales is down, because you know, our our sales people are not getting enough salaries and hence they are demotivated and hence sales is down right. I can now look at the company’s data slice it down across various geographies and and look at all those cases where you know the sales are up okay. And now, we need to find out what has happened let’s say we look at the data of eastern India the sales are up okay and the salaries are same Okay, then you start saying that was whatever causality that you have built na that’s not correct, because the entire eastern India sales is up Yes, Western Indian sales are down drastically, but in eastern India, the sales are up and then the salaries are you know, the salaries are almost the same. So, why why that should be an issue right? Then Yes, there could be further counter in eastern India, our salary is competitive and so on so forth. So, you further slice and see, so, that’s the kind of, you know, the argumentation that can happen along.with. Yeah. And many times you can, you can do experiment, if you have a doubt on causality, the best thing is, you know, the experiment, but in management, many times we look at the past or something that has happened, okay. And, and then we try to try to explain and that’s why a control group is very, very important. No control group, right, and establishing that control group through slicing the data around, you know, smaller, smaller subsets out of the global subset And seeing what is happening there you can you can establish that yes this is the cause. Okay. The next one is that you

let’s say the causality also exists

causality also existed I agree that this is the cause, but, I, I also think that there is additional cause, for example, the sales Okay, went up, okay. And you can see the sales went up, you know, why because the scheme was great, I launched a scheme and that’s how the sales went up. Yeah. Now, I would say, you know, what, I agree the sales scheme, the scheme was really great. It caused the sales to go up, but I think there is also a normal market growth that has happened in your case, okay, the normal economic growth that has happened, that is also an additional cause. Yeah, right. So, that additive cause is also a reason Or, he might say you know, what, while you launch the scheme, I have given you the best sales manager for this territory, he also contributed something right and that cannot be taken out right.

Shubham Agarwal : Yeah companies generally do not give leverage or importance to other factors.
Satyashri Mohanty : So, we need to find out now, there is a there is a warning here when you look at other factors, you need to take out the noise level factors right you know, something which is in the realm of noise, just just ignore them otherwise, you know, you are to put in countless variables there. So, you need to have a sense of what is noise and ignore them and and that’s how the additional cause also comes in and if you have a doubt or a debate on the additional cause, again, some kind of a control group checks, crunching of data saying that okay, if if this additional cause has you know, done sales here, then look at another region where that additional cause was not there. Okay. So, let’s see what has happened there. So, this is about you know, the causal entity was there or causal entity was not there. So, you need to make these kinds of checks. As I said in the in the in the initial example, what would have happened if this was not there? And that’s a very important question to be asked right. And that is why you know, in medicine, they do the randomized control trial, what would have happened if you had not taken the medicine is established at a group level by having a control group is not given that medicine right and that is how they are established that yes, the medicine is actually causing the cure, because you need to compare with a counterfactual scenario and and the golden rule is all factors must remain the same. So, if there are, if there are old people here, there should be old people there right. If If or if, you know, age is a confounding variable, which means that age can impact the level of cure. So, both groups should have the equal representation of age groups and that is how those rules are put in right. So, we need to we need to you know, do these kind of checks after the event has happened, I think we will have enough data points to look into the past to look at smaller subset geography wise, you know, off of a past period and do these kinds of checks, that you know, if you’re claiming the causes, they are in effect is there what happens when the cause is not there, the effect should not be there, right? Correct. That is how causality is established, right. So, same check is there for the additional cause. So, this is the level two checks that is done.
Shubham Agarwal : Is looking at control groups the only method to establish causality?
Satyashri Mohanty : If there is an effect and you are trying to find which one out of many potential. You have to retrospectively do control group analysis.
If you have powerful new idea, which explains an unique effect which is beyond the realm of any noise. Then one experiment is enough – even without a control group – example DBR for low touch time….
If you have an idea which has an effect which is not unique – many other interfering variables can create same effect – then you need to set up a control group to establish causality
So, level one checks are just to repeat is you know clarity, clarity and entity existence level two checks are the difficult ones, which we are establishing the causality is existence, does it really exist? And there’s any additional additional costs right, the level three checks is is more about a theoretical explanation, right. So, you say that my scheme has led to increase in sales, okay. And you understand what his scheme what is sales? And you understand that yes, it has caused it, but you need to establish Why do I claim what is the theoretical explanation that the scheme will lead to an increase in sales? Right, what is there in the scheme that people would be motivated and why do you claim that people will be motivated right? And why any partners?my retailers, my channel partners, want to make more moneycorrect you need to put in there that you know, my scheme allows them to earn more Okay. Now, that that reasoning has to be also put in place, that reasoning also has to be well defined made very, very clear, you can say the scheme has led to this okay. I agree all that but what is your explanation, you have to say, you know, what, my scheme has these features, which enables my dealer to make more money. And I also claim that no other scheme is as powerful as the scheme. Right, and why do I say so? Right. So I should write down as my explanations my theoretical explanations, my conceptual explanations as to why I, why I claim that A lead to B is because of this, you know, theoretical construct that I gave. And so that’s the level three check. Another example is, you know, we in TOC implementations when we go to the plant, we typically see a behaviour, which is called efficiency syndrome in the plants. What it means is that you give people different kinds of things to manufacture people pick up the easy ones to manufacture they pick up over others, okay to, to show extra output of that Work Centre. Right. Now, I want to define this in terms of our logical construct, I want to draw this out. Now, this is what I’m saying is that because of the efficiency syndrome, okay, there is a cherry picking, that is happening, this theoretical explanation is very, very important that people behave in the way they are measured, that there’s a theoretical construct that people behave in the way they are measured. So if you have efficiency, as your prime measurement, right, people would do everything possible to pick up this easy to manufacture and show a local output. So my theoretical concert here is people behave in the way they are measured. And hence, the efficiency measurement would lead to cherry picking.

And that’s very important to do this level three checks. Once you have done all this, right, you have got your basic checks of cause and effect with you. Now you are ready to use the various tools of TOC,

Shubham Agarwal : Thank you everyone
Comments

Your Comment

Your email address will not be published. Required fields are marked *