Distrust Simplicity:

whitehead_painting

In today’s post, I will be looking at the famous quote from the famous English mathematician and philosopher, Alfred Whitehead.

Seek simplicity, and then distrust it.

This quote comes from his 1920 collection of lectures, The Concept of Nature. The quote is embedded in the paragraph below:

Nature appears as a complex system whose factors are dimly discerned by us. But, as I ask you, Is not this the very truth? Should we not distrust the jaunty assurance with which every age prides itself that it at last has hit upon the ultimate concepts in which all that happens can be formulated? The aim of science is to seek the simplest explanations of complex facts. We are apt to fall into the error of thinking that the facts are simple because simplicity is the goal of our quest. The guiding motto in the life of every natural philosopher should be, Seek simplicity and distrust it.

I like this idea a lot. We are all asked to keep things simple, and to not make things complicated. Whitehead is asking us to seek simplicity first, and then distrust it. Whitehead talks about “bifurcation of nature” – nature as we perceive it, and the nature as it is. Thus, our perception of reality is an abstraction or a simplification based on our perceptions. We need this abstraction to start understanding nature. However, once we start this understanding process, we should not stop. We should build upon it. This is the scientific method – plan the prototype, build it, assess the gap, and continue improving based on feedback.

As I was reading The Concept of Nature, several other concepts came to my mind. The first one was Occam’s razor – the idea that Entities should not be multiplied unnecessarily. Seek the simplest explanation, when all things are equal. At the same time, we should keep Epicurus’ Principle of Multiple Explanations in mind – If more than one theory is consistent with the observations, keep all theories. I also feel that Whitehead was talking about systems and complexity. As complexity increases, our ability to fully understand the numerous relationships decreases. As the wonderful American Systems thinker Donella Meadows said:

We can’t impose our will on a system. We can listen to what the system tells us and discover how its properties and our values can work together to bring forth something much better than could ever be produced by our will alone.

Seeking simplicity is about the attempt to have a starting point to understand complexity. We should continue to evolve our understanding and not stop at the first abstraction we developed. One of the famous Zen story is about the teacher pointing his finger at the moon. I have talked about this here. We should not look at the finger and stop there. We should look at where the finger is pointing. The finger is the road sign and not the destination itself. The simplicity is a representation and not the real thing. We should immediately distrust it because it is a weak copy. Seeking simplicity is not a bad thing but stopping there is. Simplicity is our comfort zone, and Whitehead is asking us to distrust it so that can keep improving our situation – continuous improvement. Whitehead in his later 1929 book, The Function of Reason, states:

The higher forms of life are actively engaged in modifying their environment… (to) (i) to live, (ii) to live well, (iii) to live better.

Final Words:

In seeking simplicity, we are trying to be “less wrong”. In distrusting our simplified abstraction, we are seeking to be “more right”. I will finish with a Zen story.

A Zen master lay dying. His monks had all gathered around his bed, from the most senior to the most novice monk. The senior monk leaned over to ask the dying master if he had any final words of advice for his monks.

The old master slowly opened his eyes and in a weak voice whispered, “Tell them Truth is like a river.”

The senior monk passed on this bit of wisdom in turn to the monk next to him, and it circulated around the room from one monk to another.

When the words reached the youngest monk he asked, “What does he mean, ‘Truth is like a river?’”

The question was passed back around the room to the senior monk who leaned over the bed and asked, “Master, what do you mean, ‘Truth is like a river?’” Slowly the master opened his eyes and in a weak voice whispered, “OK, truth is not like a river.”

Always keep on learning…

In case you missed it, my last post was Cannon’s Polarity Principle:

Advertisements

Cannon’s Polarity Principle:

arrows

I recently read the wonderful book “On the Design of Stable Systems”, by Jerry Weinberg and Daniela Weinberg. I came across a principle that I had not heard of before called “Cannon’s Polarity Principle”. Cannon’s Polarity Principle can be stated as the strategy that a system can use to overcome noise by supplying its own opposing actions. If a system relies on an uncertain environment to supply the opposing factor to one of its regulatory mechanisms, that mechanism must have a much more refined model. By supplying its own opposing factor, it can get away with a much simpler model of the environment.

This principle is one of those things that is profound yet very simple. The Weinbergs give the example of a sticky knob on a gas stove to explain this idea. If the knob is sticky then it is tricky to raise the flame to the precise point we would like it to be. Due to the “stickiness” we will try to apply much more force than needed and inadvertently overshoot, going past the desired point. The result is that the flame is at a much higher setting. When we try to turn the flame down we are still in the same situation and again go past the point where we would like to be.

What we can do instead is to use one hand to push against the direction we would like and then slowly try to turn the knob with our other hand. With this approach we can be much more refined and be at our desired position. By working “against” our own goal, we make precise adjustment possible in the face of an unknown, but small, amount of stickiness.

This got me thinking. There are several times where we apply opposing forces to slow us down, to take the time to reach the correct decision (precise adjustment). One of my favorite Toyotaism is – Go slow to go fast. This makes a lot of sense in the light of the Polarity Principle. Any time we are doing a root cause analysis, we are prone to a plethora of biases including confirmation bias – selectively looking for ideas that reinforce our thinking, and availability bias – latching on to the first idea because that was the immediate idea we came up with. These biases might make us jump to unwarranted conclusions to address symptoms, and not addressing the root problem(s). The Polarity Principle would advise us to slow down.

I will finish this short and sweet with an apt Zen saying:

The one who is good at shooting does not hit the center of the arrow.

Always keep on learning…

In case you missed it, my last post was Contextual Why:

Contextual Why:

Láminas_8_y_9_del_Códice_de_Dresden

One of the scientists that I have referenced in my posts a lot is the American physicist Richard Feynman. I particularly love his imaginary depiction of Mayan astronomy. Feynman went to Mexico for his second honeymoon and came across a copy of the Dresden Codex (one of the oldest surviving books from the Americas). He was particularly interested in the bars and dots in the codex. He was able to decipher the number system that the Mayans used to depict Venus’ trajectory in the solar system. He was so good at it that he was able to find that some of the versions were actually fakes. Feynman imagined the Mayans counting and putting nuts in a pot to make predictions of where Venus would be on a given day. Feynman was curious whether the Mayans actually knew what was happening (why it was happening) or whether they were going by the rules and making predictions based on a rule-based system of counting and manipulating numbers. Feynman stated that the Mayans may have gotten really good with counting but they must not have understood how the celestial bodies worked.

The push for following rules without understanding the context is unfortunate. Yet this is very prevalent. The rigidity of the rules cannot be sustained when a complex situation arises. The rigidity of rules indicates a direct linear relationship where cause and effect are clearly noted. This is the push for standardization and having one best way of doing things. This leads to stagnation, since this stymies creativity and the push for innovation. Rigid rules always break. Another way to look at this is as the push for robustness – avoiding failure by any means. We will put redundant steps, perform multiple inspections and implement punishments for not following rules. In the complex world, we should accept that things will fail – the push should be for resilience, getting back up in a short time. The rules are dictated top-down since the rules are created by the experts. These rules do not have the requisite variety to tackle the uncertainties of day-to-day dealings. The contexts of these rules do not match the actual context where the action takes place – the context at the gemba. Context is what brings out the meaning in a situation. The focus on rules and efficiency through best practice does not lead to having the requisite variety to change the context as needed to address a problem when it arises. We are involved in complex adaptive systems on a day-to-day basis. We need to change context as needed and adapt to respond to unanticipated events. Evolution requires that we have variety. This response is not always rule-based and is developed depending upon the context. We should allow room for bottom-up heuristics, since these are based on experience and local context.

As a simple example, let’s look at 5S, one of the most commonly identified lean tools, to look into this more. 5S is translated from Japanese as Sort, Straighten, Shine, Standardize and Sustain. The rules are provided to us and they are clear cut. Similar to the Mayan story, do we actually know the context for 5S? Toyota did not have 5S. The last few S’s were added on later. This has now changed into 6S and even 7S. The “sort” step in 5S is to have only the required tools needed at the station. The “straighten” step is to identify/label the tools so that operators from other shifts or job rotations can easily find the tools. The third step is “shine” where the work station is cleaned by the operator. This allows the operator to find any spills or other signs of wear and tear that may not be seen by a cleaning crew. These three steps help the operator to identify problems as they occur, raises awareness and helps to take pride in the work. The fourth step is “standardize” and this is mainly a regulatory function to ensure that the first three steps are followed. The last step is “sustain”, which means to integrate the first three steps so that they become the normal routine and if they are not followed, one feels like something is missing. The context is to help the operator do his or her job better and be effective. The context is that a problem is made visible immediately so that it can be addressed and people can be developed. The context is not following rules. The context is not applying 5S in areas where it does not make sense. The context certainly is not policing people. When the context of what the operator does is not made clear, they do what makes sense to them in their context – at that time with the limited information they have. Empty actions do not have context and are thus meaningless and non-value adding.

Seek to understand the perspectives of your employees. Seek to understand their local context. Seek to make them understand your context, and the context of the shared goals and objectives. Heed to their stories. Develop your employees to see problems.

I will finish with an interesting question that was posed by some French researchers in the late 1970’s.

“On a boat, there are 26 sheep and 10 goats. What is the age of the captain?”

Perhaps, you might see this as a trick question. Perhaps, you may use the two numbers given and come up with the answer as 36. The answer 36 sounds right. The answer that the researchers expected was “I do not have enough information to give the answer.”

To the researchers’ surprise, very few subjects challenged the question. Most of them reasoned in their context and came up with a number that made sense in their mind. We are not trained to ask the contextual questions.

Always keep on learning and ask contextual questions…

In case you missed it, my last post was MTTF Reliability, Cricket and Baseball:

MTTF Reliability, Cricket and Baseball:

bradman last

I originally hail from India, which means that I was eating, drinking and sleeping Cricket at least for a good part of my childhood. Growing up, I used to “get sick” and stay home when the one TV channel that we had broadcasted Cricket matches. One thing I never truly understood then was how the batting average was calculated in Cricket. The formula is straightforward:

Batting average = Total Number of Runs Scored/ Total Number of Outs

Here “out” indicates that the batsman had to stop his play because he was unable to keep his wicket. In Baseball terms, this will be similar to a strike out or a catch where the player has to leave the field. The part that I could not understand was when the Cricket batsman did not get out. The runs he scored was added to the numerator but there was no changes made to the denominator. I could not see this as a true indicator of the player’s batting average.

When I started learning about Reliability Engineering, I finally understood why the batting average calculation was bothering me. The way the batting average in Cricket is calculated is very similar to the MTTF (Mean Time To Failure) calculation. MTTF is calculated as follows:

MTTF = Total time on testing/Number of failures

For a simple example, if we were testing 10 motors for 100 hours and three of them failed at 50, 60 and 70 hours respectively, we can calculate MTTF as 293.33 hours. The problem with this is that the data is a right-censored data. This means that we still have samples where the failure has not occurred and we stopped the testing. This is similar to the case where we do not include the number of innings where the batsman did not get out. A key concept to grasp here is that the MTTF or the MTBF (Mean Time Between Failure) metric is not for a single unit. There is more to this than just saying that on average a motor is going to last 293.33 hours.

When we do reliability calculations, we should be aware whether censored data is being used and use appropriate survival analysis to make a “reliability specific statement” – we can expect that 95% of the motor population will survive x hours. Another good approach is to calculate the lower bound confidence intervals based on the MTBF. A good resource is https://www.itl.nist.gov/div898/handbook/apr/section4/apr451.htm.

Ty Cobb. Don Bradman and Sachin Tendulkar:

We can compare the batting averages in Cricket to Baseball. My understanding is that the batting average in Baseball is calculated as follows:

Batting Average = Number of Hits/Number of Bats

Here the hit can be in the form of singles, home runs etc. Apparently, this statistic was initially brought up by an English statistician Henry Chadwick. Chadwick was a keen Cricket fan.

I want to now look at the greats of Baseball and Cricket, and look at a different approach to their batting capabilities. I have chosen Ty Cobb, Don Bradman and Sachin Tendulkar for my analyses. Ty Cobb has the largest Baseball batting average in American Baseball. Don Bradman, an Australian Cricketer often called the best Cricket player ever, has the largest batting average in Test Cricket. Sachin Tendulkar, an Indian Cricketer and one of the best Cricket players of recent times, has the largest number of runs scored in Test Cricket. The batting averages of the three players are shown below:

averages

As we discussed in the last post regarding calculating reliability with Bayesian approach, we can make reliability statements in place of batting averages. Based on 4191 hits in 11420 bats, we could make a statement that – with 95% confidence Ty Cobb is 36% likely to make a hit in the next bat. We can utilize the batting average concept in Baseball to Cricket. In Cricket, hitting fifty runs is a sign of a good batsman. Bradman has hit fifty or more runs on 56 occasions in 80 innings (70%). Similarly Tendulkar has hit fifty or more runs on 125 occasions in 329 innings (38%).

We could state that with 95% confidence, Bradman was 61% likely to score fifty or more runs in the next inning. Similarly, Sachin was 34% likely to score fifty runs or more in the next inning at 95% confidence level.

Final Words:

As we discussed earlier, similar to MTTF, batting average is not a good estimation for a single inning. It is an attempt for a point estimate for reliability but we need additional information regarding this. This should not be looked at it as a single metric in isolation. We cannot expect that Don Bradman would score 99.94 runs per innings. In fact, in the last very match that Bradman played, all he had to do was score 4 single runs to achieve the immaculate batting average of 100. He had been out only 69 times and he just needed four measly runs to complete 7000 runs and even if he got out on that inning, he would have achieved the spectacular batting average of 100. He was one of the best players ever. His highest score was 334. This is called “triple century” in Cricket, and this is a rare achievement. As indicated earlier, he was 61% likely to have scored fifty runs or more in the next inning. In fact, Bradman had scored more than four runs 69 times in 79 innings.

bradman last

Everyone expected Bradman to cross the 100 mark easily. As fate would have it, Bradman scored zero runs as he was bowled out (the batsman misses and the ball hits the wicket) by the English bowler Eric Hollies, in the second ball he faced. He had hit 635 fours in his career. A four is where the batsman scores four runs by hitting the ball so that it rolls over the boundary of the field. All Bradman needed was one four to achieve the “100”. Bradman proved that to be human is to be fallible. He still remains the best that ever was and his record is far from broken. At this time, the batsman with the second best batting average is 61.87.

Always keep on learning…

In case you missed it, my last post was Reliability/Sample Size Calculation Based on Bayesian Inference:

Reliability/Sample Size Calculation Based on Bayesian Inference:

Bayesian

I have written about sample size calculations many times before. One of the most common questions a statistician is asked is “how many samples do I need – is a sample size of 30 appropriate?” The appropriate answer to such a question is always – “it depends!”

In today’s post, I have attached a spreadsheet that calculates the reliability based on Bayesian Inference. Ideally, one would want to have some confidence that the widgets being produced is x% reliable, or in other words, it is x% probable that the widget would function as intended. There is the ubiquitous 90/90 or 95/95 confidence/reliability sample size table that is used for this purpose.

90-95

In Bayesian Inference, we do not assume that the parameter (the value that we are calculating like Reliability) is fixed. In the non-Bayesian (Frequentist) world, the parameter is assumed to be fixed, and we need to take many samples of data to make an inference regarding the parameter. For example, we may flip a coin 100 times and calculate the number of heads to determine the probability of heads with the coin (if we believe it is a loaded coin). In the non-Bayesian world, we may calculate confidence intervals. The confidence interval does not provide a lot of practical value. My favorite explanation for confidence interval is with the analogy of an archer. Let’s say that the archer shot an arrow and it hit the bulls-eye. We can draw a 3” circle around this and call that as our confidence interval based on the first shot. Now let’s assume that the archer shot 99 more arrows and they all missed the bull-eye. For each shot, we drew a 3” circle around the hit resulting in 100 circles. A 95% confidence interval simply means that 95 of the circles drawn contain the first bulls-eye that we drew. In other words, if we repeated the study a lot of times, 95% of the confidence intervals calculated will contain the true parameter that we are after. This would indicate that the one study we did may or may not contain the true parameter. Compared to this, in the Bayesian world, we calculate the credible interval. This practically means that we can be 95% confident that the parameter is inside the 95% credible interval we calculated.

In the Bayesian world, we can have a prior belief and make an inference based on our prior belief. However, if your prior belief is very conservative, the Bayesian inference might make a slightly liberal inference. Similarly, if your prior belief is very liberal, the inference made will be slightly conservative. As the sample size goes up, impact of this prior belief is minimized. A common method in Bayesian inference is to use the uninformed prior. This means that we are assuming equal likelihood for all the events. For a binomial distribution we can use beta distribution to model our prior belief. We will use (1, 1) to assume the uninformed prior. This is shown below:

uniform prior

For example, if we use 59 widgets as our samples and all of them met the inspection criteria, then we can calculate the 95% lower bound credible interval as 95.13%. This is assuming the (1, 1) beta values. Now let’s say that we are very confident of the process because we have historical data. Now we can assume a stronger prior belief with the beta values as (22,1). The new prior plot is shown below:

22-1 prior

Based on this, if we had 0 rejects for the 59 samples, then the 95% lower bound credible interval is 96.37%. A slightly higher reliability is estimated based on the strong prior.

We can also calculate a very conservative case of (1, 22) where we assume very low reliability to begin with. This is shown below:

1-22 Prior

Now when we have 0 rejects with 59 samples, we are pleasantly surprised because we were expecting our reliability to be around 8-10%. The newly calculated 95% lower bound credible interval is 64.9%.

I have created a spreadsheet that you can play around with. Enter the data in the yellow cells. For a stronger prior (liberal), enter a higher a_prior value. Similarly, for a conservative prior, enter a higher b_prior value. If you are unsure, retain the (1, 1) value to have a uniform prior. The spreadsheet also calculates the maximum expected rejects per million value as well.

You can download the spreadsheet here.

I will finish with my favorite confidence interval joke.

“Excuse me, professor. Why do we always calculate 95% confidence interval and not a 94% or 96% interval?”, asked the student.

“Shut up,” explained the professor.

Always keep on learning…

In case you missed it, my last post was Mismatched Complexity and KISS:

Mismatched Complexity and KISS:

mismatch

*work-in-process*

In today’s post, I will be looking at complexity from the standpoint of organizational communication and KISS. For the purpose of this post, I am defining complexity as a measure of computational effort needed to describe your intent. This idea of complexity is loosely based on Kolmogorov’s definition of “Complexity” from an algorithm standpoint.

To give a very simple example, let’s say that I would like to convey two messages, M1 and M2:

M1 = 010101

M2 = 100111

From the complexity standpoint, M2 requires more effort to explain because there is no discerning pattern in the string of numbers. M1, on the other hand, is easier to describe. I can just say, “Repeat 01 three times.” For M2, I have no choice but say the entire string of numbers. In this regard, I could say that M2 is more complex than M1.

Let’s look at another example, M3:

M3 = 1415926535

Here, it may look like there is no discerning pattern to the string of numbers. However, this can be easily described as “first 10 decimal values of pi without 3. Thus, this message also has low complexity. We can easily see a direct linear relationship or know the content just by observation/empirical evidence.

The examples so far have been examples of low complexity messages. These are easy to generate, diffuse and convey. From the complexity standpoint, these are Simple messages. If I were to create a message that explained Einstein’s relativity, it may not be easily understood if the receiver of the message does not have a good grasp of Physics and advanced math. This is an example of medium complexity or a complicated topic. The relationship is evident with all of the information available.

Now let’s say that I would like to create a message about a complex topic – solve poverty or solve global warming. There is no evident relationship here that can be manipulated with an equation to solve the problem. These are examples of wicked problems – there are no solutions to these problems. There are options but none of the options will fully solve the many intricate problems that are entangled with each other. Such a topic is unlikely to be explained in a message.

The common thread in communication or solving problems is the emphasis on KISS (Keep It Simple Stupid). However, in an effort to keeping things simple, we often engage in mismatched complexity. Complex ideas should not be exclusively conveyed as simple statements. The ideal state is that we use the optimal message – adjust complexity of the message to match the complexity of the content. This is detailed in the schematic below. The optimal message is the 45 degree line between the two axes. A highly complex topic should not be expressed using a low complex message such as a slogan or policy statement. In a similar fashion, a low complexity topic does not need a high complexity message method such as an hour-long meeting to discuss something fundamental.

message diagram

The highly complex topic can use both low and medium message methods to ensure that the complex idea is conveyed properly. The diffusion of the highly complex topic can build upon both low and medium message methods. The diffusion of a highly complex topic also requires redundancy, which means that the message must be conveyed as many times as needed and use of metaphors and analogies. One definition of “communication” from the great Gregory Bateson is – Communication is what the receiver understands, not what the sender says.

A good example to explain this is Toyota Production System. The concept of a production system for the entire plant is a complex concept. Toyota Production System was once called “the Ohno method” since it was not implemented company-wide and there was doubt as to the success of the system being a long-term plan. Ohno’s message was not written down anywhere and the employees did not learn that from a manual or a video. Ohno conveyed his ideas by being at the gemba (actual work place), implementing ideas and learning from them. He also developed employees by constantly challenging them to find a better way with less. Ohno used to draw a chalk circle on the floor for supervisors/engineers to make them see what he saw. He developed the Toyota Production System and with continuous mentoring, nurtured it together with the employees. Today there are over 1000 books at Amazon regarding “Lean Manufacturing”. When top management is looking at implementing lean, the message should match the complexity of the content. Low complex message methods like slogans or placards will not work. Medium complex message methods like newsletters, books etc will not work. This will require constant on-the-floor interactive mentoring. At the same time, slogans and newsletters can aid in the diffusion process.

Final Words:

I have always felt that KISS and Poka-Yoke have a similar story to tell from a respect-for-people standpoint. Poka-Yoke (Error proofing) was initially termed as Baka-Yoke to indicate “fool proofing”. Shigeo Shingo changed it to Poke-Yoke to indicate error proofing after an employee asked him “have I been such a fool?” In a similar fashion, KISS was initially put forth as “Keep It Simple Stupid” (without the comma). Nowadays, this has been changed to “Keep It Short and Simple” and “Keep It Simple Straightforward”.

It is good to keep things simple and to view at a problem from a 10,000 feet level. However, we should not stop there. We need to understand the context and complexity of the problem and then create this information in such a manner that it can be diffused across the organization. This can be repeated as many times as needed. Do not insist on simplicity without understanding the complexity of the problem. Asking to keep things simple is an attempt to keep round pegs in familiar square holes. When there is a mismatch of complexity it leads to incorrect solutions and setbacks. As Einstein may have said,everything should be as simple as it can, but not simpler”.

We can also view the complexity/message diagram in the light of the Feynman (Nobel-prize winning physicist Richard Feynman) technique of studying hard subjects. Feynman came up with a method where he would start studying and making notes pretending to prepare a lecture for a class. He would use simple terms and analogies to explain the subject. When he got stuck he would go back and try to understand it even better. He would then proceed with making notes. He would repeat the steps many times until he got the concept thoroughly. Moving from High to Medium to Low in the diagram, and going back-and-forth helps to connect the dots and gain a better understanding.

I will finish with another quote, attributed to Lotfi Zadeh (father of Fuzzy Logic):

“As complexity rises, precise statements lose meaning and meaningful statements lose precision.”

Always keep on learning…

In case you missed it, my last post was Flat Earth Lean:

Flat Earth Lean:

pipe

How many interpreters does it take to change a light bulb?

It depends on the context!

In today’s post, I will be looking at what I call “Flat Earth Lean” and “Contextual Lean”. I recently came across the concept of “Flat Earth View” in organizational communication. Matthew Koschmann, currently an associate professor at the University of Colorado, talks about the one-dimensional approach to organization communication where the big picture is not used. It is a linear approach without looking at the contexts or the social aspects. Koschmann explains – What I mean by a flat earth approach is a perspective that seems correct from a limited vantage point because it works for much of our day to day lives, but ultimately it fails to account for the complexity of a situation. For much of human history we got by just fine thinking the earth was flat, even though it was always round. And even with our 21st century sophistication where we know the earth is round, most of us can actually get by with flat earth assumptions much of the time. But what about when things get more complex? If you want to put a satellite into space or take a transcontinental flight, flat earth assumptions are not going to be very helpful. Remember in elementary school when you compared a globe to a map and realized, for example, that it s quicker to fly from New York to Moscow by flying over the North Pole instead of across the Atlantic? What seems counter intuitive from a flat earth perspective actually makes perfect sense from a round earth perspective.”

I would like to draw an analogy to Lean. Perhaps, the concept of flat earth exists in Lean as well. This could be looked at as the tools approach or copying Toyota’s solutions to apply them blindly. The linear approach implies a direct cause and effect relationship. From the Complexity Science standpoint, the linear relationship makes sense only in the simple and complicated domains. This is the view that everything is mechanistic, utilizing the metaphor of a machine – press this button here to make something happen on the other side with no unintended consequence or adverse effects. In this world, things are thought to be predictable, they can be standardized with one-glove-fits-all solutions, and every part is easily replaceable. Such a view is very simplistic and normally cares only about efficiency. This is an approach that is used for technical systems. There is limited or no focus on context. Hajime Ohba, a Toyota veteran, used to say that simply copying Toyota’s methods is like creating the image of Buddha and forgetting to inject soul in it. In Flat Earth Lean, the assumption is that end goal is clearly visible and that it is as easy as going from HERE to THERE. The insistence is always to KISS (keep it simple stupid). In many regards, this reductionist approach was working in the past. Information generation was minimal and the created information was kept local in the hands of the experts. In today’s global economy, organizations do not have the leisure to keep using the reductionist approach. Today, organizations not only have to ensure that information is diffused properly, they also have to rely on their employees to generate new information on a frequent basis. The focus needs to be shifted to organizations being socio-technical systems where things are not entirely predictable.

Here to There

Karl Weick, an American organizational theorist, advises to “complicate yourself”. He cautions us to not rely on oversimplification. We need to understand the context of what we are doing, and then challenge our assumptions. We have to look for contradictions and paradoxes. They are the golden nuggets that help us to understand our systems. In Contextual Lean, we have to understand our problems first and then look for ways to make things better. Implementing 5S with the aim of being “Lean” is the Flat Earth Approach. Implementing 5S and other visualization methods to make sense of our world, and making problems visible so that we can address them is “Contextual Lean”. If there is such a thing as “going Lean” for an organization, it is surely a collective expression. “Lean” does not exist in isolation in a department or in a cabinet; let alone in one Manager or an employee. To paraphrase the great philosopher, Ludwig Wittgenstein, the meaning of an expression exists only in context. Context gives meaning. Toyota’s “Lean” has limited meaning in relation to your organization since it makes sense only in the context of the problems that Toyota has. Thus, when the Top Management pushes for Lean initiation, it has to be in the context of the problems that the organization has. Understanding context requires self-reflection and continuous learning for the organization. This again is a collective expression and does not exist without involving the employees. Interestingly, Contextual Lean has to utilize Flat Earth approach as needed.

Flat Earth and Contextual Lean have some similarities to the late American business theorist Chris Argyris’ ideas of Single and Double Loop learning. Single Loop learning is the concept of correcting an error by using the existing mental models, norms and practices. Argyris gives the example of a thermostat to explain this – Single loop learning can be compared with a thermostat that learns when it is too hot or too cold and then turns the heat on or off. The thermostat is able to perform this task because it can receive information (the temperature of the room) and therefore take corrective action. Double Loop Learning, on the other hand, involves a reflective phase that challenges the existing mental models, norms and practices, and modifies them to correct the error. In Chris Argyris’ words –If the thermostat could question itself about whether it should be set at 68 degrees, it would be capable not only of detecting error but of questioning the underlying policies and goals as well as its own program. That is a second and more comprehensive inquiry; hence it might be called double loop learning. Single Loop Learning has some similarities to Flat Earth Lean in that it wants to take a simplistic approach and does not want to modify the mental models. It wants to keep doing what is told and to use an old analogy – only bring your hands to work and leave your brains outside. Single Loop Learning is a superficial approach to solve problems symptomatically. Double Loop Learning has some similarities to Contextual Lean in that it is not one-dimensional and results in modifying the mental models as needed. It is a continuous learning and adapting cycle. Argyris also believed that organizations learn when its people learn – Organizational learning occurs when individuals, acting from their times and maps, detect a match or mismatch of outcome to expectation which confirms or disconfirms organizational theory-in-use.

I will finish with a fitting contextual story about change.

Mulla Nasrudhin was now an old man. People used to gather around to hear him talk. One day a young man asked for some words of wisdom.

Mulla replied, “When I was young I was very strong minded- I wanted to awaken everyone. I prayed to God to give me the strength to change the world. As time went on, I became middle aged and I realized that I did not change the world. Then I prayed to God to give me strength so that I can at least change those close around me. Now that I am older and perhaps wiser, my prayer has become simpler. I say – God, please grant me the strength to change at least myself.”

Always keep on learning…

In case you missed it, my last post was The Purpose of Visualization:

The Purpose of Visualization:

1845

Many men go fishing all of their lives without knowing that it is not the fish they are after.” – a quote misattributed to Henry David Thoreau.

What is the purpose of visualization? Before answering this, let’s look at what is visualization. Visualization is making information visible at the gemba. The information could be in the form of daily production boards or it could be non-conforming components or other artifacts placed on a table on the production floor. Another phrase that is used in place of visualization is “visibilization”. I had talked about this in the post – Visibilization: Crime Fighting, Magic and Mieruka. The purpose of visualization or visibilization is to make waste visible so that appropriate action can be pursued. Or is it?

I recently came across the paper “Defining Insight for Visual Analytics” by Chang, Ziemkiewicz et al. I enjoyed the several insights I was able to gain from this paper. The purpose of visualization is to enable and discover insight. This may seem fairly logical and straightforward. Chang et al. details that there are two types of insights – knowledge building insight and spontaneous insight. The knowledge building insight is a linear continuous process where the operator can use established problem solving methods and heuristics to solve a problem and gain insight into the process. The spontaneous insight does not come from gradual learning heuristics or problem solving methods. The spontaneous insight results in “aha!” moments and usually new knowledge. The spontaneous insight often occurs when the operator has tried the normal problem solving routines without success. The spontaneous insight happens in frustration after several attempts when the mind breaks off from normal routines. Researchers are able to study the two insights by using electroencephalogram (EEG) and functional magnetic resonance imaging (fMRI) on the participants’ brains.

Chang et al. notes that – In normal problem solving, the activity in the temporal lobe is continuous and mostly localized in the left hemisphere, which is thought to encode more detailed information in tightly related semantic networks. This indicates that normal problem solving involves a narrow but continuous focus on information that is highly relevant to the problem at hand. In contrast, when participants solve a problem with spontaneous insight, the right temporal lobe shows a sharp burst of activity, specifically in an area known as the superior temporal gyrus. Unlike the left temporal lobe, the right temporal lobe is thought to encode information in coarse, loosely associated semantic networks. This suggests that spontaneous insight occurs through sudden activation of less clearly relevant information through weak semantic networks, which corresponds to a participant’s paradigm shift following an impasse.

The findings indicate that the spontaneous insight is qualitatively different from the knowledge building insight. The knowledge building insight is using the normal routines and increasing the existing knowledge, while the spontaneous insight is breaking away from the normal routines and creating new knowledge. Spontaneous insight is a form of problem solving that is used to find solutions to difficult and seemingly incomprehensible problems. Knowledge-building insight, on the other hand, is a form learning that builds a relationally semantic knowledge base through a variety of problem-solving and reasoning heuristics.

In the light of the two insights, which one is better? The point is not to identify what is better, but to understand that both types of insights are important and are both related to one another. Chang et al. theorizes that one can only gain spontaneous insights only from routine knowledge building insights. In their words – Einstein didn’t come up with the Theory of Relativity out of thin air but rather based it on experiments inconsistent with existing theories and previous mathematical work. The existence of deep, complex knowledge about a subject increases the likelihood that a novel connection can be made within that knowledge. Likewise, each major spontaneous insight opens up the possibility of new directions for knowledge-building. Together, the two types of insights support each other in a loop that allows human learning to be both flexible and scalable.

Chang et al. hypothesizes that there is a positive non-linear relationship between gaining insights and the knowledge that the operator already possesses. The more knowledge the operator has, the more likelihood that the operator will gain further insights with visualization. In this light, the purpose of visualization is to develop your employees, and in some regards demonstrates respect for people. Making the problems/waste visible allows them to engage in daily/ frequent problem solving routines that builds knowledge building insights, which then leads to spontaneous insights to improving their processes. In other words, it is about building the continuous improvement muscle! The problems on the floor can vary in their complexities. There can be routine problems with known linear relationships (simple to complicated problems), and there can be problems where there are no known solutions and are intricately woven with non-linear relationships (complicated to complex problems). Solving the routine problems can help with gaining valuable spontaneous insights to tackle the complex problems.

I will finish off with a quote from the great Carl Sagan when he went on The Tonight Show with Johnny Carson:

For most of history of life on this planet, almost all the information they had to deal with was in their (organisms’) genes.  Then about 100 million years ago or a little longer than that, there came to be a reptile that for the first time in the history of life had more information in its brains than in its genes. That was a major step symbolically in the evolution of life on this planet. Well, now we have an organism – us, which can store more information outside the body altogether than inside the body – that is in books and computers and television and video cassettes. And that extraordinarily expands our abilities to understand what is happening and to manipulate and control our environment, if we do it wisely, for human benefit.

Always keep on learning…

In case you missed it, my last post was Looking at Kaizen and Kaikaku:

Looking at Kaizen and Kaikaku:

trotoise-hare

In today’s post, I will be looking at the “Kaizen” and “Kaikaku” in light of the Explore/Exploit model. Kaizen is often translated from Japanese as “continuous improvement” or “change for better”. “Kaikaku”, another Japanese term, is translated as “radical change/improvement”. “Kakushin” is another Japanese word that is used synonymously with “Kaikaku”. “Kakushin” means “innovation” in Japanese. Kaikaku got more attention from Lean practitioners when the previous Toyota President and CEO, Katsuaki Watnabe said  in 2007- Toyota could achieve its goals through Kaizen. In today’s world, however, when the rate of change is too slow, we have no choice but to resort to drastic changes or reform: Kaikaku

The explore/exploit model is based on a famous mathematical problem. I will use the example from Brian Christian and Tom Griffith’s wonderful 2016 book “Algorithms to Live By: The Computer Science of Human Decisions”. Let’s say that you are very hungry and do not feel like cooking. Which restaurant should you go to? Your favorite Italian restaurant or the new Thai place that just opened up? Would your decision capabilities be impacted if you are traveling? Sticking with what you know and being safe is the “exploit” model. Trying out new things and taking risks is the “explore” model. The dilemma comes because you have to choose between the two. The optimal solution depends on how much time you have on your hands. If you are traveling and you are at a new place for two weeks, you should try out different things at the beginning (explore). As days go by and you only have a few more days left, you should definitely stick with what you know to be the best choice so far (exploit). Christian and Griffith stated in the book – Simply put, exploration is gathering information, and exploitation is using the information you have to get a known good result.

From an organization’s standpoint, the explore/exploit dilemma is very important. The exploit model is where the organization continues to focus on efficiency and discipline in what they already manufacture. The explore model on the other hand, is focusing on innovation and new grounds. The exploit model does not like risk and uncertainty. The exploit model does not necessarily mean maintaining status-quo or not rocking the boat. The exploit model is getting better at what you already do. One way that I have heard the differentiation between the two explained is like this – exploitation is like playing in the same sandbox and getting better at the games you play inside the sandbox. Exploration is like venturing outside of your sandbox and finding new sandboxes to play with and creating new games.

Some strategies used for the exploit model are:

  • Optimize the organization for current organizational rules and structure
  • Make sure standards are in place and the established rules are followed in order to achieve efficiency
  • Make incremental improvements for existing processes better and still stay within the current organizational structures
  • Keep making more of the current product portfolio

The explore model is about breaking new grounds. Some strategies used for the explore model are:

  • Break away from the current organizational rules and structure
  • Develop new structures to allow for diversity and discovery
  • Make radical improvements to overhaul current processes, rules and structures
  • Add new product portfolios altogether

The exploit model relies on current constraints, rules and structures. The exploration model relies on the willingness to break away from the current constraints, rules and structures. A perfect balance between the two models and oscillating between both models or engaging in both models simultaneously is very important for an organization to thrive. The organizations that can do both are called “ambidextrous”.

The explore/exploit model has some similarities to Kaizen and Kaikaku. Kaizen is about getting better at what we do incrementally. It is a personal development model. Kaikaku, on the other hand, is about breaking the mold and overhauling the organization in some cases. Launching a Lean initiative can be viewed as Kaikaku. Kaizen could be an ideal strategy for exploitation and Kaikaku for exploration. I came across a paper from Yuji Yamamoto called “Kaikaku in Production in Japan: An Analysis of Kaikaku in Terms of Ambidexterity” that further shed light on this. The paper is part of the collection called “Innovative Quality Improvements in Operations”. Yamamoto points out that while Kaizen is incremental; Kaikaku entails large-scale changes to both the social and technical systems of an organization. Kaizen is often viewed as an opportunity and Kaikaku may sometimes be viewed as a necessity. Kaizen is also viewed as a bottom-up activity with autonomy, and Kaikaku on the other hand can be viewed as top-down activity with direction from the top management. Kaikaku may be continual (with definite timelines and stops) and Kaizen is continuous. Kaizen is described as engaging everybody in improvement every day, everywhere in the organization.

Yamamoto discussed data from 65 case studies where Kaikaku activities were implemented at Japanese manufacturing companies. Yamamoto noted that the defining characteristic for Kaikaku based on the studies was that Kaikaku requires everybody’s exploration effort. In the 65 reports, the importance of everyone in the organization having a specific mental mode related to exploration, for instance, a challenging spirit, give-it-a-try mentality, and unlearning, is frequently mentioned. In the Kaikaku activities, managers often encouraged everyone in the organizations to think and act in a more explorative way than they were used to. Apparently, companies used the word Kaikaku as a way to make managers and employees be aware of this mental stance toward exploration.

Yamamoto used the exploit/explore model to further differentiate Kaizen and Kaikaku. The figure below is adapted from Yamamoto. The figure shows different degrees of exploitation and exploration activities. Problem solving with a high degree of innovativeness tends to involve more exploration than exploitation.

K and K

Some key takeaways from Yamamoto’s paper are:

  • Kaikaku and Kaizen are complementary and reinforce each other. Effective Kaizen often has a positive influence on Kaikaku, and Kaikaku can stimulate Kaizen.
  • Employees engaged in iterative problem solving activities in Kaizen and Kaikaku develop exploitation and exploration abilities as part of a learning cycle. The beginning of this learning cycle is about making problems and challenges visible to increase the sense of urgency. Once they are resolved, the results are made visible throughout the organization. The organizations in the case studies created an environment for keeping the learning cycle going with opportunities to engage in improvement and innovation.
  • The participants of Kaikaku activities reflect on and learn from their successes and failures. They achieve a sense of achievement and are motivated to tackle challenges that are even more difficult.
  • Problem solving activities often lead to identifying further improvement opportunities.
  • Some companies in the report used Kaikaku to enhance Kaizen because Kaizen had been slow and reactive. While some other companies initiated Kaikaku to make employees more competent in innovation.

I will end with a Zen quote with focus on when we should be doing more:

You should sit in meditation for 20 minutes a day, unless you are too busy. In that case, you should meditate for an hour a day.

Always keep on learning…

In case you missed it, my last post was Hammurabi, Hawaii and Icarus:

Hammurabi, Hawaii and Icarus:

patent

In today’s post, I will be looking at Human Error. In November 2017, The US state of Hawaii reinstated the Cold War era nuclear warning signs due to the growing fears of a nuclear attack from North Korea. On January 13, 2018, an employee from the Hawaii Emergency Management Agency sent out an alert through the communication system – “BALLISTIC MISSILE THREAT INBOUND TO HAWAII. SEEK IMMEDIATE SHELTER. THIS IS NOT A DRILL.” The employee was supposed to take part in a drill where the emergency missile warning system is tested. The alert message was not supposed to go to the general public. The cause for the mishap was soon determined to be human error. The employee in the spotlight and few others left the agency soon afterwards. Even the Hawaiian governor, David Ige, came under scrutiny because he had forgotten his Twitter password and could not update his Twitter feed about the false alarm. I do not have all of the facts for this event, and it would not be right of me to determine what went wrong. Instead, I will focus on the topic of human error.

One of the first proponents of the concept of human error in the modern times is the American Industry Safety pioneer, Herbert William Heinrich. In his seminal 1931 book, Industrial Accident Prevention, he proposed the concept of Domino theory to explain industry accidents. Heinrich reviewed several industrial accidents of his time, and came up with the following percentages for proximate causes:

  • 88% are from unsafe acts of persons (human error),
  • 10% are from unsafe mechanical or physical conditions, and
  • 2% are “acts of God” and unpreventable.

The reader may find it interesting to learn that Heinrich was working as the Assistant Superintendent of the Engineering and Inspection Division of Travelers Insurance Company, when we wrote the book in 1931. The data that Heinrich collected was somehow lost after the book was published. Heinrich’s domino theory explains an injury from an accident as a linear sequence of events associated with five factors – Ancestry and social environment, Fault of person, Unsafe act and/or mechanical or Unsafe performance of persons, Accident and Injury.

H1

He hypothesized that taking away one domino from the chain can prevent the industrial injury from happening. He wrote – If one single factor of the entire sequence is to be selected as the most important, it would undoubtedly be the one indicated by the unsafe act of the person or the existing mechanical hazard. I was taken aback by the example he gave to illustrate his point. As an example, he talked about an operator fracturing his skull as the result of a fall from a ladder. The investigation revealed that the operator descended the ladder with his back to it and caught his heel on one of the upper rungs. Heinrich noted that the effort to train and instruct him and to supervise his work was not effective enough to prevent this unsafe practice.  “Further inquiry also indicated that his social environment was conducive to the forming of unsafe habits and that his family record was such as to justify the belief that reckless tendencies had been inherited.

One of the main criticisms to Heinrich’s Domino model is its simplistic nature to explain a complex phenomenon. The Domino model is reflective of the mechanistic view prevalent at that time. The modern view of “human error” is based on cognitive psychology and systems thinking. In this view, accidents are seen as a by-product of the normal functioning of the sociotechnical system. Human error is seen as a symptom and not a cause. This new view uses the approach of “no-view” when it comes to human error. This means that the human error should not be its own category for a root cause. The process is not perfectly built, and the human variability that might result in a failure is the same that results in the ongoing success of the process. The operator has to adapt to meet the unexpected challenges, pressures and demands that arise on a day-to-day basis. The use of human error as a root cause is a fundamental attribution error – focusing on the human trait of the operator as being reckless or careless; rather than focusing on the situation that the operator was in.

One concept that may help in explaining this further is Local Rationality. Local Rationality starts with the basic assumption that everybody wants to do a good job, and we try to do the best (be rational) with the information that is available to us at a given time. If this decision led to an error, instead of looking at where the operator went wrong, we need to look at why he made the decisions that made sense to him at that point in time. The operator is in the “sharp end” of the system. James Reason, Professor Emeritus of Psychology at the University of Manchester in England, came up with the concept of Sharp End and Blunt End. Sharp end is similar to the concept of Gemba in Lean, where the actual action is taking place. This is mainly where the accident happens and is thus in the spotlight during an investigation. Blunt end, on the other hand, is removed and away in space and time. The blunt end is responsible for the policies and constraints that shape the situation for the sharp end. The blunt end consists of top management, regulators, administrators etc. Professor Reason noted that the blunt end of the system controls the resources and constraints that confront the practitioner at the sharp end, shaping and presenting sometimes conflicting incentives and demands. The operators in the sharp end of the sociotechnical system inherits the defects in the system due to the actions and policies set by blunt end and can be the last line of defense instead of being the main proponents or instigators of the accidents. Professor Reason also noted that – rather than being the main instigators of an accident, operators tend to be the inheritors of system defects. Their part is that of adding the final garnish to a lethal brew whose ingredients have already been long in the cooking. I encourage the reader to research the works of Jens Rasmussen, James Reason, Erik Hollnagel and Sydney Dekker since I have tried to only scratch the surface.

Final Words:

Perhaps the oldest source of human error causation is the Code of Hammurabi, the code of ancient Mesopotamian laws dating back to 1754 BC. The Code of Hammurabi consisted of 282 laws. Some examples of human error are given below.

  • If a builder builds a house for someone, and does not construct it properly, and the house which he built falls in and kill its owner, then that builder shall be put to death.
  • If a man rents his boat to a sailor, and the sailor is careless, and the boat is wrecked or goes aground, the sailor shall give the owner of the boat another boat as compensation.
  • If a man lets in water and the water overflows the plantation of his neighbor, he shall pay ten gur of corn for every ten gan of land.

I will finish off with the story of Icarus. In Greek mythology, Icarus was the creator of the labyrinth in the island of Minos. Icarus’ father was the master craftsman Daedalus. King Minos of Crete imprisoned Daedalus and Icarus in Crete. The ingenious Daedalus observed the birds flying and invented a set of wings made from bird feathers and candle wax. He tested the wings out and made a pair for his son Icarus. Daedalus and Icarus planned their escape. Daedalus was a good Engineer since he studied the failure modes of his design and identified the limits. Daedalus instructed Icarus to follow him closely and asked him to not fly too close to the sea since the moisture can dampen the wings, and not fly too close to the sun since the heat from sun can melt the wings. As the story goes, Icarus was excited with his ability to fly and got carried away (maybe reckless). He flew too close to the sun, and the wax melted from his wings causing him to fall down to his untimely death.

Perhaps, the death of Icarus could be viewed as a human error since he was reckless and did not follow directions. However, Stephen Barlay in his 1969 book, Aircrash Detective: International Report on the Quest for Air Safety, looked at this story closely. At the high altitude that Icarus was flying, the temperature will actually be cold rather than warm. Thus, the failure would actually be from the cold temperature that would make the wax brittle and break instead of wax melting as indicated in the story. If this was true, during cold weathers the wings would have broken down and Icarus would have died at another time even if he had followed his father’s advice.

Always keep on learning…

In case you missed it, my last post was A Fuzzy 2018 Wish