In the earlier post we dealt with the concept of predicting a ‘black swan‘.
In this post, I intend to explore the concept a bit more: what exactly we monitor to notice a ‘black swan’ in time?
In doing so we would be forced to consider the natural response of a system.
The starting point of our exploration would be to understand how any system, as a whole, whether natural or engineered, would disturbed by a ‘black swan’. A system is disturbed in three possible ways, which are as follows:
a) A system loses energy till it reaches a tipping point
b) A system gains more and more energy till it crosses the point of system resilience
c) A part of a system emits more energy than it is normally supposed to, that is going beyond the linear response of the part.
So the natural way to watch a system to expect a ‘black swan’ in time, is to keep a tab on the ‘energy’ of a system in the following ways:
a) Monitor the entropy of a system. As a system functions the entropy of a system gradually rises till it hits a threshold limit indicating the appearance of a ‘black swan’ or an outlier.
b) Monitor the energy gain of a system till it crosses the ‘resilience’ point to give birth to a ‘black swan’, outlier or a ‘wicked problem’.
c) Monitor critical parts of a system for excess emission of energy till it goes beyond the linear response of a part.
It is useful to remember that energy is transferred in ‘quanta‘ or in packets of energy. Therefore, it is natural to expect jumps of energy levels as we record by capturing the different manifestation of energy levels on monitoring trend charts. So when a ‘jump’ is big enough to cross a threshold limit or resilience point or linear response level indicated by its presence outside the Gaussian distribution range we can be quite sure that a ‘black swan’ or an outlier or a ‘wicked problem’ would soon arrive on the scene. We call such an indicator as a signal.
Therefore, the central idea is to capture such signals in time, just before a ‘black swan’ makes it way to appear on the scene to dominate and change the system.
However, the question is how early can we detect that signal to effectively deal with the inherent ‘black swan’ in a system, which is yet to appear on the scene?
Few days back I submitted a written consulting report to an owner of an NGO. This was to improve productivity of the process of making natural eco-friendly fertilizers. With rising costs he was increasingly finding it difficult to sustain the ‘bottom of the pyramid’ business. So improving productivity was his only means to not only save his business but also to save his customers – the poor farmers. The farmers use this eco-friendly product to improve farm productivity and also feed their live stock.
So the project was important to me. If I can help him improve productivity then I would also be able to contribute to sustainable and resilient rural communities.
I visited them. And clinically went through the present process. Then I wrote my ideas and recommendations in a written report and emailed it.
I quote the relevant extracts from the report:
Quote
A) Reduce the size of the batch being processed in the initial stage.
Present:
At present a large quantity of raw material is processed in one go. It is slow, time taking and perhaps non-uniform. It suffers from all the 3 Mudas.
Suggested:
Break it up into smaller chunks.
For example a single batch may contain just the right amount of material that can be processed into finished product in a day. For instance if only 1T of finished product can be processed in a day then the batch sizes would be of 1T only.
A 1. Stagger
Present:
The whole batch is processed in one go. That is the whole batch is left for fermentation for the next 7 to 10 days depending on the prevailing season of the year before it is taken out for further sorting and processing. It suffers from all the 3 Mudas.
Suggested:
Since it takes 7 to 10 days (varies between summer and winter) for the raw meal to be ready for further processing. The batches can be planned in such a way depending on the season of the year — where preparation and production are in sync and it can start after the first batch is ready. It means the first batch is taken out after the 7th day and the next batch (prepared a day later) is taken out on the 8th day from the start of the production and so on…
The preparation is done in a separate area. This is followed by storing for fermentation which is carried out in a separate area in plastic drums.
Suggested
Combine the two processes.
Prepare the raw meal in the fermentation drum directly. It combines two activities in one stroke.
A suitable size of drum (used oil drums cut into halves) may be used for the purpose. The drums may be designed like the boiler drum available in the xxx office which would allow for stirring and mixing as well as act as air tight drums, which is necessary for fermentation.
The sprinkling of the chemicals that is required during preparation of the raw meal might be carried out in the way mosquito repellents are sprayed with the drum placed on the back of the sprayer. Similar drum containing the chemical can be strapped onto the back of the operator who sprays the required amount of chemical into the raw meal.
There is a need to constantly stir the raw meal during preparation. This can be done manually by means of a gear drive made of plastic or wood (so that heavy lubrication is avoided, which may spill into the raw meal – thus contaminating it).
After the meal is prepared the drums can be shut tight (air tight) by a similar mechanism used for the xxx boiler drum.
The same drum can be placed on a titling mechanism (similar to platform used for industrial gases) which can be lowered when needed to the required height for packing and weighing.
Such a jugaard (half drum, detachable and movable cover, gear driven manual stirrer, chemical spray dispenser, air tight mechanism, mounting on a tilting table like the way industrial gas cylinders are mounted) might be created for processing smaller batches.
C) Filling up the packets by calibrated mugs.
Present
At present the operator fills up the packets by haunch/feel. Then the right amount is adjusted after weighment. This is a trail and error method that entails a loss of productive time.
Suggested
The material can be filled by help of calibrated mugs that would contain either 1 kg of raw meal (for 1 kg packs) or 2.5 kg of raw meal (for 5 kg packs). This improves accuracy, speed of operation and minimizes time losses from the system.
Presently the sequence of the processing is bit haphazard. This cause time loss and motion loss.
Suggested
The entire process may be laid out in a orderly and logical sequence to eliminate motion loss, time loss and loss due to unevenness of loading.
The floor can be raised and leveled for that purpose complete with suitable and reliable electrical connections wherever needed. This would also prevent inundation of the work place by rain water during monsoon.
Unquote
I am sure you found it difficult to follow what I recommended.
The owner, though he knew about the process, also found it difficult to follow the report. He called me up to say, ‘Hey, my men and I did not quite get what you want us to do in your report. Is it possible for you to come down and explain? I would be obliged if you may.”
Coming from a man who passed out from one of the best MBA schools in the world rang a bell in my mind. So if he can’t get it surely others might also find it difficult. And clearly there was any story here.
However, I wanted to avoid a visit (I never love physical travel) and thought what else might I do to make it clear to him and his men.What followed was a pictorial representation of what I wrote, which is as follows:
After he got the pictorial view of the story I called him up to ask how he liked it. “Simply brilliant story!’ he replied.
Note:
I thank my colleagues Rick De and T R Khan for creating this moving visual story.
Here is one of many toys I use in my classes on Leadership in Complexity to demonstrate complexity through play. It is a simple and common toy – a double pendulum. It is interesting to see how interactions between few elements really produce complexity. So, the question that I ask at the beginning of a session – ‘Can we predict what is going to happen?‘
We have made a video demonstration of it. It is about 5 mins. Hope you would find it engaging. You may choose to skip it if you like. I suggest a try. While you are viewing it mentally start predicting what might happen the next instant…
On the face of it it appears that it isn’t predictable at all. The movements of the loose limbs of the double pendulum simply go crazy. It is not or nearly not possible to predict. Every time we think something like this might happen it usually turns out to be something else. It appears that there are no definite patterns about it. It is too random to make sense. No doubt this is what always happens in complex adaptive systems.
But then I show how complexity can be predicted along with many of its principles.
At first it feels rather strange to realize how all complex systems or complex adaptive systems are inherently predictable as an ensemble in the short run and how they all follow the same rules of the game.
That is really fascinating. It gives us tremendous hope to embrace complexity with faith. There is no point in ignoring complexity since we are entangled with it every moment of our lives. But once we embrace it knowing fully well how to read, learn and go about it — life is simple indeed. The objective of learning about complexity and applying its principles is to make life simpler; not more complex.
That promises us an alternative way to lead our own lives through creativity and adaptation.
This alternative Leadership path can be summed up by three simple rules, which are —
1. Explain what is happening.
2. Institute methods to Foresee what might happen in the short term
3. Envision desired Interventions to make the system flow in the right direction.
Three of the best designed interventions that I found are a) Education b) Interactions c) Design. These give long term ongoing benefit for many.
So what do you feel and think about it?
Acknowledgement:
(I personally thank my colleague Trichur for prototyping complexity through this model. )
Here is a beautiful story of a Complex Creative System. Venice is beautiful. But what lies hidden is still more beautiful. The constantly unfolding story of interactions between the sea, weather, people, houses, ideas and actions to adapt to changing situations would delight and inspire anyone to live better. The city and its people are constantly creating and re-creating their destiny in such clever ingenious ways. It is a fascinating but continuing story that is being told for centuries.
Complexity is always beautiful as it is infinitely creative. It constantly creates new situations and therefore demands us to be aware of what is happening all around compelling us to be insanely creative too. Else we would soon be out of step with the system’s creativity. Going out of sync with system’s creativity spells disaster. Systems would never listen to us; we need to listen to systems.
However, the fundamental process of being and remaining creative is to follow the principles of 1) Design/Redesign 2) Maintain 3) Destroy what is unnecessary.
There is no harm embracing complexity. In fact, that is precisely the story of our human survival through intelligent adaptation. Nothing more. Nothing less.
Recently, I have been playing with a Traffic model (basic) on Netlogo and thinking as to what we might learn from such a self organizing event in order to adapt and survive in a better way. In this post, I would like to explore whether the lessons of adaptation gained from this simple model can be applied to a broad base of applications covering different fields.
A general screenshot of the game/model is given below.
Learning from Traffic Jams
What I want to do is to show how the “RGBwaves” concept, the central theme of Nemetics (1), is applied to obtain deeper understanding of patterns of the events (Red Wave). Then I would like to examine the reason that drives such behavior (Green Wave). Having done that I would then like to explore the nature of the collective intelligence (Blue Wave) and how to change that for better adaptation in complex environments. Once done, we would then try to see whether the insights along with the wisdom this model (2) offers can be usefully applied to different fields of study in some practical manner. Along with this I would also explore what precisely needs to be changed or adapted. In short, we would like to explore enough to get to the bottom of things enabling effective adaptation.
It might be useful to notice that the R (Red), G (Green) & B (Blue) waves are interconnected and interdependent, i.e. one affects the other. Hence a change of modulation or adaptation in any wave can bring about effective changes or results. However, deciding about which wave(s) to tweak and by how much (adaptation) to bring about long-term on-going collective benefits would of course depend on the nature and character of the situation we face.
However, one need not be overwhelmed by the RGB waves and how to get around them. The whole concept of RGB waves is carefully embedded in the following NEME poster that starts with the observation of the R wave at the ‘Notice’ stage and gently leads us to the B wave at the ‘Exchange’ stage. We would then explore events through the NEME process (or the Nemetic process as it is called) as depicted below.
The Nemetic Process
A Brief understanding of the Nemetic Process
In the Notice stage we actually observe the R wave – that is its ’emergence’, ‘movement’ and ‘shape’.
In the Engage stage we observe the objects and the fields they create through their interconnections and interdependence.
At the Mull stage we find the ‘rules’ that govern the behavior of the individual and the group. We understand the given constraints under which the whole system operates and then use models to explore through mathematical visualization different possibilities that emerge to help us get to the bottom of things. For this we look to the G wave.
Finally, the Exchange stage surfaces our crucial concern for adaptation that is done through tweaks of different waves or modulation of the waves through the process of redesign, which is iterative in nature. This is done against authentic constraints by involving the stakeholders involved to co-create solutions. Usually the best solutions are obtained by leveraging the most crucial part of a phenomenon where care is taken to keep both effort and costs to the bare minimum. The idea is to effectively resolve a given paradox to obtain on-going benefits. This involves both exploration and exploitation of the B wave (collective intelligence). In case of our traffic jam problem the paradox is – the system always moves forward but the jam grows longer backwards. The same paradox can be extended to in the case of a developing economy — as an economy grows poverty grows too.
The Experiment
In this case we are trying to notice ‘traffic jam’ as emergence. The movement that we are noticing is the flow of traffic along with its speed, acceleration, deceleration and the number of cars on the road. In terms of movement we also notice whether the movement approaches a fixed outcome (fixed attractor), or is the movement periodic (sync attractor) or quasi-periodic (a torus attractor) or something quite strange is happening (strange attractor). We also notice the shape, which in this case turns out to be a tube (3) containing the stretch of road (a highway), cars which are in some way coupled to one another through invisible strings. In this case we consider that this tube is not connected to other tubes operating in the same space and time. And the issue that we are about to observe is Spacio-Temporal in nature i.e. the event is playing out in both Space and Time, which makes things complex indeed calling for instant moment to moment adaptation.
Let us start the exploration through modeling. In this case we have a fixed number of cars running on the road against time. We can of course change the number of cars plying the road during the experiment to see whether that affects the outcome. Our car of interest is the red car. The rule is simple. If there is a clear stretch of road ahead we accelerate else we decelerate if the space between cars is insufficient to maintain speed. So, the driver takes the decision to either accelerate or decelerate depending on the situation.
In this experiment the variables I have tuned are: a) acceleration b) deceleration c) number of cars on the road. Given below is a representative sample set of the outcomes that might be just enough for the critical insight to pop out of the dark.
To go along with the experiment just think which of the above three factors (number of cars on the road, acceleration or deceleration) would cause a traffic jam or snarl. It can be more than one variable or parameter or a combination. Make a guess to start with.
The effect of the Torus AttractorWhat causes the jamWhat causes or ‘uncauses’ Traffic Jams?
I hope reflecting on these three representative samples (5) from the experiment (R waves) would help us see the underlying G (Green) and B (Blue) waves.
Interpretation of the findings
Incidentally, I saw the following:
1. Clearly the pattern is that of a ‘saw tooth’ wave (represented by the red graph, which shows the velocity of the red car at various points of time). Seeing such a ‘saw tooth’ wave instantly points us to the ‘quasi-periodic’ nature of the movement of traffic flow. This is also the characteristic signature of a ‘Torus Attractor’ (4), which is commonly found in reality. In simple terms it means that system after some time dances around lock stepped with this attractor and that is what keeps producing the events (R Wave) in a quasi periodic manner. The dance is at times periodic and at times non-periodic.
2. The acceleration and deceleration, the apparent opposites, combine to produce the boundary and the effects that are visible. So the boundary that is created by these opposites create the authentic constraints under which the system operates.
3. By tweaking the values of the authentic constraints the quality of the system changes. For instance it becomes clear that bigger the gap between ‘acceleration’ and ‘deceleration’ more would be the number of traffic jams with longer lengths that extend backwards (both frequency and amplitude of the jams increase). For certain ratios of acceleration and deceleration the jams minimizes or maximizes (I have not shown those ratios here to keep the post to a manageable size). However it is interesting to note that as we increase the minimum group velocity the incidences of jam reduces and so does the length of jams. When that happens the quasi-periodic nature of the event suddenly transforms to that of a periodic nature. That is what determines the group behavior or the ‘collective intelligence’ (the B wave).
So far so good. But how do we use this insight of ‘raising the minimum group velocity’ for practical applications in different fields? Let us explore.
A) EngineeringTechnical
The same happens for a Blast Furnace of a Steel Works. Whether the furnace would ‘hang’ or ‘slip’ (big problems for Blast Furnace) worries both Blast Furnace engineers and management alike. With this insight when I looked at the phenomenon I saw the same pattern of jamming followed by sudden opening up. The system is caught by a Torus attractor characterized by a quasi periodic movement. And the most surprising thing I noticed was that when deceleration was high the undesirable event continued for longer duration. And this was specifically induced when the blast temperature was below a certain limit. There is more to this which I may be blogging about later. But as of now we seem to have cracked a century old unsolved problem opening up a way to predict and take appropriate actions as to when the Blast Furnace would ‘hang’ or ‘slip’.
B) Production and Organization systems
Production shops are designed for ‘flow’. But as is often the case, we see material (finished as well as work in progress) jamming up the shop floor, which signals the most important source of productivity and profitability problems. So far, managers have tried many methods. But every technique seem to work for some time and then suddenly stops giving results. Why is that? This is because manufacturing units often operate under ‘dancing landscape’. Now ‘dancing landscape’ is a term used in complexity science which means that the system is operating in an uncertain environment. How is that? Most business units operate in uncertain conditions. Orders vary, customers vary, the level of service demanded varies, order sizes vary, logistics vary etc. But the system that is designed to offer goods and services often remain invariant. So naturally the system and its operating environment quickly goes out of ‘sync’ and trouble ensues.
However the insight gained from Traffic flow might help. The simple idea would be to increase the average speed of the manufacturing unit and not to concentrate specifically on any particular point, station or unit. Increasing the group velocity is the determining factor in design and operation of a business unit keeping in ‘sync’ with the operating environment. As of now the improvement ideas have revolved around increasing the velocity of the slowest unit, not the group as a whole. Well managers might now think differently.
C) Flood, Stock Market, Education, Economic Health.
I am putting all of these issues together since the basic nature of these seemingly unrelated systems is exactly the same. Can we predict floods much in advance so that we can save lives in a more effective manner? We presently do it by measuring the rise of water level. But what happens if we measure the velocity of a river at different points along its length in order to detect a significant change in the deceleration component enabling us to predict a flood much in advance to save valuable lives? I believe it would be possible to do so.
Similarly, no one has got the prediction of stock markets right. That is what appears to be the case till date. We gather trends in different ways. We trend the rise or fall of stock indices. We trend the rise and fall of the major stocks. We do all that in order to predict, which at its best is at times awfully off the mark. What we might do instead (only a suggestion) is to trend the acceleration and deceleration of the various market indices to predict better. I believe this would be useful.
How do we know how well a class or a school is doing? Noticing the highest and the minimum scores might be a way. Measuring percentile scores might yet be another way. I am sure most schools employ some sort of ranking measures and scores to keep tab on their effectiveness of teaching and growth in the capability of students. But what happens if we choose to consider trending the ‘acceleration’ and ‘deceleration’ of such scores over a period of time (since education is also a dynamic process)? May be we would have a better understanding of how the students are faring and how well the school is doing. I believe this would provide a useful understanding in the field of education.
The same insight might be applicable to Economic forecasts and predictions as well. Presently we measure the strength of an economy by measuring its GDP, PPP, growth rates etc. What might happen if along with the usual measures we measure the ‘acceleration’ and ‘deceleration’ of the economy. My guess is that it would reveal the presence of the same Torus attractor, which would not only help us to predict which way the economy is going but also help us understand better the actions needed to make necessary improvements. This too I believe would be very effective.
Conclusion
However, it is fair to say that though I have personally experienced how this insight helped me improve Engineering, Production and Organizational systems I haven’t yet applied the insight to other potential fields I highlighted like Flood Management, Stock Market predictions, Educational effectiveness and Economic health. But I believe that it would work just as successfully as it worked in engineering, production and manufacturing systems. What gives me this confidence? Simply because the pattern these systems generate would be just the same since they would all be governed by the same type of attractor, ie. the Torus attractor.
Notes
1. Nemetics is a thought model that allows us to take maths and insights from a designed model and apply them to any complex adaptive system – psychology, organizational sociology, economics, engineering, design, system design, manufacturing systems, system reliability etc. (contributed by Michael Joseforwicz)
2. Use of models (physical, mental, software) in Nemetics is a way to train the eye and mind to see the world in its dynamical state using a nemetical lens to understand the imperfection in the performance so as to take actions to adapt effectively. The contention is that every aspect of the physical and social worlds are best framed as Complex Adaptive Systems. Given that fact a content independent way to see and analyze CAS through models and maths is invaluable for whatever you choose as your speciality. (contributed by Michael Joseforwicz)
3. Tube is taken as a topological shape (a very common shape) where action takes place through independent actors (drivers in this case) which distorts or de-shapes the tube over time through a dynamic process.
4. It is important to explain the concept of attractor in simple terms since this is possibly the first time Chaos and Complexity which are generally treated separately are being brought together by linking the concept of ‘attractor’ to group or network behavior as seen in complexity. There are four types of attractors as we know today. These are the following:
a) Fixed attractor – think of a damped oscillation, It tries to settle at one point only. Example: A married man falls in love with a lady who is not his wife. His mind is occupied 24×7 with this new lady. Nothing on earth can buzz him off the fixed point. He is hell bent on bedding this lady.
b) Oscillating Attractor: I call this a ‘sync’ attractor — an attractor which brings in harmony. It is a periodic wave like a sine wave or a cos wave. It is something similar to – get up in the morning, go to office, come back from office, have dinner, do facebook, go to bed, make love … and this continues day after day.. So periodic. Or it is like a woman’s menstrual period.
c) Torus attractor — the characteristic is ”quasi periodicity”. Simple idea — I like to take holidays two times a year. But it is not very periodic. In the first year it is in July and December. In the next year it is in June and Nov. Third year it is in May and October. Note the quasi periodicity . Appears random but it retains the frequency of twice a year.
d) Strange Attractor – Now we don’t know what exactly might happen. Such an attractor is unusual and appears all of a sudden or is not anticipated by us. Hence the term ‘strange’. Examples – heart attack, epilepsy, etc.
5. I chose the screen shots from my experiment to depict two things — 1) the saw tooth wave (the red graph) and how the group velocity of the cars changes while moving within the constraints defined by the upper and lower limit of group velocity determined by the acceleration and deceleration of the cars. Notice how the red graph varied as I changed the acceleration and deceleration of the group thus affecting the group dynamics.
Reliability Management Consultant Pvt. Ltd (RMCPL) 