The Corporate Startup Experiment

“Computer Science is no more about computers than astronomy is about telescopes” Edsger W. Dijkstra

I’ve been reading a lot lately about the concept of a Corporate Startup. Not the book (, which doesn’t yet exist, but the concept itself. The idea that large businesses need a way to stay nimble and look forward without necessarily sacrificing their core business is not particularly new.

A typical large business in the past could not move fast enough. Great ideas were lost in the bureaucracy or killed outright for fear of encroaching on their core business. Greatest example of all times: Kodak. A company which invented the first digital camera back in 1975, and was later put out of business by its own invention. If you want to learn more about, I recommend watching Dr. Tendayi Viki’s talk on Youtube

The reality is, big companies need to be both fast and nimble when generating and exploring new ideas, and slow and meticulous when it comes to their core business. A popular Harvard Business Review article on innovation accounting, commonly referred to as Nagji and Tuff, even suggests that a 70-20-10 investment (70% in core business, 20% in adjacent areas, and 10% new research), when used by “successful” companies, can bring a reverse, 10-20-70 return.

Finance folks like to talk numbers, and in that sense 70-20-10 is a perfect rule. While important, alone, it’s not going to make a company successful. Reading between the lines, a “successful” company spends 70-20-10 to return 10-20-70. The reverse is not necessarily true. The bigger questions are when to switch gears, how to align business, so the companies actually get an idea of how to invest those 70-20-10.

What interests me the most, perhaps, is none of that, rather how to run the 20-10. If we run them like the rest 70, we are doomed to fail long-term. That got me to think to think about the power of analogies. I’ve long been a huge fan of looking inside myself for ideas. Simon Sinek talks a lot about how we should look at the human history and physiology, and reflect it on everything from leadership to company structure. The Toyota Way is taking a very similar approach bringing together various aspects of humanity, Japanese culture, and simple common sense. Daniel Kahneman’s “Thinking Fast and Slow” made me think of a striking similarity between Human Psychology and The Corporate Startup.

In “Thinking Fast and Slow”, Daniel Kahneman introduces what has long been known as a Dual Process Theory. The theory itself has existed for quite some time, but for his ground-breaking research, Kahneman received a Nobel Prize in Economics in 2002. If you haven’t read the book, I strongly encourage you to do so. The basis of the Dual Process Theory is that we follow two kinds of thinking: associative and reasoning. Associative is what Kahneman calls System 1. It’s quick, intuitive, and requires no reasoning whatsoever. Stepping back when a car next to you is about to hit a puddle is an example of quick, associative thinking. This system is based on a gut feel, and will often produce a wrong answer or even and answer to a wrong question. Reasoning, or what he calls System 2, is slow and meticulous, where we have to build a line of reasoning and come up to a logical result.

Established businesses have an existing customer base and a revenue stream they do not want to endanger. They need this revenue stream for a variety of reasons, including a cashflow to help discover new revenue streams. Upsetting old loyal customers is typically not a good idea: it costs a lot more to attract new ones. That’s why the Corporate System 2 is a dominant factor for the the core investment (70%).

Investment in adjacent areas (20%) could use a hybrid approach. We could definitely use a combination of a gut feel (System 1) and some hard logical thinking (System 2). For example choosing and potentially profitable adjacent area definitely requires both the gut feel to look for direction and reasoning to determine its potential.

The un-adjacent area, the 10%, are of course, the most interesting and exciting one. You are like a firefighter entering a burning house. Did Google manufacture cars before they started their driver-less car business? They’ve entered a totally new area, with little or no in-house expertise in it, on a gut feel (also called a corporate vision these days) that this is something that could potentially change the future. And I am sure that some very hard reasoning went into justifying the short-term costs, such as useful research that could help existing Google products, such as computer vision, pattern recognition, robotics, AI and a lot more. Then System 2 would also get employed in finding appropriate personnel and resources, and building an environment where those employees could be the most productive.

There are 3 major rats to system 1: overconfidence, risk-aversion, and overestimating negative effects. Ever noticed that the stock market tanks on a negative news, but if the news is good, its effect is diminished? Someone told me that it’s because the the positive news was already expected and was priced into the stock. Perhaps, but there is another explanation: we like to panic once our overconfidence has been shattered. And then we get so risk-averse we start missing on opportunities. Our reaction to the news is driven by System 1.

A successful company is not the one that’s a near monopoly sitting on a pile of cash. It knows how to switch from System 1 to System 2: when it’s time to go into a panic mode and cut the losses, and when to invest in the future… More than 50% of the time, at least. That’s what made the likes of Goggle, Facebook and Apple successful so far, and the likes of Kodak, Motorola, Chrysler, and many others to wither away.

I didn’t start writing all of this just because I read Kahneman’s book and got excited. I feel privileged to participate in an experiment at one of those large companies that are open to running themselves as a corporate startup, by giving employees enough leeway to do the best job they possibly can, while keeping traditional controls in place to guarantee a painless switch to System 2 when necessary. We have a very small and lean team of data engineers operating in a Bitbucket/HipChat/Trello mode, in addition to a regular PM keeping his hand on the pulse of things. It’s an interesting experiment that I hope will set a precedent.


Self-healing Organizations

Just the other day, I watched a great talk by Joe Armstrong about self-healing systems. And I got to thinking: as we have figured out how to build effective nearly self-healing software systems, why can’t we figure out how to build effective nearly self-healing software companies?

We are not yet ready to have a completely self-healing system. A failover as a result of a serious hardware failure of a component in a large cluster does not remove the need to replace a failed hardware component. It just delays the need for human action until later while keeping the system operational and service uninterrupted. Even in the age of robotics, where the robots can do a lot of things that humans can do, there are still parts to be ordered, shipped, or even just pulled from a warehouse. Robots can remotely move entire racks in a million square-foot amazon warehouse, but they are not yet capable of healing a failed hard-drive in a data center. In today’s world, human intervention is still absolutely required.

Per Joe Armstrong, in a distributed system, “each module is a unit of service and a unit of failure”. A fault-tolerant system needs to “fail fast” (provide an immediate notification), so that we know that there is a fault; and “fail early, fail often”, so that we can make changes to the system sooner when it’s easier to diagnose and remedy problems. A fault-tolerant system like this also creates a sort of horizontal transparency, that allows it to self-heal (failover), and a vertical transparency that enables human intervention (diagnose the problem, pull a spare part from the shelf and replace the failed one and re-order a spare).

Vertical transparency is especially hard. Let’s turn to Sal Khan and take a brief excursion into compound probability of independent events. I’m joking of course, but a consistent failure to see the big picture opens the door for little slip ups to start compounding unnoticed, thus reducing the likelihood of the long-term success. A small perceived slip up in a product quality may produce serious reputation damage. Don’t handle that situation correctly, and the reputational damage could be even worse. As a result, you could be losing repeat customers and damage could last for years to come. At the same time, these risks corrected on-time, while still serious on their own, would at least remove the compounding danger.

Here is a Life of a Bug story. I remember that almost 20 years ago, a company I worked for had an issue with a version control system. The issue itself would not have been a big deal, if in addition to that, they have realized that their backup system was failing unnoticed for weeks, and there was no reliable backup to restore it from. Which in return caused lost code, slipped schedules, and an avalanche of other problems… Fail to manage horizontal transparency: and risks start staggering.

A successful organization, is much like a self-healing, scalable, distributed system. It shares many of the same characteristics. Just read Eric Ries’ “The Lean Startup” and you will see that “fail fast” and “fail early, fail often” are at the core of his “pivot” concept. agile startups are based on testing hypotheses, collecting data, and constantly iterating products”. I would replace “startups” with organizations in this particular quote. Truly agile organizations can survive many near-death events during their lifetime and come out even stronger each time.

Let’s rewind back a few years, all the way to the Toyota sudden-acceleration fiasco. In addition to the cost of human suffering, the crisis did cost Toyota billions in fines, recalls, reduced demand, as well as reputational risk. Let’s be blunt, for the vast majority of big and small corporations this could have been a near-terminal event. Banking crisis in the US caused banks to shatter entire divisions and to layoff thousands of permanent employees. What they did was an equivalent of performing an autopsy on a live person. They took little time trying to understand what they did wrong and why, and how to make their businesses more competitive when economy recovers. They chopped sick limbs off hoping that that would reduce the suffering. Yet, while Toyota had to let go many temporary workers, it also found a way to preserve most of permanent jobs. Toyota viewed this hardship as an opportunity to improve, to optimize its factory floors instead of shutting them down, in order to be ready for a sunny day.

Just over a week ago, I traveled to Florida to seek employment with an amazing company called Ultimate Software. I didn’t get the job: it wasn’t quite where my strengths were or what I wanted to do. But I came back having learnt about their “People First” culture. The idea comes from the Lean principles of the Toyota Production System (TPS), and it’s not necessarily a novel idea either. In his book “Leaders Eat Last”, Simon Sinek shows a multitude of examples of companies that realized that it’s the employees who are their most valuable asset. If we look at the “house” diagram of The Toyota Way, the two pillars under the roof are Respect for People and Continuous Improvement. Toyota has always put People First, and not just its employees, but everybody: employees, suppliers, customers, … everybody who could possibly be involved in the process. In fact, one of the conclusions that Toyota made as a result of the sudden-acceleration fiasco was that they indeed needed a better way of collecting and analyzing the customer perspective. For Toyota, to have overlooked the customer in such a way was definitely a huge failure: TPS simply was not trained to look at an additional angle. Yet the same system, TPS, was trained to sense that there is a problem, get to the bottom of it, and engage in a series of corrective measures that clearly worked. Let’s just think for a second: how many of us work for a company that has such a system in place?

One of the specific traits of the Toyota Way, is the emphasis on hiring the right people. Principle 10: “Develop exceptional people and teams who follow your company’s philosophy”. I believe in hiring right people with right mentality. Hiring people who believe in building the horizontal and vertical transparency, rather than the arcane bureaucracy where each module is hiding its risks and failures from others, and where the hierarchies consistently fail to take proper corrective measures because they are either not empowered or are afraid to ask for help. A bright-looking project plan does not help create the culture of trust and transparency. It is the culture of trust and transparency, that makes a project plan be worth the paper it’s printed on.

A self-healing human system hires the right People and then puts them First. It can’t prevent every possible occurrence of every possible problem, but it observes and rectify problems quickly, at all levels: at the level of a floor-shop, and at a level of a CEO of a company, and can extend its function all the way to the customer, as Toyota managers found out. A self-healing system that is transparent and hires the right people will be able to pivot and survive.

User Sympathy=Mechanical Sympathy!!!

Early in the morning,I headed to Sheraton hotel downtown Chicago for Cassandra Days Chicago. Not to say that I was able to learn lots of technical details that I haven’t already know in the two hours before I had to head back to work, but I was able to carry away the general idea, the spirit of Cassandra that no book can really emphasize enough. Now that you know that I am absolutely crazy about this particular NoSQL database, here comes a shocker. Remember the beginnings of each quarter or semester in college, when a new professor would tell you something along the lines: “Last year you studied Math. This year we will learn History. Math and History are so much alike…” Sounds crazy? It did to me back then, and it still does today. That professor was right though, and here is why.

Mechanical Sympathy is a concept that literally means that you need to love and understand your hardware (or any domain for that matter). And if and when you do, there isn’t a problem that cannot be solved. There is a great InfoQ presentation about it. When it comes to software, for instance Cassandra, I extrapolate Mechanical Sympathy to be deep understanding of the implementation domain of that software. Cassandra was brilliantly architected for a very specific domain: be very fast writing and reading, yet for very specific usage patterns. Try to use Cassandra as a sort of quasi-relational database, and it will bite back with errors (if lucky) or significantly handicapped performance. Writes are the fastest when they are sequential, and bulk reads are fastest when we need to retrieve large quantities of sequential data. This means that one would have to understand how to properly use partition keys and clustering columns and how indices work in Cassandra. That’s an important technical detail, but perhaps not the most important. The most important is this: don’t think about how to store your data, but rather how the data will be used! De-normalize! NoSQL databases exist to make impossible possible, that is to allow us to store and analyze unlimited quantities of data quickly.

Let’s stop and ponder for a moment how we present data to a user. Whether its a sophisticated user interface, or a complex mathematical computation, nicely-normalized data is of no use. We grab a fully de-normalized view and present it to the user. In the old relational reality, we had to do all the joins and selects first, and while we were doing all that, the user has to sit and wait. And so we resorted to tricks: precomputing data, temporary tables, client-server, caching, background processing, even using temporary files for storage. NoSQL databases, on the other hand, are very good at storing de-normalized views that so conveniently can be presented to end users. Simply said, by dissecting what users need and want, we can figure out the best way to layout the data, not vice versa. Truly, User Sympathy=Mechanical Sympathy!!!

So, why even bringing up Mechanical Sympathy. Once, there was a very popular agile technique, DDD or domain-driven development. DDD focused on thoroughly studying the problem domain, gaining the understanding, and using that deep understanding to come up with an ultimate solution. Later today, I participated in a meeting that was called to find a solution to a very interesting problem. Instead of modeling customer behavior, we again modeled the data, and stored it in a very fast object store. The problem: that data model could not be presented to the user, hence a lot of performance-prohibiting work needed to be done to re-shape it, and the users weren’t happy.

I’ve written about HCD (Human Centered Design) before. HCD is centered on 3 pillars: feasibility, viability, and desirability. If you think about it, Software Design isn’t that different from Human Centered Design. In fact, we should follow HCD to design software. We need to have User Sympathy to understand the user domain and to provide a service that the user desires, and Mechanical Sympathy, so that such a service can be implemented, and not break the bank.

A Factory of Good to Great!

The true delight is in the finding out rather than in the knowing.

Isaac Asimov

How often did you receive a feedback from your manager or HR that you are very good, but… not so great. It’s all about paperwork, questioners, isn’t it? Your manager has to say smth good, and smth bad. And then he needs to sit down and figure out how you fit into his overall compensation plans, and adjust the review based on what he can afford at the moment. He is the king of his own kingdom, and of course, there is another king over his head, and so on. It’s a system based on merit, decided by a variety of factors, such as ability, luck, politics, and a lot of other very personal things to the very personal people. In other words it is subjective, and therefore imperfect. Now, imagine that you could work in a system that is completely objective, i.e. 100% pure orange juice and not from concentrate! In this system, your hands are completely untied, and you could reach for the stars. Even better, the system would propel you to get even better: sort of a flywheel that just keeps the good energy running. Well, Holacracy isn’t one of them: not designed, not supposed to be. But with objectivity being the ultimate prize, it gets pretty close.

First a disclaimer. I am not an expert on Holacracy. The first time I’ve read about it was just a few weeks ago, when Zappos finally decided to go all the way in adopting it. And what I’ve read was an immediate mixed-to-negative reaction from our mainstream media: “The holes in Holacracy”, “Making sense of Zappos and Holacracy”, “My company adopted a Holacracy. It kind of sucked”. Let us be fair. All these articles are right in their arguments: Holacracy has some very serious issues. Well, so does a Democracy, and yet we prefer living in a democratic country rather than autocratic. Raise your hands, how many of you would rather live in United States than, say ChinaRussia?

Though the United States is not a pure Democracy (we are a Democratic Republic), we vote for our representatives at all kinds of local and higher level circles all the way to The President. Those circles are divided by Purpose, rather than Authority. We vote for township councils, school districts, counties, states, and so on. It takes us a long time to come up to a consensus. Our votes can be skewed and misguided, so we elect the wrong politicians. But these our votes, that we’ve cast, and we bear full responsibility for them. These votes make us better, we learn from them. Holacracy is even better. There is also a hierarchy of circles, and the lead-link (see Holacracy Constitution, Sec 2.2) at each level has a vast amount of authority. In fact it’s not a whole lot different from any other company, since the lead-link of enclosed circle is selected by members of enclosing circle. So the bosses sort of select the underlings, except they do it with a rather transparent due-process described in the Company Constitution. But there is a huge difference. First of all it allows a lateral shift. Each employee usually participates in a multitude of circles: could be a lead-link in one, and just fill a role in another. Got bored in a Design Circle, resign and go the Architecture circle. If you are good at it, you will succeed in the new circle. And if not, then, perhaps, move on to a different circle. Nobody can encroach on your autonomy within that circle just because they have a special relationship with the boss: it’s clearly defined and documented, and to be changed, it takes the whole circle to make a decision.

Pretty Agile, isn’t it? Holacracy is a very lightweight process. Outside of the constitution, the circles themselves decide the limits of individual autonomy, charters, and so on. None of that is pre-set: its a living book. Even better, ever tried to decertify a taxing body in my home State of Illinois? All it takes is a decision by the enclosing circle to disband an enclosed circle, or all of the members of the enclosed circle leave. Holacracy is driven by a concept of tension. When there is tension, change will automatically come. And it comes within the bounds of each individual’s role or a circle’s authority.

Holacracy != Democracy. There is really no need for soldiers in the Army to elect their 4-star general: they have little understanding of his tactical abilities. I’ve always wondered why I get to vote for a judge or a sherif in my home state? I don’t understand how to evaluate judge’s qualifications. The the fact that that sheriff is tough on crime, doesn’t mean that he won’t let criminals go free and send innocents in jail: those kinds of decisions must be made by experts who will evaluate that person’s qualifications in great detail and make an educated decision. This is exactly the role played by circles.

One of the bigger complains that I’ve heard about Holacracy so far, is the difficulty to make a decision. Lots of meetings are often required, especially at first. Here is my answer to this: Holacracy is a Lean process. Learn/Build/Measure, and as the time goes by you should be able to adjust living processes, and focus on tension where it needs to be, rather than on making communication work.

Here is what I, personally, look for in Holacracy: that goodness flywheel that can take a company from Good to Great. It’s about Teamwork and Autonomy, where everyone gets to learn how to be that humble servant leader that cares about the common good and big picture rather than individual success. If this doesn’t sound familiar, check out the definition of a Level 5 Leader in one of my past posts. My hope is that Holacracies can become factories of Level 5 leaders, taking their respective companies from Good to Great!

Motivate thyself to Succeed

“You are essentially who you create yourself to be and all that occurs in your life is the result of your own making.”

Stephen Richards, Think Your way to Success: Let Your Dreams Run

I spend most of my day at work. Likely, so do you. Say, you spend 8 hours at work, and another couple of hours on commute. By the time we get home, a typical adult working full time has only 3-5 hours to spend with family.

Work is an important part of our social life. Imagine how terrible it would be if we just hated getting up and going to work every morning? It wouldn’t matter how much we are getting paid: turnover would be sky high. Does everybody have a perfect job: always exciting and interesting work, nurturing boss, and supportive co-workers? Clearly that’s where we want to be. That’s the Nirvana that no company will ever be able to accomplish.

You may have heard about intrinsic vs. extrinsic motivation before. The difference is simple: extrinsic motivation is a treat, while intrinsic is a habit. If your dog knows she would get a treat every time she does a trick, she will be watching for that treat. If you don’t have it in your hands, she is not going to do it. That’s extrinsic motivation. On the other hand, if you like to paint, or sing karaoke, or run a marathon, you will keep doing it for the sake of it. Being compensated, say by winning a competition becomes secondary to doing it. I write this blog simply because I enjoy writing down my thoughts. It’s important for me to know if someone finds them useful, but I would keep writing even if no one cared.

As I was reading Susan Fowler’s book “Why motivating people doesn’t work”, I realized that even the brightest of us got it all wrong. Lean and Agile are about creating frameworks that motivate people to excel. Susan Fowler talks about managers or leaders doing it for the rest of us. Both approaches are based on the right science and work great. The key fault in both is an assumption that the rest of us mortals need to be shepherded to get there. I wonder why, in a truly Lean or Agile environment, where we value “individuals and interactions over processes and tools”, each of us cannot be our own leader and try to achieve the level of intrinsic motivation that will get us excited to go to work every morning and excel at what we do.

Look at this article from the fast company that lists 6 motivational outlooks that Susan Fowler described in her book. The first 3, disinterested, external, and imposed, are purely extrinsic: we would end up inventing various carrots and sticks to get ourselves going from one occasion to another. Next, aligned and integrated, are still somewhat extrinsic: we are still looking for a motivating factor, except now, instead of carrot and stick, it’s more of a shared purpose or vision. The last, inherent, is truly intrinsic: we do what we do because we like it.

According to Fowler’s book, there are 3 things that are absolutely necessary to get us excited about our work. They are what she calls ARC: autonomy, relatedness, and competence.


Leaders need to first look at themselves, understand what motivational outlook they belong to, figure out what needs to be done to improve it (learn) by basically asking 5-why’s, shift (build), and reflect (measure); and keep repeating this cycle (sounds pretty Lean, doesn’t it!). And then they have to help everybody else do the same.

Servant Leadership, implies leading through helping to create a fertile environment, similar to the one described by inherent outlook. Such would be a job of a Scrum Master for example. Except, nothing would help if Team Members aren’t perceptive to Scrum Master’s guidance. This is why I am a strong advocate of everyone becoming own servant leader, own motivator. Don’t wait  for someone to tell you how to get there. Try figuring out what motivates you, and don’t forget ask for help when in trouble. In fact, ask your neighbor, friend, team-mate before going to your manager or Scrum Master: the more help you get the better.

In trading, a strategy will break even if it works more than 50% of the time. Taking into account costs, let’s say a break even trading strategy needs to succeed 55% of the time. An organization, where 55% of employees are able to attain some degree of intrinsic motivation has a chance to be successful. Anything below is a failure, and anything above is the degree of success. This degree of success depends on 2 factors: effective leadership that creates an environment that nourishes success, and ourselves who should be using all available resources to succeed.

The Do (Spirit) of Perfection

It’s late morning. I am at a resort restaurant in Mexico, sipping from a cup of coffee, bored. Suddenly, I notice a waiter replacing a table cloth on a table, a few feet away. He swipes the dirty table cloth off with a single precise move, and then unfolds the new one in its place in a just a few more. Much like watching David Copperfield perform a magic show, I watch him in awe as he repeats his moves over and over again.

Same restaurant a few hours later. It’s lunch time and now I intensively study waiters at work hoping for the repeat of the morning show. It’s another waiter’s time to clean tables, and he does his job well. Except that this time the show does not happen. The moves do not express waiters confidence in what he does, they were different and there were more of them, and they seemed to differ somewhat from one table to another.

Why did I bring this tale of the two waiters? Principle 6 of The Toyota Way says: “Standardized tasks and processes are the foundation for continuous improvement and employee empowerment.” Whereas, Principle 7 says: “Use visual controls so no problems are hidden.” Both principles are tied together in a way. Waiter 1 seemed to have polished his steps to perfection, while waiter 2 did not. As I was watching both in action, I could easily see  the difference.

Principle 6 was not devised by some bureaucrat at a large corporation. It is rooted deeply in Japanese culture. As Zen monks “needed to pass on instructions to greater numbers of learners and to later generations… it became necessary to train many teachers, or masters, and to be able to do this, the formalization of do practices began in the mid-Edo period” (sometime in 17th century) So, per Davis and Ikeno, the process of mastering an art involves 5 steps:

  1. keishikika (formalization)
  2. hampuku (constant repetition of pattern or form)
  3. mastering the pattern or form
  4. kanzen shugi (achieving complete perfection)
  5. toitsu (becoming one with the pattern or form, going beyond it)

To make things a little simpler, a pattern needs to be identified, followed, perfected, and then we can go beyond it. Standardized tasks in the Toyota way are just such a pattern, which needs to be learned and perfected. As we reach the toitsu (become one with the task), we can now reflect and improve the task as needed.

Both waiters did get their jobs done. Even though waiter 1 was probably working a little faster, let’s for a moment assume that both could do their 3 tables in the same amount of time. Which waiter do you think I would ask for next time I come back to that restaurant? I would always pick waiter 1 because I appreciate his mastery. I think that majority would subconsciously make the same choice. Clearly, a win for the Spirit of Perfection.

Can a Spirit of Perfection win when the definition of perfection is fuzzy? Classical music lover may not appreciate rap, but a rap lover would appreciate good wrap and a classical music lover would appreciate good classics. Formalization process has been invented and perfected for a variety of arts, from fencing to calligraphy to flower arrangement. Thus applying formalization to any other creative process should just be a matter of defining the forms and processes.

In calligraphy, a pattern is perhaps a letter or its component and in martial arts it’s a particular position or move. Just as a painting consists out of visual patterns and techniques, software consists of patterns,  architectural components, and techniques. Just as a painter perfects his technique and becomes ONE with his patterns and starts inventing his own style, a software developer gains mastery of software patterns and becomes ONE with his skills and can start taking the art of software development in an entirely new direction.

Besides their skill level, there seemed to have been one more thing that differentiated waiter 1 from waiter 2. I think it was a Spirit of Perfection: waiter 1 seemed to have been determined to be great, while waiter 2 was just doing his job. The Spirit of Perfection was in the air and that’s what I think truly caught my attention. And while there is no guarantee that waiter 1 will one day run that restaurant, the Spirit of Perfection is definitely one of the determinant factors for future success.

Saihatsu Boshi or Culture of Prevention

“Those who forget the mistakes of the past are doomed to repeat them”

George Santayana

Saihatsu Boshi stands for “prevention of reoccurrence”. It is not just about prevention though. It’s about a culture of learning how to get better. It’s about 5-whys, getting to the bottom of every issue and making sure it doesn’t happen again. In other words, it’s about an insatiable, perfectionist drive to get better.

Is perfectionism good or bad? One of the main things I admire in people that we call Role Models is their insatiable drive to make the world a better place. Whether in Sports, Art, Science, or Business, they set the bar so high, that it seems impossible to reach it. And when they succeed, they change the world.

Culture is an integrated system of learned behavior patterns which are characteristic of the members of a society I.e. it’s a learned behavior that we follow subconsciously. In Japanese culture, Saihatsu Boshi is one of those behaviors, the main driver behind Hansei and Kaizen, and the cornerstone of the Toyota Way and Lean Thinking.

Culture is a learned behavior; therefore cultures evolve. Cultures spread their influences and affect other cultures. Western culture has been affecting the cultures of people in Asia, Africa, and South America and vice versa. In modern days, cultures of Central and South America have a profound effect on American culture. Japanese  culture has been subliminally affecting ours for decades. Not just by means of Sushi and Anime, but from Lean Principles that took hold of our manufacturing and service sectors over the past decades.

Japanese don’t need to prove themselves right. Clearly our traditional western thinking reached a limit when we had to import back Lean ideas to restore our science, industry, cinema, software development. But when importing new ideas, we often trust the interpreters, without clearly understanding why these ideas worked in the first place. People like Ken Schwaber and Jeff Sutherland gave us Agile (Scrum) with retrospectives and process improvement. Unfortunately, we looked at their teachings as a dogma. Majority never learned that the real reason Agile advocates retrospectives and reflections lies in Japanese culture, and Saihatsu Boshi.

Some argue that in a fast-paced scratch-and-go Agile environment, there is no need to get to the bottom of every bug. They argue that the old and buggy code will likely be thrown out before the fix is even needed. They try to rationalize if and when the bug may or may not be affecting customers. Lean teaches us to study every problem, apply a temporary fix, and then work on implementing a permanent solution a.k.a counter-measure. Lean doesn’t tell us how much time we should spend on every counter-measure. It’s all about common sense. What’s important is to get to the bottom of every issue and to learn from it. The counter-measure itself could be a small change or a complete architectural makeover. The actual solution is not as important as a subconscious desire to continuously learn and improve.

Today, saying “don’t fix it if it ain’t broken” is no longer enough. When is it broken enough to say that something needs to be fixed? Does it matter? Ayn Rand once said: A culture is made — or destroyed — by its articulate voices.“ Perhaps I’m a bit of a dreamer who believes that each of us should be one of those articulate voices. It will take a lot of voices to drive cultural change. We need to set the bar high and start.

PS. A friend recently showed me the writings of Rochelle Kopp: SAIHATSU BOSHI – LEARNING FROM MISTAKES TO PREVENT REPEAT PROBLEMS”  and SAIHATSU BOSHI – “THOSE WHO FORGET THE MISTAKES OF THE PAST ARE DOOMED TO REPEAT THEM” I would recommend these 2 rather small articles to anyone who wants to understand what Saihatsu Boshi is about.