It was the antiquated method (largely paper) with which the FBI managed the data that prevented from connecting the dots that could have avoided 9/11.
One office wondered why so many foreigners took flight lessons, another signaled the entry into the country of various Al Qaeda activists, a third raised suspicions about a specific person. But their information remained isolated, no analyst was able to access it. According to the 9/11 Commission, the FBI’s IT system proved inadequate to the situation.
Since then, various attempts have been made to solve the problem, including Virtual Case File, which cost $ 170 million: not even a line of code from that computer system has ever been used.
Then it was Sentinel’s turn, in 2005, an estimated cost of 541 million dollars, which had to be operational since 2009. But in 2010, Jeff Johnson was called to solve this situation: 405 million had been spent, just half had been developed of the program, an additional 6/8 years of work and an additional 350 million investment were estimated.
The cause was not the quality of people’s work, but their way of working. The project had been organized according to a method called waterfall, based on Gantt diagrams – named after their inventor, who had developed them in 1910.
Gantt charts were used for the first time to supply General Crozier’s troops during the Great War. In practice, we abandoned the trench warfare for the war with drones, but we pretended to build the FBI computer system with the methods of the last century.
When the method slows down the project
Sentinel’s problem was that new bugs (software problems) emerged faster than old ones were fixed. The project was out of step.
Jeff Johnson and the FBI CIO, Chad Fulgham, decided to save the salvable by bringing the project “home”. They identified the 9 main problems and focused on them, but first of all they established a change of management.
Managers wanted two things: control and predictability. This led to huge planning efforts, futile attempts to avoid change, to know the unknowable. Months of time and resources were wasted.
Scrum captures uncertainty and creativity.
Its essence says: start the project and check it frequently and regularly, check if you are going in the right direction, check if there is a way to make things go faster. This is what is called the “inspect and adapt” cycle.
At short intervals, we stop, review, correct, recover. Scrum can be used successfully to build cars, run a laundry, send a rocket into space.
Let’s go back to the FBI. A thousand and one hundred specifications, expressed in a pile of sheets several centimeters high, composed the documentation. Johnson and Fulgham put order and set absolute priorities, which would give greater value to the project.
In the fall of 2010, the FBI says it can complete Sentinel within a year by applying an “agile methodology”, using fewer employees and spending $ 20 million.
Scrum is the process that puts into practice the values expressed in the Agile Manifesto, first of all: responding to change rather than following a plan. It works by removing obstacles, an idea that comes from Taiichi Ohno’s Toyota Production System, for which the key concept in development processes is flow.
Production must flow quickly through the process and one of the main tasks of management is to identify the obstacles that hinder this flow.
To complete Sentinel, Jeff Johnson organized the work in two-week cycles. These periods in Scrum are called Sprint, at the beginning of each one a meeting takes place during which the working group decides what to do in the next two weeks.
These activities are extracted from a list of priorities and displayed on a wall, so that everyone can see them. At the end of the Sprint, the team meets for a meeting called Demo, during which it shows the work done, so that we can verify that it is going in the right direction.
According to Johnson, the Demo, that is the demonstration of the progress of the activities, was the most powerful element for the “salvation” of Sentinel: nobody believed that with 5% of the budget and in just 20 months the team was managing to do what it was It was impossible to achieve by investing 90% of the budget over 10 years.
Once on the move, the Sentinel team increased its speed by 3 times: they understood what slowed them down and with each new cycle they managed to eliminate some obstacles. In July 2012 Sentinel was launched. Its effect on the entire work of the FBI has been enormous.
Management in step with the changing world
The world is changing and the same happens to our work. Under the hood of a new Ford there are as many lines of code that in Facebook and Twitter put together. And every time people are involved in a complex creative effort, traditional management simply doesn’t work.
Instead, these simple rules work:
- planning is useful, blindly following a plan no: it is necessary to inspect and adapt;
- clinging to the command and control method will lead to disaster;
- failing quickly, that is, verifying that you are going in the wrong direction, is a key factor for success;
- work that does not produce real value is wasteful.
A structure that allows the team and managers to be successful
Teams are complex and adaptive systems: when something is changed, the whole system goes into a state of chaos, then the organization settles down again. What matters is that the new structure is better than the previous one.
Scrum is a set of rules that guides teams in this change, which leads organizations to settle into a more productive and happier state.
In 1986 the Harvard Business Review published an article signed by business analysts Hirotaka Takeuchi and Ikujiro Nonaka. Entitled The New New Product Development Game , that article explained the results of studies conducted on the world’s most innovative companies.
The development processes of these companies were flexible, cross-cutting and autonomous teams. Managers were servant leaders , and acted as facilitators. The two authors compare the work of these teams to a rugby team: the ball is passed within the team that moves as a single entity on the field.
Seven years after the article was published, no American manager had managed to make sense of it and put it into practice. It is from that basis that Scrum was formally born, presented in 1995, together with Ken Schwaber, as the Scrum development process .
The Plan Do Check Adapt production cycle
The PDCA Plan Do Check Adapt cycle (plan, act, adapt) of Deming, an expert in operations management of the fifties, is still usable today in the production of anything.
When defined, it immediately established itself as a revolutionary and became the basis of the Toyota Production System , a method that allowed the Japanese car manufacturer to become the first in the world.
To teach Scrum, one of the simplest activities is to apply PDCA to build paper airplanes. The demonstration of improvement, i.e. improvement of the process and growth of results, is at the heart of this exercise, which demonstrates in a practical way how self-management, process analysis and small changes lead to amazing results.
Scrum and Japanese culture
Scrum’s roots lie in Japan and its culture. Like Aikido martial art, Scrum can only be learned by practicing it.
The body, mind and spirit align and progress through practice in constant improvement. In martial arts, Shu Ha Ri is followed , which illustrates various types of mastery.
In the Shu state we learn the rules, the forms, the pupil repeats them like the steps of a dance without modifying them, so that the body absorbs them. In the Ha state , changes can be made.
In the Ri state , forms can be abandoned because they are now mastered: each movement expresses the essence of Aikido.
The characteristics of the great teams
The characteristics that make a team great are coded and reproducible. Hirotaka Takeuchi and Ikujiro Nonaka describe the teams that have seen the best companies in the world at work:
- they are transcendent. They have a sense of purpose superior to the ordinary. They are committed, reject mediumship and aim for greatness, a decision that modifies the idea they have of themselves;
- they are autonomous: self-organized and self-managed, they make decisions and put them into practice;
- they are cross-functional: within the team there are all the skills necessary for the realization of the project, and those skills are reinforcing and feeding each other.
The All Blacks, New Zealand rugby team, are a transcendent team. Before playing they perform in the haka , the Maori war dance.
Let’s examine what the team expresses during the dance:
- great focus on the goal, supported and strengthened by singing;
- total collaboration: people move like one man;
- hunger for victory;
- excitement about the challenge.
The team size is important, the optimal size is 7 +/- 2, that is more than 5 plus but less than 9.
According to Brooks’ law, if you add a person to a delayed project, the delay increases.
To understand why, one must look at the functioning of the human brain: there is a structural limit of short-term memory. There are basically two problems when times get longer when we put too many people on a project: it takes time to bring everyone to the maximum level of efficiency, and secondly, the number of information channels that our brain has to manage grows.
The formula is:
communication channels = number of people x (number of people – 1) / 2
So for a team of 5 people there will be 10 channels, for a group of 8 we will have 28, for a group of 10 they will increase to 45. As the channels grow, the difficulty of people to circulate information increases, slowing down the work of all.
The role of the Scrum Master
To ensure the efficiency of a process supported by meetings where plans are made, critical evaluations and retrospective analyzes are expressed, a manager is not needed, but a leader at the service of the team, a middle ground between captain and coach.
The name chosen for this role is Scrum Master (melee expert), his task is to facilitate meetings, ensure transparency, help the team to discover obstacles and ways to overcome them.
One of the main tasks of the Scrum Master is to manage the meeting in which, at the end of the Sprint, after the Demo, the team examines how it worked in the previous weeks in search of errors to be eliminated and possible improvements.
Use the Retrospective to improve the Sprint
This meeting is called a Retrospective, and it is essential that during its development the Sprint is judged without committing what the experts call “fundamental attribution error”.
Psychologists have found that when analyzing one’s own error, many situational factors are taken into account, and the reasons that led us to take a certain direction are cited. We do not do the same when analyzing the behavior of others.
It is a phenomenon known as a fundamental attribution error, a dynamic already studied in the seventies, according to which we tend to look for the culprit, we do not look at the system trying to understand what generated the problem to solve it.
During the retrospective we start from the concept that dysfunctional systems create problems, people are only part of the system, and it is it that must be corrected.
Sprint and time
The idea of the Sprint comes from the MIT Media Lab, where, in the early nineties, they had given themselves a rule: every 3 weeks the researchers showed what they were working on. They were open demonstrations, anyone could attend.
The Sprint are “time boxes” (time boxes), have a predetermined duration and should be constant. You don’t do a one-week Sprint and then one of three.
The work rate must be established so that people can make valuable releases, working pieces, useful progress for the user of the product.
The 15 key minutes of the morning
Each morning the team meets and asks three key questions. Everyone says what he did the day before, what he will do that day and brings out possible problems called impediments. And that’s that.
If this meeting takes longer than 15 minutes, it means you are doing it the wrong way. The purpose of the Stand up is to help the whole team understand exactly where we are.
The idea is that the team focuses on the tactics to be used, as a rugby team would do: “I have problems with that outside”, the receiver could say, and a defender would reply: “I will take care of it”. The team has to get out of the meeting saying “let’s fix this, let’s do that”.
There are three Japanese words for waste, and all three are taken into account in the Toyota Production System : walls , waste due to unreasonableness, walls due to inconsistency, mud due to results.
A waste that tends to be underestimated is that caused by the lack of focus.
People don’t do multiple things at a time because they do multitasking , but because they are distracted. Scientist Harold Pusher demonstrated in the early 1990s what is called “double task interference”: the experiment required pressing a button when a light was turned on.
The time doubled if asked to press a button or another depending on the color of the light that came on.
Magnetic resonance imaging shows that thinking is not simultaneous, but serial: the brain switches from one task to the other continuously.
It is essential to remember that work in progress is wasteful: you think you have to rake the garden, unload the groceries from the car, pay the bank transfer. If you start doing everything together, going from one task to another, when you have done them all partially, you will not have created any value.
James Womack, founder of the Lean Enterprise Institute at MIT, says the waste due to the reworking (rework), with his analysis of the production in the luxury car sector in Japan, Honda, Nissan and Toyota had an average of 16.8 hours manufacture a luxury car with 34 defects for every 100 cars; in Europe, Mercedes, Audi and BMW showed other numbers: 57 hours and 78.7 defects per 100 cars.
The secret lies in the power that every employee on the Toyota production line has: he is authorized to stop and everyone rushes to the stop point to solve the problem that has arisen there. If it did not do so, the flaw would extend to hundreds of vehicles.
The Toyota method says: solve it once and for all.
In the Mercedes factories, however, at the end of the production line, a team in a white apron works with the task of solving the problems, perhaps finding the same defect on dozens of vehicles.
The German factory fixes the newly built car, the Japanese factory brings the perfect car out of the production line.
In a time-based system, estimating activities is important. In Scrum the relative dimensional estimate is used, that is the dimensional comparison between two tasks.
Various methods can be used to help people estimate the work, one of the most popular being the size of dogs. One wonders: is this problem a dachshund or a Dane?
Each breed is assigned a numerical value. A dachshund is worth 1, a labrador 5. Values are assigned using the Fibonacci series, which is everywhere in nature and therefore instinctively understandable. It is a trend in which the next number is the sum of the two that precede it 1, 3, 5, 8, 13 …
The values are far enough apart to allow us to understand the difference instinctively. Our mind does not consider marginal increases, we are better at perceiving clear passages. To estimate a task, it is easier to determine if it is a 5 or an 8 than to distinguish between a 5 and a 6.
Understand the story before defining what to do
With Scrum it is decisive not to start from what there is to do, but from the understanding of the history behind a project. The first questions asked by the team are: for whom do we do this job, and why? Once this is understood, it is clearer to define what should be done.
It is the Product Owner (product manager) to prepare the list of specifications in a list called backlog , transforming them into stories: the description of the activity to be done becomes the story of a need, of a problem that must be solved, for example : “As a user of this game app, I want to be able to access the list of the last games I’ve played”.
The team analyzes this problem and translates it into as many activities as there are professionals involved: from the designer to the UI (User Interface), from the database developer to the editor.
At the end of the Sprint, the Product Owner will decide whether the request made through his stories and transformed into a Sprint goal, has been fulfilled and if there is a “deliverable” value to the user. Once this is done, the cycle starts again.