The Wisdom of Crowds Book (Part-2)
The Art of Decentralisation
After the attack on Pearl Harbor, America needed a strong Secret Service. Many generals were in support of creating a central body which would have complete control of every activity related to intelligence. Therefore, the Central Intelligence Agency (CIA) was created in 1947.
But, the American Intelligence Agency was never actually centralized. In addition to the CIA, there were several independent secret services, including the FBI (Federal Bureau of Investigation), NSA (National Security Agency), and DIA (Defense Intelligence Agency). All three branches of his military also had their own separate intelligence operations.
These independent groups were decentralized. This means that they were not controlled by any single authority. America’s different intelligence agencies rarely worked together. However, the purpose of all of them was to protect against any attack against America.
In the 1980s, decentralization became a popular concept. This was supported by the growth of the Internet, which is the world’s largest decentralized system. File sharing services like ‘Napster’ started shaking the centralized industry. Now companies started focusing on self-managed teams instead of traditional corporate structures.
- Advertisement -
The main strength of decentralization is that it promotes independence. In this, specialized people can work alone on a problem and then coordinate with others. However, its disadvantage is that some information never reaches that community, so it is very important to have a system to collect and share information about new things.
The Linux operating system is a great example of this model. Finnish hacker Linus Torvalds developed Linux in 1991. This was their own version of the ‘Unix Operating System’. Then, Torvalds released Linux for free to the public. He wanted some suggestions and improvements for his code.
5 out of 10 people who downloaded Linux reported improved code and bug fixes. Over time, thousands of programmers started adding features to Linux. Today, it is the biggest challenger to Microsoft’s Windows operating system.
But as is the case at Microsoft, programmers are not paid to improve Linux. That means if there is a software problem, it is not fixed until someone volunteers to fix it. Linux is a decentralized system. It has no official organization and anyone from any corner of the world can suggest improvements in it.
Linux has been able to work so well because decentralization promotes diversity. A company gives projects to its employees. When he works on problems he manages them. Therefore, Microsoft employees are able to examine only limited solutions.
Additionally, Linux has a surprising number of volunteers. This means that there is a greater chance that more people will be working on a problem. It will not be beneficial for any company to employ thousands of programmers to solve the same problem. But, having many independent thinkers is the perfect way to find the best solutions.
But how does Linux collect that information about who its volunteers are? Even though it is a decentralized system, all the codes reach the same group. Torvalds himself leads this team of programmers. It itself accepts or rejects any improvement added to Linux.
The method of aggregation (collecting and combining information) is necessary for decentralization. Imagine a group of programmers working on a problem. Without aggregation, the best solution would be the one found by the most intelligent programmer. But aggregation finds the best features of all the different solutions available.
The example of Linux proves that for decentralization to work, certain conditions must be met. Regretfully, there is no aggregation system in America’s Intelligence Community. They are failing to anticipate major terrorist threats, such as the 9/11 terrorist attacks.
Decentralized organizations are powerful on their own, but they do not have any mechanism to share information. Centralization was not the solution because larger groups are always more intelligent. But, if different data is not brought together and aggregated, then no crowd or group can be successful.
Science: Collaboration and Competition
What do you think the process of scientific discovery would be like? Most of us think that great inventions are made by a single genius in his lab. In fact, scientists mostly work in teams.
Since the mid-20th century, solo scientific projects have become even less frequent. It has become normal now for 10-20 people to write research papers. In 1994, credit for the discovery of the top quark was given to 450 physicists. Collaboration for scientific research has become a global standard.
Just look at the SARS virus research project of 2003. Between November 2002 and February 2003, 305 Chinese citizens became strangely ill. He had symptoms of cold, but the test showed that it was some other disease. Because some patients started dying, WHO (World Health Organization) started investigating that virus.
11 research laboratories joined this project. These were around the world in Europe, East Asia and North America. These labs worked independently but met daily to discuss what they discovered. The results of each lab were posted in the WHO database.
Many labs had studied the same samples. This thing improved the speed and efficiency of his analysis. This project started in 2003 on 16 March. Five days later, a lab in Hong Kong isolated some samples of the virus. By April 16, they had identified the virus responsible for SARS and confirmed it in animal trials.
The SARS project was a huge success in medical research. Labs that work alone take months or sometimes years to identify and isolate new viruses, but in 2003, these 11 labs completed this challenge in just one month.
The interesting thing is that there was no one in-charge of the collaboration of these labs. Even though WHO had organized this project, they did not give instructions to the labs. He had organized himself in an independent manner. The result was that the labs found the most efficient way to divide the work among themselves.
This example shows why scientists collaborate with each other. It reduces the burden on one person by dividing the research work. Moreover, with time many branches of science have become highly specialized. It is difficult for a single person to understand all aspects of a project. Collaboration helps scientists combine their unique skills and knowledge.
Joint projects also have a chance of being more successful because they increase the diversity of the research team. Having different viewpoints means more new ideas will be tested. In the case of SARS, each lab initially had a unique theory. With more groups, the chances that one of them would detect the virus had increased.
Going deeper, we realize that collaborating with others is the only way to be successful. A study conducted in 1996 revealed that researchers who made the most discoveries were the ones who collaborated most often.
Scientists do not run after new inventions because they care about society. Usually, they are most concerned about getting recognition for the research they have done. But ultimately we only benefit from this sailfish behavior. To get recognition, a researcher has to publish his work publicly. Every new project makes the scientific community smarter than before.
Just like the SARS team, scientific research centers are generally decentralised. For example, there are successful government research programs like the Manhattan Project. However, scientists are generally not given directions for this.
So, what is the system of aggregation in science? The scientific community reads newly published research and then decides how it should be used. However, this highlights a disadvantage of the decentralized structure of science.
Most published research is rarely read. Only those works which are published by well-known scientists are seen the most. When famous scientists work in a team, they get most of the credit. The scientific community acts like an intelligent crowd, but they are not using their full potential.
Maximum collaboration should be done to fix this problem.
Committees, Juries, and Teams
The Columbia Space Shuttle made its 28th flight on January 16, 2003. When it reached the atmosphere, a large piece of foam broke off from Columbia’s fuel tank. Before falling down, the piece of foam hit the left wing of the Space Shuttle.
A group of engineers was called to study the damage caused by that accident. But, the evidence obtained from the camera was not clear enough to reach any conclusion. The engineering team informed NASA that the damage could be very high.
The Mission Management Team (MMT), led by Linda Ham, was responsible for the flights. When Hamm received the damage report he decided that the foam impact was not a big deal, so he said at the meeting that even if the wing was damaged, there was nothing he could do about it.
On February 1, Columbia returned from space to Earth’s atmosphere. Due to the hot air, the heat shield on the left wing started burning. Then the shuttle broke into two parts and 7 members present in it died.
MMT is an example of how small groups should not work. These lessons are important because such teams are an important part of our lives too. Just as juries decide whether a person is guilty or innocent, boards of directors make company strategies. Most projects are handled by teams.
One of the main features of small groups is lack of independence. People are affected by stock market prices. But, they do not have any deep relationship with companies or other traders. Small groups are different from this. Even if it is temporary, each member takes on the identity of the group. They all have a huge impact on each other.
The disadvantage of this type of group is that they can take very bad decisions. Without independence, he cannot use the intelligence he receives from the crowd. However, the relationship between them can encourage group members to perform better. In well-functioning groups, people work harder and smarter.
Let’s go back to the example of MMT. Their first mistake was that everyone followed Linda Hamm’s suggestion. Ham made a mistake by starting the discussion with results. He had already decided that nothing could be done about that accident.
Moreover, MMT did not use all the information it had. He ignored all the warning signs given by the engineers. One of the main reasons for this was that MMT was not very diverse. All its members were from the same academic background. There was no one in it who could challenge the group’s decision by thinking differently.
Many studies show that diversity is very important for the success of a group. A jury consisting of people who do not agree with each other on everything takes decisions more thoughtfully. In this case, the minority opinion is not necessarily intelligent or unbiased. Just by being present in the group, they make the group smarter than before.
The second feature missing in MMT was an aggregating system. Even though team members had different ideas, they were not expressed in meetings. During every discussion, Haim used to dominate them and ask questions to the members. Instead, he would have gotten more done if the team voted on each issue.
As you can see, small groups have many disadvantages. This makes us think whether we should remove them. The work of a small group can be given to a trusted person.
But, experiments have shown that groups can take intelligent decisions more quickly. Alan Blinder was the Vice Chairman of the US Federal Reserve Bank. His committee was often accused of not being efficient.
After retiring, Blinder designed a study with the help of economics students. He asked them to play the role of the Central Bank. According to unemployment and inflation, the interest rate for those students had to be increased or decreased. On average, groups made better decisions than individuals alone.
Also, Blinder found that the success of a group was greater than that of the smartest individual, so the performance of an intelligent economist is improved by being in a group. This tells us that in many cases group decisions are more valuable. However, for the group to function better, certain conditions must be fulfilled.
The Company
Corporations organize large groups of people to provide products and services. They should also be designed efficiently to reduce the costs involved in reaching their targets. But beyond these factors, there are no limits to how a company can operate.
However, in the early 20th century, the structure of a company was clearly defined. First, it was designed so that the corporation could control every step of its sales process. For example, Ford controlled the mining of iron ore and sand used in manufacturing.
Next, a corporation was built according to a hierarchy. This means that there was a ranking system among the employees. In management, those who were at higher levels had the responsibility of those at lower levels. More complex problems were sent to higher levels.
Eventually, all corporations had to be centralized. This means that the company’s head-quarters controlled the operations of each division. Only the central body could take decisions on major strategy. Often, this power rests with the CEO.
Because of these features, many American corporations have not been able to harness the wisdom of crowds. Employees’ ideas had to pass through many levels to reach the boss. Every manager would make changes to that idea to increase its chances of being accepted. In the management of Ford’s factory, there was a distance of 15 levels between the supervisor and the chairman.
Let’s look at an example of a new headlight design at General Motors. 15 meetings had to be held to approve this change. The CEO was present in 5 of these meetings. Even though this company promotes discussion, it was still very inefficient.
To utilize group intelligence in a corporation, its employees must work independently. However, corporate incentive systems generally do not encourage such behavior.
For example, junior employees want to be promoted, but they cannot do so by going against their boss. Therefore, they hide bad information like employee budget issues and product failures.
Regardless of intelligence, no one can make good decisions without honestly obtained information. But, almost all corporate incentives encourage employees to be dishonest. Let us look at the example of bonus given to employees.
A sociologist named Donald Roy worked in a machine shop during the 1950s. There the workers were paid according to their production rate. When he accomplished a specific target, he was given more money for each part of the machine.
However, the workers did not want to work very efficiently because then the management would increase the target. The result was that workers did not focus on working as hard as possible. Instead, he focused on increasing his bonuses while reducing his work.
When you ask managers to reach a target, their behavior becomes even worse. First, the manager will at least try to set targets. In this way he will easily get the bonus.
Once the target is set the manager will do whatever is necessary to get there. This includes manipulating accounts and giving false reports.
An effective way to motivate employees is to give them stock options. This means giving shares of the company to the employees. This makes him feel as if he is also the owner of that corporation. The result is that they work independently to make the best possible choices.
An employee alone cannot do anything to increase the value of the company. However, handing out stock options is a powerful way to motivate. It has been seen that this increases corporate profits and productivity.
A corporation can also become more efficient through decentralization. Let’s look at the example of Toyota. While Ford had 15 levels of management, Toyota had only 5 levels. During this time, he introduced the famous ‘Toyota Production System (TPS)’.
Earlier car manufacturers used an assembly line. In this, each worker used to assemble a single part in the car and did not know anything about the entire machine. TPS’s focus was to teach its workers more skills. He had to understand every step of the production process.
TPS created a team of mechanics who put the car together. Each team could develop its own production techniques. The result was that Toyota’s manufacturing facilities became more efficient than its competitors.
To utilize group intelligence, corporations must reduce the influence of the CEO. The 20th century was filled with many CEOs whose companies went bankrupt. This happened not because he was a bad leader but because a single person cannot always be right. When only one person is taking all the decisions then the failure of that company is certain.
Instead, companies should aim to aggregate their employees’ input. Hewlett-Packard tested it in the late 1990s. He called employees from different divisions to casually estimate the sales of his printer. At that time, 75% of the time the group’s forecast was better than the company’s official forecast.
Conclusion
First of all, you learned about crowd intelligence. Large groups made up of average people can outperform experts. There are many examples of this in real life, ranging from game shows to the Internet and the stock market.
Second, you learned about the importance of diversity. Diverse groups develop more ideas, increasing the chances of getting the best idea. Diversity also reduces destructive behaviors like “groupthink.”
Third, you learned how independence improves a group’s decisions. This prevents people from making the same mistakes and helps them use unique information.
However, it is impossible to eliminate the habit of copying others in humans. In the information cascade we follow the crowd, but imitation can also be helpful if done intelligently.
Fourth, you learned about decentralization. This concept promotes local and specialized solutions. But, without a good aggregation system, decentralization will fail. Open-source projects like Linux follow a decentralized model.
Fifth, you learned how group intelligence is used in scientific research. Scientists these days rarely work alone. They collaborate to form high-skilled groups. Researchers work independently and publish their research. Then the scientific community aggregates or collects these new discoveries.
Sixth, you learned about the harms caused by small groups and how to combat them. You have to reduce the influence of the group leader and promote independence. An aggregating system will also have to be introduced to collect ideas. Maintain diversity to promote ideas that go against each other.
Seventh, you learned about the use of group intelligence in companies. The old hierarchical corporate structures are no longer efficient. Only the CEO should not have the power to take decisions. Companies should also develop incentives that encourage employees to think and work independently.
Group intelligence is not an effect that we find in any experiment. It is an expression of human potential that can have a positive impact on our lives. Unlock the power of diversity, independence, and decentralization in your team. Today, only the wisdom of the people in the crowd can help solve our biggest challenges.
Anyway, see you again with the powerful explanation of the new movie.
