The algorithmic economy is emerging very quickly not only because we collect a lot of data, but also because algorithms offer fast, fact-based and consistent decision making. Most often, the process is even automated – as in algorithmic trading, for example – thus posing a relevant concern about the role of the human decision maker in such an economy. It’s worth taking a closer look at why algorithms are so appealing.
First, algorithms work extremely fast on very large volumes of data. Therefore, decisions can be made within seconds, and even milliseconds. Obviously, human analytic processes cannot compete with this speed. Second, algorithms evaluate only data, that is, the facts collected and stored in systems, so every algorithmic decision is based solely on the data and is not impacted by subjectivity or the “gut feelings” that can make human decisions prone to errors.
Third, algorithms eliminate the inconsistency in decision making. When presented with the same facts, algorithms will make the same decisions. In contrast, if a group of human decision makers is presented with the same facts, they are likely to make different decisions. This poses a problem for organizations that are looking to standardize decision-making processes across large data sets and significantly sized sales forces in order to achieve efficiencies and scale.
Today, many people argue that every employee should be converted into a data analyst, and that businesses should empower them to perform data analysis with self-service tools. But the algorithmic economy is progressing in exactly the opposite direction, as algorithms are marginalizing the need for DIY analysis.
This indicates an important error in the approach of making everyone a data analyst: We should not strive to make everyone a data analyst. We should aim to make everyone a decision maker.
Today, analytics is equated with data analysis strictly. That means that an analyst gets a particular data set and uses various tools to massage and analyze the data to extract some insights or knowledge from the data. In other words, analysis is viewed as “playing” with data. This can create hype around the belief that all employees in an organization will become knowledge workers, extracting data and analyzing it. However, most employees aren’t in an analysis role. They must perform operational tasks, so involving them in analysis will cause a decline in productivity.
Toyota has captured this in principle No. 7 of The Toyota Way. Inherent in this principle is making everything immediately visible so decision makers can see the facts and make decisions quickly. In other words, do not distract an engineer with analyzing data. Give that engineer the facts from the data so she is able to make decision in fewer than 30 seconds. That is not diminishing the analytical power of the engineer; on the contrary, it is empowering her to use her tacit knowledge and combine her expertise with the facts to make a decision. Hence, we should strive to make everyone a more effective decision maker.
The question about humans’ role in the future of decision making is of huge importance and a philosophical one. Without question, the algorithmic economy will reduce the role of human involvement in the execution of decisions. The driverless car is just one example of the execution of decision making being delegated to a computer. Algorithms will evaluate the facts and present them to the human decision maker, who then will do meta-analysis to connect the facts to the context – the environment and the goals that drive the decision making. But humans will still have a critical role in design decisions, which will not be delegated anytime soon.
This approach is well captured in a Bloomberg BusinessWeek interview with Lyor Cohen, arguably the most successful record company executive in America. In 2013, he spoke to MIT students about his approach to finding new talent. He found acts to sign by monitoring radio stations. According to the article, Cohen said he plans to use the Internet to do this more efficiently and faster. Yet, using algorithms that follow traffic is not enough. As he stated: “All you smart [MIT] people, you could come up with an algorithm, but somebody still has to show up and say, ‘Yeah, I feel that.’ ”
Music is Cohen’s business, which you may think leads to greater emotional context, yet the process is still built on a foundation of data. But it captures the essence of human decision making. As the algorithmic economy emerges as a stronger force, the time of human decision makers is best spent in performing meta-analysis and not in attempting to compete with algorithms to process and analyze data. For those in business intelligence and analytics, the growing reliance on algorithms should push us to rethink how we will deliver information to users in the future.
Dr. Rado Kotorov is chief innovation officer at Information Builders, a business intelligence and analytics provider.
Subscribe to Data Informed for the latest information and news on big data and analytics for the enterprise.