"In God we trust; all others bring data."
– W. Edwards Deming

Bringing an analytical view to business is one of the most significant business trends in the last decade. Almost every job profile requires analytical skills, business intelligence has been at the top of the CIO agenda for many years, and fact-based decision-making is the business model for a complete industry of consultants and analysts. In fact, I used to be an analyst at analyst house Gartner for many years.

Competing on Analytics: The New Science of Winning, by Davenport and Harris, has become a very influential book. The authors argue that outsmarting the competition is the biggest distinctive capability for organizations in the years to come. They provide a roadmap to becoming an analytical competitor, describe an analytical process, provide insights in how to manage analytical people, and spend time on how to define the right technology foundation.

This idea is not new. Frederick Taylor is well known for introducing scientific management in the early twentieth century, particularly in production environments. Most of us will immediately remember Charlie Chaplin's "Modern Times" movie, which took productivity improvements to the absurd. In World War II, operations research came to be, including techniques such as simulation, decision trees, linear programming (maximizing results while working with limited resources), and dynamic programming (e.g., to find the shortest route for deliveries). These techniques are the basis of today’s analytical toolkit.

Originating in the medical world, evidence-based management is now becoming popular. This movement also suggests the use scientific methods for making decisions, but evidence-based management also includes the attention to ethical considerations and a focus on behavioral economics. Behavioral economics breaks with the widespread idea that people make rational decisions as proposed by decision theory, but instead are influenced by a wide variety of circumstances. By better understanding these circumstances, behaviors and decisions can be influenced.

What could be against this approach? CEOs are responsible for their employees and their families; we wouldn't want them to make far-reaching decisions based on their gut instincts. Although intuition, experience and other soft factors may play a role, at the minimum they should be tested and validated before an actual decision is implemented. So how can we make the best of analytics?

Analytical Philosophy

The old philosophers would have loved this direction that business decision making is taking. According to the dictionary, analysis is nothing more than "the separating of any material or abstract entity into its constituent elements" – taking something apart in order to understand what it is, what it does, what elements it has, and how these elements relate. Analytics is the science or process of doing all this. This is a core competence for philosophers. They take fundamental questions or concepts and break them down into smaller questions until there is a certain understanding or logic revealed. IT people would call this process "functional decomposition.”

The love for analysis that philosophers display is only logical. Many of them were mathematicians as well. Aristotle (384 BC – 322 BC) formulated the laws of logic. Gottfried Leibniz (1646-1716) and Sir Isaac Newton (1643-1727) invented calculus independently of each other. Rene Descartes (1596-1650) was the father of Cartesian geometry. Over the years, philosophy has become increasingly analytical in nature. In fact, led by Bertrand Russell (1872-1970) and Ludwig Wittgenstein (1889-1951), a large part of twentieth-century philosophy is known as analytical philosophy. This branch of philosophy creates formal logic including a notation (language), based on the analysis of language, to create clarity of argument. You could say this formal logic is very much like what SQL is to databases. In fact, in a narrow sense, this school of thought rejects the idea of thinking big thoughts and examining the fundamental purpose of life. Analytical philosophy proponents state that philosophy is nothing else but trying to create logical clarification of thoughts. IT people would call this “requirements analysis” and “building the query.”

Perhaps citing analytical philosophy seems a bit much, but we can definitely learn from the philosophers.

Do We Think, Or Do We Know?

You would think that people promoting fact-based decision making and a more scientific approach to management would base those recommendations on, well...facts. In a recent article in the Journal of the Academy of Management, researchers found in a large meta-study (analyzing the available studies) that the evidence for evidence-based management was poor and anecdotal at best. Or consider another example. Good academic papers always have a "limitations" paragraph in which the shortcomings of a research design are described, or under which conditions the conclusions are valid and under which conditions they are not. Business literature rarely has such disclaimers. On the contrary, the sky is the limit.

Perhaps the best recent case of unlimited skies comes from Wired, an influential American technology magazine. In 2008, editor-in-chief Chris Anderson wrote an article called "The End of Theory: The Data Deluge Makes the Scientific Method Obsolete." Although it wasn't clear if he was trying to be provocative or was actually serious, his central thesis was that in the near future we will no longer have to create models  for analyzing phenomena. With computers already powerful enough, we will simply feed it all the data there is without the need  for any context or hypothesis. The computer will do the statistical analysis and provide an in-depth description of reality instead of a theory. As Anderson noted in the article, "With enough data, the numbers speak for themselves.” In science, we are taught that correlation doesn't equal causality. It is necessary to understand the underlying mechanisms – just a correlation could be a coincidence. However, if we have all the existent data, there is no room for coincidence. Every single case, every instance simply has been covered. We could ask the computer to analyze all the data and see what it shows, what descriptive algorithms it comes up with. Anderson uses examples from physics, biology, and other fields to point out that more advanced technology has already created more precise descriptions of reality for the last centuries. Why not foresee the last step?

The early Enlightenment philosophers would have been on Anderson's side. In the Age of Enlightenment there was an unshakeable belief in technological advancement. Benedict de Spinoza (1632-1677) asked if there would still be a need for God if everything that happens in the universe can be explained in scientific terms. Newton didn't see that contradiction–he posited that God created the universe and now it is up to us to figure out how it works. The metaphor of the universe in those days was that of a machine, albeit an incredibly complicated one. But even the most complex problems can be solved if you unravel them bit by bit.

"Do we think, or do we know?" is a famous quote from Gary Loveman, the CEO of Harrah's Entertainment. He uses this question to test if a proposal is truly fact-based or "just" a good idea. But it raises an interesting question. What do we really know? What is really true? It seems like every philosopher has explored an angle to these important questions, but for the purposes of the subject of this article, I have chosen the most relevant angles. Gottfried Leibniz (1646-1716) identified two types of truth: truths of reasoning and truths of fact. Truths of reasoning simply follow logic, such as if A is true, so must B. If all cows are mammals, then one particular cow must be a mammal too. If there are four apples on the table and you divide them equally among two people, each gets two apples.

Alternatively, based on reasoning we can be sure that certain statements are not true, such as "The deaf person turned around when he heard footsteps", or "I recently had an interesting chat with a dead man.” Truths of reasoning are hard to dispute or deny. This is the analytical world. You create a model, based on one or more hypotheses and/or available data, and start to build in rules for how the model works. The logic can be described and transferred to other people as well. A model can be called correct (or not), and the outcomes can be predicted just by examining the rules, even before actually running it. In other words, truths of reasoning can exist before the event – a priori.

Then there are truths of fact. These are based on direct observation. I could claim there are four apples on the table, but in fact there might be only three. I could claim I lost four kilos in the last three weeks because of a new diet. This can only be observed if I stood on the scale three weeks ago, measured my weight, and stood on the scalejust now. But who is to say the loss of weight was caused by the diet alone? There might have been other reasons as well, such as stress, sports or anything else. And did I lose those kilos in a somewhat linear sense, or did I gain a kilo and lose five over that period of time? Truths of fact can only be measured after the event – a posteriori.

You could object to the use of the terms truths of reasoning and truths of fact. What is a fact anyway? We can observe only through our senses, namely sight, hearing, smell, taste and touch. Most philosophers today seem to agree that as a result there are no real facts, just interpretations. And if you measure through some kind of recording device, such as scales, sensors, computers, video or audio equipment, the observation is limited by the capabilities of the device. It would only measure what it is designed to measure, based on a concept of reality that exists in our minds. It might measure precisely, but does it also measure completely? The British philosopher John Locke (1632-1704), although a product of the Enlightenment, also agreed there is always room for error. Even if a phenomenon is observed by multiple people, it can still be wrong, as people's senses are limited.
 
Locke’s observations also invalidate Anderson's reasoning in the “End of Theory” article, notably based on Anderson's own arguments. Locke observed that even carefully constructed knowledge over the centuries turned out to be untrue, or imprecise. As insight grows and technology progresses, we simply discover more layers of complexity. As Anderson describes, quantum mechanics corrected Newtonian models and there are many discoveries to be made in the field of genetics. Perhaps a simple example is our insight that the earth does not contain only four elements: air, fire, earth and water. Who can truthfully claim that if computers capture all the data, it is really all the data? We just don't know. And even if we did, data still doesn't equal reality. The best we can claim is that results are probable.

So truths of fact are not that factual. At least we still have truths of reason, right? But in the end,  a model is always an abstract of reality. And even if we could construct the perfect model, as Anderson claims, the outcome of running such a model can only be as precise as the data in it – the truths of fact – and these cannot be validated.

In short, truths of reason can be true, but can be based on false assumptions. Truths of fact can never be true, only probable. Reality has its own unpredictable ways. As the joke goes among economists, "Reality is the exception in the model." So the reliability of data collection and data analysis is problematic. After careful analysis and weighing all the facts, the only answer we can give to Gary Loveman's question is, indeed, "I think.” What is there to know, really? No wonder philosophers are not the most popular guys on the executive floor – before you know it, you can't rely on anything anymore.

Recent articles by Frank Buytendijk

• Machiavelli and Sun Tzu on IT Governance
• Plato and the Art of IT Governance
• Solving the One Version of the Truth
• The Myth of the One Version of the Truth