TDWI | Training & Research | Business Intelligence, Analytics, Big Data, Data Warehousing

Data Driven Mad

Most meaningful business decisions are based not only on data but on a combination of fact and opinion.

• By Barry Devlin
• April 12, 2016

"Algorithms will eliminate the last weak link in decision making ... us." So said Rita Sallam, research VP at Gartner Group, in her keynote at the Gartner Business Intelligence & Analytics Summit in Texas in mid-March. I've recently focused in some depth on such algorithms in the context of applying cognitive computing to decision making. In this article, I turn my attention to the underlying concept of a data-driven business, upon which such algorithms operate.

The world of business decision making has been besotted by data for many decades. We regularly hear W. Edward Deming quoted: "without data you're just a person with an opinion." The actual source and date of this aphorism is unknown, according to the Deming Institute. The lack of these basic facts does not prevent supporters of fact-based and data-driven decision making from using the quote to support their argument.

Similarly, in his 2012 book Too Big to Know, David Weinberger points to Senator Daniel Patrick Moynihan's oft-quoted "everyone is entitled to his own opinions, but not to his own facts." A brief search of Wikipedia will suffice to show that the attribution and even the existence of this quote may or may not be a fact. The subtitle of Weinberger's book -- perhaps the longest I've ever seen -- offers his opinion on the necessity of Rethinking Knowledge Now That the Facts Aren't the Facts, Experts Are Everywhere, and the Smartest Person in the Room Is the Room.

Data-driven decision makers turn to the physical sciences to support their belief in facts. There, the facts appear to be indisputable -- at first. The fact-based scientific method dates to Francis Bacon's Novum Organum in 1620. The length of a beam at 24.67cm is a fact, as long as we all agree on the length of a centimeter. Only the most contrarian might argue. Sodium lamp streetlights emit light of wavelengths of 589.0 and 589.6nm. This measured fact is indisputable -- the experiment has been repeated often enough and produced the same result. How one measures the wavelength of light requires a good knowledge of physics. To accept these numbers unquestionably as fact, we must trust the physicists.

As science extrapolates from simple measurements to pronouncements about the past (an asteroid impact 65 million years ago wiped out the dinosaurs) and predictions of the future (average global temperatures will rise by 4 degrees by 2030), it moves from relatively agreed-on facts to more debatable opinions, known also as theories. Their acceptance depends less on the underlying facts than on the authority of those who propose and support the theories. Such authority emerges from the long-standing method of peer-reviewed scientific publishing, and is now being devalued by cheap, widespread, and un-reviewed publications.

Whether in the world of science or business, the Web has changed that dramatically: no alleged fact or opinion is too unusual or too unsupported to publish. The identity or affiliation of the writer may be unknown, hidden, or even falsified. How does one establish or judge the authority of the author? Opinions and facts are thus increasingly hard to distinguish in common usage today. The reality that everyone can publish an opinion about anything on Twitter or Facebook devalues each and every opinion. All positions across the bell curve of opinion from one extreme to the other achieve equal standing by some measure.

In a recent Forbes article cleverly entitled "Without An Opinion, You're Just Another Person With Data," Milo Jones and Philippe Silberzahn argue that "in a world where data becomes plentiful, the risk is not the lack of data, but having too much of it. The only solution to make sense of this sea of data is to have hypotheses, i.e., to have an opinion that will guide research on the mass of data." I have some sympathy with their position and hope we are beginning to see the pendulum swing back from the extreme of data-driven everything.

Sallam's pronouncement quoted above may itself not even be factually accurate. I picked it off the Twitter stream from an attendee at the Summit. She may have been misheard. She may have placed the sentence in the middle of a more nuanced paragraph. Perhaps she spoke ironically. However, my opinion is that the longstanding belief in increasingly data-driven decision making is being amplified by current advances in artificial intelligence and a growing level of hype around algorithmic business operations.

Some more simple and speedy decisions -- mostly operational in nature -- may be based wholly on indisputable facts and thus best made algorithmically. However, in my experience, most meaningful business decisions are usually based on a combination of potentially correct facts, negotiated opinions and positions, and the story the decision maker wishes to believe about the past, present, and future of the enterprise.

As a final thought, consider Malcolm Forbes' wisdom: "Anyone who says businessmen deal in facts, not fiction, has never read old five-year projections."