Bad Decisions Are Contagious

Part of the problem with writing a book is that you become associated with a simplified version of your concept. In "Smart (Enough) Systems," James Taylor and I never claimed that computers were capable of running businesses in an unassisted way, or even that many of the decisions made in an organization can be automated. Our premise was that only those decisions that are high volume and low-risk (on an individual basis) are likely to be improved through decision services. Otherwise, decisions are best left to people. Sort of.The whole field of decision-making and how, or if, we can assist it with analytics is pretty murky. This is a contentious point as promoters of business intelligence and analytics still insist on suggesting they are helping people by "getting the right information to the right person at the right time." It's undeniable that getting this timely information to the right person is useful, but there is no evidence that it is even slightly sufficient. People bring all sorts of experiences, prejudices, flawed reasoning and emotional aspects to the decisions they make. Researchers at Northwestern University recently released a study that shows bad decisions are endemic. Their suggestion is that "companies trying to reverse results of bad decisions should find true outsiders." That's good news for us consultants (assuming they don't bring their own contagiously bad decisions with them), but where does this leave us with notions of "collaborative BI?" Does it make sense for anyone to share anything if their bad decisions are, indeed, contagious? To make matters worse, a study at Wake Forest University School of Medicine found that software that generates standard and uniform reports of radiological test results do not improve, but rather decrease the completeness and accuracy of the reports. Radiology reports are typically prepared in a freeform narrative, but like other areas of analysis, there is pressure to standardize the data so that it can be studied longitudinally. But, as one of the researchers reported, ""It turns out that a structured reporting system actually decreases the accuracy and especially the completeness of reports, which is the opposite of what we expected." She went on to explain, "'Standardization seems simple, but it's not always easy or what we commonly do in medicine. The theory is that structured reporting would make the reports intrinsically better because we'd all be using the same ideas recorded in the same verbiage instead of using numerous different ways to say 'blood.' Right now, several people reading a scan may all agree that they see the same thing, but each individual will say it in a different way." What she is describing is a problem with semantics. By using standardized terms in structured reports, creators of the reports are forced to squeeze their usage and semantics into a rigid framework that doesn't always correlate with their own. This is a common problem in BI reporting, usually dealt with by "taking requirements" and then having the one person or persons who are the most disconnected from the actual process, data modelers, create the data structure that delivers less than a complete picture. There are solutions to this problem but the BI industry has been slow to adopt them, especially various kinds of semantic technology that allow for a much more flexible use of meanings.Promoters of BI and analytics still insist on helping people by "getting the right information to the right person at the right time."... The trouble is, people bring all sorts of experiences, prejudices, flawed reasoning and emotion to the decisions they make.