Although the Cold War was clearly a frightening time, much of the uncer- tainty in the 1950s and 1960s consisted of statistical uncertainty, which can
be defined as variation observed in repeatable phenomena such as defect rates, resupply time, or product development cycles.30Repeat the activity enough times, and the chances of getting the same outcome can be expressed in clear statistical probabilities such as the Six Sigma goal of only 3.4 defects per million (or 99.9997 percent perfect). As the authors ofThe Six Sigma Way suggest, there is still enough statistical uncertainty out there to justify action:
Six Sigma initiatives have tallied billions of dollars in savings, dra- matic increases in speed, strong new customer relationships—in short, remarkable results and rave reviews. Are these results for real? And is it really possible for you and your business to achieve some of the same gains? The answer is “yes.” It can happen to any type of business and, contrary to many people’s fears, you don’t have to have an in-depth background in statistical analysis.31
Even as organizations attack these statistical results, which some label as risk, they must also deal with state-of-the-world volatility. Unlike statistical uncertainty, which can be measured through quantitative methods, state-of- the-world flux resides in phenomena that are by definition immeasurable.
They are often also unobservable.
The two are roughly analogous to puzzles and mysteries. Like statis- tical uncertainty, a puzzle can almost always be solved with existing infor- mation—indeed, the answers are often provided at the bottom of the page or back of the book. Some answers might involve little more than a good guess, but the answer is out there somewhere.
Like state-of-the-world uncertainty, mysteries are sometimes unsolv- able even with massive amounts of information. By definition, as RAND’s Gregory Treverton writes, puzzles only arise after an event has occurred.
“The missiles have been built, with warheads and accuracy that may remain unknown even though they are knowable. War plans have been framed, and the attack started, though it may still come as a surprise.”32In contrast, many of the most interesting mysteries are not only unknowable at this time, but their eventual answer is a mix of hope and fear. “We often care most about events we hope to influence, or we hope to influence them because we care about them.”
The two types lead to very different organizational actions—statistical uncertainty leads toward clear effort to reduce variation, while state-of-the- world concerns should generate efforts to increase organizational responsive- ness to threats and opportunities. “In peacetime, the main kind of uncertainty is statistical, which assumes that each individual component has a failure
mode, and they all depend on one another,” Camm explains. “There’s this diffuse noise in the system, but if you look at it from a macro level, the noise is pretty constrained around a small band of variation. So you can develop good ways to estimate the performance of the system as a whole.”
In wartime the models begin to change, if only because there is an obvious change in the state of the world. “In peacetime, you don’t actually test the software and the hardware that’s used to defend against electronic attack,” Camm continues. “Otherwise, you shut down your radio stations. As a result, you never see any failures in that equipment during peacetime.
When you go to war, there are failures all over the place. Not only that, the failures all occur at the same time because you never actually tested them during peacetime.”
Even wartime volatility can be contained, however, if the war remains the same. But as the U.S. Army discovered early in the Iraq War, tank treads had a disturbing tendency to wear out. “We had this parame- ter that said tank treads last this long in combat. That was one parameter.
The other parameter said tank treads are going to spend this much time on the road.” The more time the tanks spent on the road protecting con- voys to and from Baghdad, the more the statistical uncertainty changed into state-of-the-world uncertainty. Every prediction said the tanks would be off the roads immediately after Saddam’s palaces fell, but the world would not comply.
State-of-the-world change clearly comes in different sizes, including what James Dewar, the director of RAND’s Center for Longer-Range Global Policy and the Future Human Condition, calls deep uncertainty. “There are a lot of times when we know how the system works but we don’t know what numbers to plug in,” Dewar explains. “And then there are times when we don’t even know how the system works. There, you’ve got deep uncertainty.”
Deep uncertainty is not necessarily unmanageable. Indeed, RAND’s Robert Lempert, Steven Popper, and Steven Bankes maintain that humans often do well when they confront deep uncertainty in their own lives, espe- cially if the intuition about the system in question works reasonably well.
However, intuition and what-if thinking often fail when humans confront novel situations or extensive amounts of information. “In such situations,” the three authors write in Shaping the Next One Hundred Years, “humans rapidly lose the ability to track long casual links or the competing forces that may drive the future along one path or another.” Hence, RAND’s concern with developing decision tools that help humans develop plans that do well across multiple futures.
The question for the moment is not whether humans and organiza- tions can harden themselves against assorted levels of change, however.
Rather, it is whether uncertainty has risen to the point where it deserves special attention in organizational design. If the level and mix of uncertainty has remained relatively constant over the past 50 years, for example, per- haps organizations can just accept occasional vulnerabilities based on igno- rance, inflexibility, indifference, and inconsistency as a normal cost of doing business. If, however, the level is rising and the mix is changing toward greater state-of-the-world conditions, then organizations might consider a robustness auditto see just how tough they are.