#4: Strategic uncertainty

Building the uncertainty mindset for pleasure and profit; sources of uncertainty (again); how to be in an uncertain world.

I’m Vaughn Tan; my research focuses on understanding individual and collective responses to uncertainty and I teach strategy and design at University College London’s School of Management. The Uncertainty Mindset is a weekly newsletter about uncertainty in individual and organizational life. Ideas are likely to be half-baked.

A newsletter reader—let’s call him/her X—who’s a senior researcher at one of the Big Tech Firms emailed me after last week’s issue to say that,

Actually most of the things I’m doing right now professionally are handling situations of high uncertainty on behalf of those that are very uncomfortable with high uncertainty. I can write code and train ML models just fine, but that seems to be a common skill at [redacted]. Lots of folks just do not want to make hard calls in public or deal with ambiguous and emotionally charged situations. So I find myself doing more of that.

It will soon become a tired trope of this newsletter that our default modes of education and organization are moving us individually and collectively in the direction of being unable to handle uncertainty and of denying its existence—even as the world becomes more uncertain around us. The upshot is that being able to recognize and handle uncertainty is both important and strategic for both individuals and organizations.

One sensible premise of strategy is that rare and useful resources are valuable. These two dynamics—increasing uncertainty and decreasing capacity to manage it—mean that the capacity to recognize and manage uncertainty is already rare and will become more useful over time. This capacity is linked to what I call an uncertainty mindset: one which recognizes uncertainty and can handle it with facility. An ancillary premise of strategy is that if something rare and useful is also difficult to imitate or substitute, it is sustainably strategic. Developing the uncertainty mindset isn’t easy, and nothing can really replace it. Individuals and organizations which have (or which develop) the uncertainty mindset will have a strategic advantage that will continue to grow.

I’ve only seen the uncertainty mindset develop in people and teams who have been forced to routinely confront uncertainty. In organizations, this can happen when an employee is promoted into a position of management or profit and loss responsibility, or when a team is given a minimally parameterized R&D project. (Or in education, when the assessment is not a problem set with definitely and unambiguously correct answers.)

Back to X, who is handling uncertainty on behalf of those who aren’t capable of doing it themselves. His/her experience is unlikely to be unique. In every organization that is exposed to external uncertainty (as is inevitably the case in the technology industry), a small group of often isolated employees eventually realizes that what they are doing is soaking up uncertainty for everyone else. That they have the uncertainty mindset is potentially good for them because they gradually become more valuable and powerful but is also bad for the organization. If uncertainty is growing, what organizations need is for as many employees as possible to be able to see it and deal with it.

What can organizations and managers do to “train” employees and team members to be better at recognizing and dealing with uncertainty? The quotes are there because training the uncertainty mindset won’t happen in conventional corporate training, where employees get pulled out of routine work to doze in conference rooms during 3-day workshops featuring thick binders of photocopied course material. Instead, it will probably happen through re-tooling how organizations actually work. The obsession with certainty and predictability is so fundamental that it permeates our default ideas about how to organize people. The way to fight that is by changing that default idea of how to organize—and this re-tooling can be done in small, very manageable ways.

One example is changing how organizations (for that, read: managers and HR) think about jobs and the responsibilities they entail. The default is to hire or promote people into roles that are described in great detail. These job descriptions define what each individual’s responsibilities are; they implicitly and explicitly crystallize job roles. This way of thinking about jobs as pre-defined and stable applies to the vast majority of employees. But thinking of jobs like this naturally limits how adaptable employee roles can be.

Job adaptation—modifying what people do to respond to changing demands—is essential when the environment is uncertain. When people in a team face environmental uncertainty that affects their work (for instance, an R&D team at IBM developing a quantum computer while there is still disagreement in the field about how the underlying basic technology works), they have to change their plans as new information emerges (for instance). This means that what they do should change over time as well. When the environment is uncertain, employee roles must be partly unscripted and open to improvisation to accommodate the situation as it develops. Yet the default way organizations think about jobs and hiring makes this very unlikely.

One actually implementable approach managers can use to encourage ongoing adaptation is to explicitly set team member expectations that some (probably large) proportion of what they are individually responsible for is pre-defined, but the rest of their roles are up to them to collectively figure out. The by now well-known percentage time programs at Google and other progressive technology companies (which may have started decades ago at 3M) are a variant of this. At Google, 20% of engineering staff time, nominally a day a week, is given over to projects that they can self-define and which aren’t part of their assigned work. But it’s important to not stop at giving employees free time for exploration. The essential—but almost always missing—second part is to both set the expectation and create the work contexts in which this exploration jointly optimizes for both the employee’s and the organization’s interests.

Counter-intuitively, I have most often seen this done right in high-end cutting-edge restaurants, and especially in their their R&D teams. Members of these teams work with three jointly held explicit expectations:

  1. That their roles are part-fixed and part-emergent,

  2. That they have to experiment to figure out the emergent parts of their roles, and

  3. That whatever they propose for these emergent role-parts must pass muster with their respective teams.

This prevents team members from going fully on autopilot—in fact, it forces them to explore new things to work on (as percentage time programs do) while also keeping their teams’ goals in mind as they explore (which most percentage time programs don’t). This balances between ensuring that known tasks are carried out and creating individual and team capacity to detect and respond to emergent and unexpected situations and demands.

Over time, a culture of not-going-on-autopilot emerges in these teams because the individual and collective baseline expectation is that work, to some degree, will not always be perfectly programmed and predictable. The decision (which can be taken by a manager) to set expectations that part of each team member’s role is emergent and to be negotiated can thus have long-term implications for the team and its members: they develop what might be thought of as negative capability. This makes them collectively more capable of dealing with unpredictable demands, and individually more inclined to explore new ways to work—it gradually produces the foundations of an uncertainty mindset. (There’s much more fine-grained and practical detail about implementing this kind of approach to job roles in my book, which should be available in a few months.)

Implementing systems like this that train organizations and their members to develop the uncertainty mindset requires dropping many of the assumptions about organization that we’ve held for so long that they’ve become taken for granted. By the way, my point is not to argue that all firms need to work like restaurant R&D teams. Fine-dining happens to be an industry that until recently was relatively uncontaminated by management school theories—so there was little to prevent these teams from figuring out unorthodox ways of organizing that actually worked. They offer practical insights for other industries about alternatives to these taken-for-granted assumptions about how to organize.


[half-baked] Where does uncertainty come from?

In Issue 1, I suggested that there are likely to be many different forms of the unknown. More than one person wrote to opine that maybe all uncertainty emerges from complexity. Several conversations since then in various media have led me to a preliminary conclusion that complexity is only one source of uncertainty, and that it may be less useful as a way to think about uncertainty than it at first appears. A more practically useful approach may be to think of uncertainty as unknown-ness which can emerge from at least four distinct sources:

  1. Unknown-ness in causality: When the relationship between actions taken and outcomes intended is not known. Complexity can create this form of unknown-ness. For instance, the more complex a system, the more likely that action taken on the system has unanticipated (or unanticipatable) consequences due to the proliferation of interactions between its parts (e.g., central bankers trying to stimulate national economies). Insufficient experience can also lead to unknown-ness in causality (e.g., when a seasoned winemaker can make wine under challenging conditions because she “had a vintage a lot like this back in 1972”).

  2. Unknown-ness in the range of possible actions: When it is unclear what actions are available to be taken. This can result from system complexity (e.g., using LaTeX and all of its various packages), from social dynamics (e.g., believing that your VP doesn’t want you to talk to the CEO about the new product launch), or from imperfect information (e.g., trying to use some of the extensive but nonetheless often totally inadequate LaTeX package documentation).

  3. Unknown-ness in the range of possible outcomes: When it is unclear what outcomes are available to be pursued. This can result from existing outcomes that are not known about (a form of imperfect information e.g., a college applicant not knowing about liberal arts colleges as a category) or outcomes that aren’t known about because they don’t exist yet (a different form of temporally imperfect information e.g., a company not considering a particular hardware infrastructure because it has not yet been launched or developed).

  4. Unknown-ness in desires: When it is unclear which outcomes are relatively preferable (or what level of utility is associated with each possible outcome). This can result from preferences which change through experience (e.g., swearing off the air fresheners you love after having a violent allergic reaction to one them), or discovering new possible outcomes to prefer (e.g., visiting a city so superior to your current home city in every way that it makes you want to move permanently).

The three types of unknown-ness in Issue 1 fit into these categories without much distortion. Each of these sources of unknown-ness is conceptually different, but all of them makes formal-rational decision-making impossible. For me, the existence of uncertainty from any one of these sources is a working definition of a situation representing true uncertainty.

Not unrelated miscellany

  • Jerry Neumann writes on startups and the need to manage uncertainty. Along the way, he explains Knightian uncertainty in straightforward prose. It’s only 6800 words and you should read it.

  • Individuals can force themselves to continually update their job roles. One of my favorite books is Lawrence Weschler’s biography of the artist Robert Irwin. In it, Weschler describes how Irwin repeatedly abandons an artistic medium when he becomes good at using it. Each time he abandons a medium, the process of learning a new medium is an opportunity to explore questions which are newly relevant. It’s a great book.

  • Odysseus represents the archetype of adaptability in the face of uncertainty. One way to read the Odyssey is as a story about a fundamentally uncertain world (lotophagi, malignant sea monsters and whirlpools, unexpected ship destroyers, vengeful gods, personal transformations, etc) which therefore illustrates how an individual has to be to manage that uncertainty. The opening lines provide a capsule description of Odysseus which has been variously translated as “that man skilled in all ways of contending” (Robert Fitzgerald), “a complicated man” (Emily Wilson), “a man of twists and turns” (Robert Fagles)—but his original Greek epithet is πολύτροπος (polutropon), which Tim Power and Gregory Nagy translate as “many-sided.” Odysseus survives in uncertainty by not being stuck in being a single person—instead he has the ability to change the person he presents to the environment as the situation demands.

On loop

Expansive drone music suitable for writing: The Tired Sounds of Stars of the Lid.

If you liked this issue, you could click on the little ❤️️ and share it with people who might like it too. Please send comments, suggestions, and satisfying dog videos to <uncertaintymindset@vaughntan.org>. You can also get in touch on Twitter @vaughn_tan and Instagram @vaughn.tan.