Strategy and Uncertainty

Publish date: 2020-12-28

Some notes from The Uncertainty Mindset, by Vaughn Tan. All errors and omissions mine. Anything interesting is Vaughn’s.


‘The unknown is existentially threatening, which is why people and organisations act both instinctively and rationally to reduce or avoid it.’

‘Where the future is uncertain, people and organisations have the freedom to influence what it becomes.’

The Uncertainty Mindset, Vaughn Tan.


What do startups, test-kitchens, and applied scientists have in common?

They all share the uncertainty mindset.

The uncertainty mindset has been one of the central ideas of my 2020. It’s a state of mind, and also a newsletter and a book, by Vaughn Tan. As a happy result of corona-time, I was lucky enough to work with Vaughn a bit this year — I helped teach his class Strategy by Design and I got to read his book. It was so good I read it twice.

It’s a study of the most innovative restaurants in the world, and what makes them the best. Vaughn argues that these organisations are great at innovation not because of any innate talents, but because the people in the organisations have a mindset better suited to innovation (and failure) than everyone else.

Though Vaughn’s learnings come from the restaurant industry, the insights he draws are general. They’re applicable to small teams of people in any field, who are trying to do something new — especially ‘disruptive’ startups and applied research teams in big organisations.

In this post I’ll summarise some parts of UM, in the context of doing strategy under uncertainty. First I’ll quickly recap a conventional view of strategy, before showing how this conventional view is impaired by uncertainty. Then, I’ll go on a whistle-stop tour of some of the book’s recommendations on how to design teams to handle uncertainty.

Off we go.


Strategy, a homebaked definition

Strategy is the process of working out your goals, and then working out a set of actions to take in order to achieve your goals.

Your goals depend on your values: the set of things that you believe are important and won’t compromise on. Your values provide a framework from within which to choose goals. If you value money and the respect of your work colleagues, your goal might be to get promoted as soon as possible. If you value family and your health, your goal might be to make enough money to adequately survive, while having enough time to work out and play with your kids.

Aside: It is important to articulate your values. Otherwise you end up with goals that aren’t in line with what you really care about.

The actions you can take depend on your own abilities, your resources, and your environment. They depend on the things you know how to do, can learn how to do, and the things your environment constrains you to do at the moment. You can only make lemonade if life gives you lemons; and you have enough time, sugar, and a vessel for the freshly squeezed juice.

The conventional view of strategy emphasises gathering information and planning. If you can build an accurate map of your environment, your resources, and your current abilities, you can identify all the possible actions available to you. You can break an overarching goal like ‘be the most profitable lemonade stand in the neighbourhood’ into subordinate goals, like ‘source the cheapest lemons’ and ‘put the stand on the street corner with the most traffic’. Then, picking the ‘right’ set of actions simply involves choosing those which advance you furthest towards your goal.

A ‘right’ answer implies wrong answers. Once you’ve picked the ‘right’ set of actions, you can go about eliminating all the ‘wrong’ ones, becoming lean and efficient, adapted to your environmental conditions like a prized racehorse.

The conventional view of strategy works well when the environment is static. But most environments aren’t — even the rules of horse racing change every now and then. When your environment changes, the ‘right’ set of actions changes with it. And in an uncertain environment, there isn’t a ‘right’ action at all.

Uncertainty is not Risk

The conventional view of strategy has an assumption hidden within it: It assumes the future is risky, instead of uncertain.

A risky future is one you can plan for. While you can’t know for certain what will happen, you know all the possibilities that might happen, and you can estimate how likely each one is. There is an optimal set of actions, and the better your information about the possible futures, the better the actions you will take.

An uncertain future is one you can’t plan for. Not only can you not know for certain what will happen, you can’t even know all the possibilities that might happen, and you also can’t know how likely each possibility is. There is no ‘right’ set of answers, but there are still lots of wrong ones.

Imagine you’re preparing for a weighty end-of-year exam. If you know the exam questions will be recycled from prior years’ exams, you can look at past papers, identify the questions that came up the most, and spend time on each question in proportion to its relative frequency. You can take a risk-based approach. But if the course is new, then you have no past papers to look at. You have to master the material as best you can, in preparation for questions which could be about anything. You must reconcile yourself to uncertainty.

Risky environments have countable possibilities whose likelihoods you can estimate. Uncertain environments have uncountable possibilities whose likelihoods you can’t even guess.

Uncertainty is scarier than risk, and most of human civilisation is built to insulate us from it. But uncertainty also contains opportunity: if the future is uncertain then you have the potential to change it. Strategy under uncertainty is an attempt to do just that.

Strategy when you don’t know

Most realistic environments are uncertain, not risky. E-commerce adoption seems to be following a nice steady uptick, until coronavirus comes along. Mortgage-backed securities are a safe, high-yield bet, until they aren’t. As noted philosopher of uncertainty Mike Tyson once said: “Everyone has a plan, until they get punched in the face.”

The real world is far more detailed than the best models we can make of it. It’s impossible to take account of everything. The very act of making a model of reality is choosing which bits of reality to focus on, and which variables to ignore. Run an experiment long enough, and those ignored variables will change your system, and what you thought was a great map will no longer represent the underlying territory at all. Uncertainty is unavoidable.

In UM, Vaughn recalls a conversation with a chef at Amaja, a three Michelin star restaurant known for using local Patagonian ingredients in entirely new ways:

‘He said there hadn’t been any strategic plan from the beginning. Amaja had been driven by the realisation that whatever they did would likely be relevant for only a brief instant, and that the key was to get things “right for now”, because “everything would change anyway”.’

The restaurant industry is highly uncertain. It’s driven by the incessant experimentation of fame-seeking chefs, who serve an increasingly demanding clientele composed of critics, connoisseurs, and social media influencers. Diners want novelty, and reward quality innovation whenever they taste it. In such an environment, it is tough to plan for one year, and you can’t even see the three to five year time horizon which most organisations are accustomed to.

When the future is uncertain, you have to throw your plans out the window.

In an uncertain environment you can’t break down an overarching goal into subordinate goals, because you don’t know what needs to be done before doing it. The best you can do is pick a guiding overarching goal and start following the most promising path to meeting it.

In an uncertain environment you can’t identify all the actions you’re able to take, let alone work out which one is ‘right’ or ‘wrong’. The best you can do is try something, learn from your attempt, iterate, and improve.

In an uncertain environment, there isn’t one ‘right’ thing to do. Instead, there are several roads that lead onward, and many more dead ends. The best you can do is to stay alive and keep on walking.

In an uncertain environment you can’t know what the ‘right’ targets are, so motivating your team with performance-based incentives doesn’t work. You can’t reward the winners if you can’t define what winning is. The best you can do is to leap into deep water, and work out how to swim.

In an uncertain environment, there is no advantage to efficiency, because you can’t know what to be efficient at. Being very efficient at the wrong thing is a sure path to a dead end. The best you can do is to stay nimble, build your capacity, and wait for an opportunity.

The organisations in UM follow all these principles. They didn’t just survive uncertainty, they thrived in it, and they’re still thriving today. In the next section I’ll explore how they do it.


The uncertainty mindset, a few key insights

Vaughn identifies six general principles which lead to more innovative organisations:

  1. Modular roles lead to adaptable team members
  2. Continually adjusting roles leads to adaptable teams
  3. Choosing open-ended goals permits innovation
  4. You learn how to pursue open-ended goals by doing real work
  5. Innovation work is motivated by well-chosen sticks better than carrots
  6. Carefully designed progressive overload forces organisations to adapt and innovate

I’ll expand on each one in turn.

1. Modular roles lead to adaptable team members

There is a well-worn hiring process in most corporate jobs: identify the gaps in your organisation’s competencies, write down a detailed job description for a role which will plug those gaps, then, after a rigorous application process, hire the candidate best suited to the role.

This is not at all what happens at the restaurants studied in UM. Instead, if someone is particularly keen to work with a restaurant and his skills seem complementary to the current team, he’s hired for a provisional job with a fancy title; no job description or aptitude testing required. Neither the current team members, nor the new hire himself, know what his role will be in advance.

They figure out his role as he goes along. He’s thrown into the team, and might immediately be tasked with real work, like redesigning a menu component. This is where the testing really starts.

The newbie must present his prototype dish at a regular tasting once or twice per day, and receives feedback from team members, some of whom will also be testing dishes of their own. He’ll also receive feedback informally, while cooking throughout the day. These “microtests” are are a regular part of the team’s work. Vaughn writes: ‘the low stakes, trivial nature, and sheer volume of microtests involved in routine work… made them an honest signal of a newbie’s ability.’

It may take weeks to perfect a new dish, so the team members get to see their new hire’s approach to making something new in detail. By seeing all this work in progress, they can understand how he likes to approach problems in general, as well as what his particular technical skills are.

After a few months of this kind of work, it becomes clear whether the new hire’s approach to problem solving and his existing skills are well suited to the team. If so, he’ll take a full-time role, continuing to work out how he can be most helpful, just like before.

“Knowing their roles could be modified, people in these teams actively sought out new responsibilities that they wanted to take on and worked to identify new abilities that they wanted to be known for among their colleagues.”

Continuously iterating on your role can be exhausting, but to a certain kind of person, it’s rewarding. One chef interviewed in UM says: “At first it was a bit unnerving to realise that I couldn’t just figure out what my job would be, get good at that, then coast. But once I got over that, having a job that never gets old is what keeps me here.”

Vaughn calls this approach to hiring “negotiated joining”, and its open-ended nature means team members get experience doing a lot of tasks, which makes each member, and the team as a whole, more adaptable.

2. Continually adjusting roles leads to adaptable teams

Microtesting in public creates shared knowledge. Vaughn again: ‘Tests were done in public because multiple members participated in them and saw the results.’ If you show expertise at a skill, like filleting a fish, then everyone knows you’re good at it, and everyone knows everyone else knows you’re good at it. You’ll become the go-to fish-filleting guru.

The next time a dish is prepared which calls for a perfectly filleted fish, everyone will look to you for guidance and action.

‘The same processes that built adaptable teams out of collections of members with open-ended roles also ensured that their members knew each other at a remarkable level of detail: both what their colleagues were good at and how they worked.’

The shared knowledge created by public testing allows natural hierarchies of competence to form. People’s skills are deployed where the team most needs them, not where their job descriptions dictate. And because everyone knows what everyone else is good at, when an unexpected fire starts, it is usually obvious who is best equipped to put it out.

“The unconventional provisional roles team members took on gave them time to figure out their role’s parameters and how they fit with the roles of other members. Consciously deciding to invest this time and effort paid off in the form of a team who could work interdependently like a dream — something like how a perfectly integrated jazz ensemble or basketball team might play together. And this payoff seemed to last over the team’s lifespan.”

3. Choosing open-ended goals permits innovation

Detailed, closed-form goals are a great way to scale an existing process, to go from 1-to-n, but they can’t help you innovate. Innovation is doing something both new and useful. If you already know the end result and all the steps towards it, you aren’t doing anything new. Thus, by definition, you cannot innovate.

Innovation is going from zero-to-one. To do it, it helps to specify the end result — to know what you want to end up with. But trying to specify how to get there is an exercise in futility. You can’t know what will work (and what won’t) in advance.

There are many ways to make something new, but you have to try them to see if they’ll work. Vaughn again: ‘such complexity means that you can get to the desired end result from many different starting points, and by taking many different paths, what complexity theorists call equifinality.’

And here’s the head of ThinkFoodGroup, a conglomerate of award-winning restaurants: “Unless the recipient of the feedback is very inexperienced, I try to let them figure it out on their own, but I usually let them know what the end result should be. That way they can work out how to make things to our style in their own way.”

According to Vaughn, ‘choosing to pursue open-ended goals permits and encourages familiar innovation…’ by allowing team members to use their past experience to make new things, in a style similar to the kinds of things they made before.

It’s also a recipe for long-term success. Think about how crazy a certain class of technophiles get every time a new Apple product comes out. ‘Novelty combined with distinctive familiarity makes for loyal customers, and is nearly impossible to copy.’

4. You learn how to pursue open-ended goals by doing real work

A reliable way to make things that are good is to start making things that are bad, make more things that are less bad, and keep improving until you finally make something good.

You can only learn from feedback on your work, and to get feedback you have to Always be Producing. The organisations studied in UM used every piece of work, no matter how routine, as an opportunity for learning and development.

Vaughn again: ‘Learning happened as an unintentional result of how team members gave and received prototype feedback. Instead of isolating teaching into formal training programs, these teams spread it across almost every piece of work each member did. Microteaching and microlearning happened all the time in the course of daily work. Continuously exposed to learning opportunities, members developed an accurate, detailed, and shared understanding of their team’s open-ended style remarkably quickly.’

Team members wouldn’t just give tactical feedback, on how to, say, improve at a particular technique. They’d also offer strategic feedback on how to think about which kinds of techniques to choose, or how to go about solving a particular type of problem. Team members wouldn’t just get better at isolated skills. They’d also improve their own abilities at generating solutions to problems.

Working in one of these teams is almost like being part of a hive mind: ‘These teams had so many prototype evaluation sessions that tasting and giving feedback in a group setting became practically habitual; their members were continually exposed, almost at an ambient level, to a stream of information about their respective team’s style.’

On the face of it, the organisations studied in UM appear not to do any kind of teaching or learning. But in reality, by giving frequent feedback, learning pervades everything they do. As Vaughn says: ‘It is in fact extraordinarily elegant to use work that would have been done anyway as a way to continually monitor and update how the team is structured and what its members do.’

This principle is adopted by organisations outside the culinary world. I remember Patrick Collison saying somewhere that Stripe interns ship code on their first day. And instead of reading documentation for weeks on end, in my first month at Wren I was entreated to rewrite the entire landing page copy from scratch. There’s no better way to learn how to communicate with new users than to try to do it on your companies most important public asset.

The best way to learn how to do things is to do things, then learn from what you’ve done.

5. Innovation work is motivated by well-chosen sticks better than carrots

The standard approach to setting strategic goals is to pick an audacious overarching goal like ‘become the market leading toothpaste brand’. Then, break this goal into many smaller subordinate goals, dole them out to employees, and motivate these employees with financial compensation, based on the extent to which they have met their assigned subgoals. In many cases, this works:

‘If a team’s goals are clearly understood and specified in advance, and the actions it must take to achieve those goals are also clearly specified and well-understood, the best way to motivate the team is to tie rewards to the output and to train them in the actions known to be needed.’

But if you’re trying to do something new, you don’t quite know what you need to do. Performance based incentives won’t work, because you don’t yet know what actions to encourage. But you definitely need some kind of incentive, otherwise your team members will get bored and leave.

‘Innovation teams need to be forced into doing things they don’t yet know how to do to keep their members happy and engaged and to be effective at innovation.’ The kinds of people who enjoy innovation are ‘oriented to exploration and amenable to uncertainty and failure’, but, like everyone, they’re a little lazy. To innovate, they need a kick up the backside.

This kick in the backside is provided by doing what Vaughn calls “desperation by design”. Essentially, the team takes on a project far beyond their current capacities, and publicly commits to doing it.

Noma, a 3-Michelin star Danish restaurant, committed to running a festival about the frontiers of food, even though they’d never run one before. The Fat Duck, Blumenthal’s flagship 3-Michelin star London restaurant, committed to radically overhauling its menu in only 3 months, when most restaurants would take 9 or more.

‘Successful desperation projects manage to find the point where the project’s demands push the team beyond their current capacities, but not so far that they snap.’ Making a public commitment to an ambitious new project is like throwing yourself into rough, cold seas. Once you hit the water, you’ve got to work out how to swim to shore.

6. Carefully designed progressive overload forces organisations to adapt and innovate

Once you’ve committed to a desperation project, there’s no going back. You can either sink or swim.

ThinkFoodGroup committed to opening three restaurants serving different cuisines in Las Vegas, a new location for them, all at the same time. ‘When simultaneously opening three restaurants, thousands of emergent and potentially interacting problems must be discovered and resolved.’

The fear of public embarrassment stimulates team members into action, resulting in effective problem solving and rapid learning. Every team member’s reputation is on the line. The fact that they all have a personal stake in their team’s success makes them maniacally look for problems to fix.

After committing to opening the restaurants: ‘there was barely even a discussion of whether roles would change, it was taken for granted that they would. Team members were repeatedly thrust against their will into situations that gave them individual opportunities for growth and learning.’

Desperation projects force organisations to learn how to do things that they couldn’t previously do. People working on desperation projects find them enormously stressful, but exhilarating and rewarding.

After completing a desperation project, teams are exhausted. Like an athlete after the olympics, they need to spend some time resting. But where it takes weeks for sportspeople to recuperate, it takes months for the innovation teams studied in UM to cognitively and emotionally recover.

And when they do, to stay innovative, they have to be ready to do it all again.

They select another, more ambitious project, and get ready to hurl themselves into the void once more. It’s a tiring process — much like a weightlifter progressively overloading his muscle fibers with heavier weights. But after a few years of successful desperation projects, the teams expand their capacities beyond what seems possible. Like a strongman lifting objects many times his bodyweight, these teams seem to have superpowers, made possible by seeking productive discomfort over and over again.

It’s all in your head

Many of Vaughn’s recommendations can be found in conventional management theory, but the most important part of his thesis isn’t mentioned anywhere else.

Here’s the secret: It’s all in your head.

To make organisations more adaptable, resilient, and innovative, you first need to view the world in a different way.

‘These counterintuitive methods of organisation, founded on a view of the future as being dominated by uncertainty, were in turn responsible for their success in innovating continuously.’

Adopting the uncertainty mindset — realising the world is unpredictable and the future is unknowable, is a prerequisite to implementing any of the principles above. If you believe your environment can be predicted and controlled, the recommendations seem wasteful or useless. But if you believe the world is uncertain, then the recommendations follow naturally.

To change the world, you must first change yourself. And to change yourself, you must first change your mindset.