Em/mình đang cày series này, đang đọc đi đọc lại để thẩm thấu. Chia sẻ cho nhà mình cùng cày chung :)))
Not-knowings: A thought and action manual
The organisations that we depend on — our governments, central banks, health agencies, multilaterals, major corporations — keep failing precisely when it’s most important that they succeed at navigating uncertainty. This is not a failure of competence or resources. It is a failure of mindset and tools.
The mindset and toolkit these organisations rely on for managing unknowns — risk registers, expected-value calculations, scenario matrices, probability-weighted forecasts — was built for formal risk: quantifiable unknowns in situations where all possible actions, all possible outcomes, and all the probabilities connecting actions to outcomes are knowable in advance.
Truly uncertain situations are inherently unquantifiable: we don’t know all possible actions and outcomes in advance, and we can’t be confident of either the causal connections between actions and outcomes or what outcomes will be worth. In truly uncertain situations, we face four different types of not-knowing that aren’t risky.
The WHO’s response to COVID-19 in early 2020 and the banking sector’s management of risk before 2008 are two prominent examples of what happens when formal risk tools are applied to genuinely uncertain situations: not approximate answers, but false confidence followed by catastrophic failure. Applying the wrong framework produces worse decisions than having no framework at all.
The essays collected below are the first robust map of this territory of not-knowing that I’m aware of. It is conceptually new work developed over three years and sharpened by a series of discussions in 2023 and 2024 with people dealing with not-knowing in their work for governments, multilaterals, corporations, NGOs, and the military. The ideas and tools in this series give you and your organisation the ability to see clearly what types of not-knowing you’re facing, and choose appropriate tools for dealing with them.
Overview essays
A pair of short essays to introduce the concept of not-knowing and synthesise across the essays in the main series.
Introducing not-knowing: The word ‘risk’ describes situations of quantifiable unknowns … but it gets used in ways that stretch that meaning past breaking point. It is currently often used to describe at least four conceptually distinct types of uncertainty: not-knowing about possible actions, possible outcomes, causation between actions and outcomes, and relative value. Treating all four as though they are the same, and as if they are all formally quantifiable risks, produces incoherent decisions, as pandemic-era government responses repeatedly demonstrated. This essay introduces the four-type framework and argues that getting clear on the differences is both a cognitive task (learning new categories) and an emotional one (accepting that much is genuinely unknown). 29 November 2023. Read a discussion summary of this article.
A not-knowing synthesis: Most organisations treat not-knowing as a single thing requiring a single response. It is not. Four structurally distinct types exist — about what actions are possible, what outcomes are imaginable, how actions cause outcomes, and what outcomes are actually worth — each arising from different sources and each demanding a different response. Applying the wrong approach is not merely inefficient; it produces worse outcomes than no framework at all. This essay maps all four types and their sources, and outlines the four-part toolkit that follows from seeing them clearly. 3 August 2025.
Pre-readings
Most people tolerate not-knowing the way bad woodworkers tolerate a knotty plank — grudgingly, working around it. The better approach is to understand not-knowing well enough to use it: as the basis for being generative, as the condition that makes genuine innovation possible, and as a path to being less unnecessarily unhappy. Before that approach is possible, one conceptual confusion has to be cleared up: risk and true uncertainty are not the same thing, and treating them as if they are is one of the most costly and common mistakes in modern decision-making.
True uncertainty ≠ formal risk: Formal risk and true uncertainty are both types of not-knowing — but they are fundamentally different, and treating one as the other is a form of delusion about the world. In formal risk, you know the possible actions, outcomes, and their probabilities. In true uncertainty, at least one of those is missing entirely. The distinction matters because the tools that work for risk are not just useless for uncertainty — they produce false confidence that leads to actively worse decisions. 7 April 2025.
Why (and how) I think about not-knowing: Most leaders treat not-knowing the way novice woodworkers treat irregular timber: remove the imperfections, then grudgingly adapt the plan to what couldn’t be removed. Master furnituremakers like Krenov and Maloof did the opposite — they studied the material’s irregularities first, and let those irregularities shape what the piece became. The essay argues that not-knowing works the same way. Grudging accommodation produces dead, suboptimal work. Actively understanding not-knowing and letting it inform decisions produces something different: work that is alive. The alternative approach is harder, but it is the better approach. 4 March 2023.
Part 1: Motivation
The world is producing more not-knowing faster than at any previous point in history — through interconnection, through the growing power of our actions, and through a widening gap between what we can actually know and what we believe we should know. Understanding it better has three distinct payoffs: it is the only condition that makes genuine innovation possible; it is the basis for being generative rather than merely reactive; and — perhaps most surprisingly — learning to engage with it rather than resist it is one of the most reliable paths to equanimity.
Not-knowing as a path to being happy (Part 1) and Not-knowing as a path to being happy (Part 2): In 1967, psychologists conditioned dogs to accept inescapable electric shocks — then watched them fail to escape shocks that were perfectly escapable, because the conditioning held even when the situation changed. Social and cultural conditioning does the same thing to humans with not-knowing: we are trained to deny it, suppress it, or misname it as quantifiable risk, and that conditioning persists even as the world becomes more genuinely uncertain. Part 1 names this as learned helplessness and identifies it as one of the central obstacles to leading well. Part 2 makes the constructive case: the four states that constitute happiness — curiosity, freedom, effectiveness, and contentment — each depend entirely on willingness to acknowledge not-knowing. The conditioning is the obstacle. Breaking it is the path. 2-3 November 2022. Read a discussion summary of these two articles.
Innovation and not-knowing: The only thing definitionally true of all innovation — whether by a government, NGO, artist, or startup — is that it requires not-knowing: you don’t know what the new thing will be, how to make it, how it will work, how the world will respond, or whether you will value it. Most innovation management ignores this, copying what big businesses do instead. That is why it rarely works — and why the real task is building organisations that can flourish in not-knowing. 28 September 2022.
Generative uncertainty: The standard view of uncertainty as purely a threat to be eliminated is wrong — and the mistake is costly. Uncertainty is the necessary precondition for anything genuinely new to emerge: without it, innovation is definitionally impossible. The essay distinguishes generative uncertainty (open-minded, designed-in, productive) from bad uncertainty and provides three concrete design principles for creating it: clear guardrails, active encouragement of emergence, and flexible support. Critically, it shows why powerful stakeholders will resist all of this and explains how to introduce generative uncertainty incrementally, below the threshold of institutional resistance. 15 June 2023.
Part 2: Clearing the ground
Clear thinking about not-knowing is blocked at two levels before it even starts. The first is affective: an evolved fear response that makes not-knowing feel like danger rather than information. The second is cognitive: a long-standing habit of mislabelling all unknowns as quantifiable risk, reinforced by a growing industry of methods that claim to handle true uncertainty but don’t.
Why not-knowing feels so hard: Not-knowing is hard to face because of a two-layer problem that compounds on itself. First: human bodies evolved to treat situations of not-knowing as physical threats, flooding us with stress hormones that prepare us to fight or flee. Second: social and cultural norms have taught us to label that bodily stress as shameful specifically in the situations where acknowledging not-knowing matters most — leadership, learning, career. The result is denial and paralysis exactly when clear-eyed acknowledgement is most needed. The essay identifies this mind-body nexus as the root cause, and the first step toward reconditioning. 6 March 2023. Read a discussion summary of this article.
How to think more clearly about risk: We use ‘risk’ to describe almost every type of not-knowing — which automatically triggers quantitative tools that only work for formal risk, a rare case where every possible outcome, every action, and every probability is known. It almost never exists outside a casino. The WHO in February 2020 and major banks before 2008 applied these tools to situations that were not formally risky, with catastrophic results. Your risk managers are trained the same way. Thinking clearly begins with better names for different types of not-knowing. 6 October 2022. Read a discussion summary of this article.
The consequences of mindset mismatch: Mindset mismatch doesn’t just produce wrong decisions — it prevents you from seeing the situation clearly in the first place. A mismatched mindset makes you blind before it makes you wrong. Most mindsets operate unconsciously and feel most natural precisely when they most need to be challenged. Some mismatches are merely unhelpful. Others are actively counterproductive, producing worse outcomes than no framework at all. The formal risk mindset is one of those. 12 October 2022.
The insidiousness of the formal risk mindset: The formal risk mindset is dangerous not because it is wrong, but because it feels right. It is so instinctively appealing that even expert decisionmakers adopt it unconsciously in situations where it doesn’t apply. The Bank of England’s management of the UK economy illustrates this precisely: macroeconomic theory is nested inside the formal risk mindset, encoding the belief that a vastly complex system can be abstracted to quantifiable targets and controlled through known levers. When that belief is wrong, actions don’t produce expected results — and often produce actively unwanted ones. Vigilance, not technical skill, is the remedy. 19 October 2022.
Overloading and appropriation and False advertising: The words we use to name a situation determine how we respond to it. Our vocabulary for not-knowing is broken in two directions simultaneously, and both breaks produce the same illusion. ‘Risk’ gets stretched to cover genuinely unquantifiable situations, triggering analytical tools that only work when all outcomes and probabilities are already known. ‘Uncertainty’ gets appropriated to describe quantifiable partial knowledge — most visibly in AI and machine learning, where researchers routinely claim their systems handle ‘uncertainty’ when almost every technically sophisticated usage resolves, on inspection, to formal risk in disguise. Both moves create false confidence that what is genuinely uncontrollable is actually under control. What makes the AI case particularly dangerous: the appropriation comes from reputable fields and researchers, giving the illusion institutional credibility it doesn’t deserve. 9-10 June 2023. Read a discussion summary of these articles.
Part 3: Laying the foundations
Not-knowing about what actions are available is a different problem from not-knowing what outcomes are possible — which is different again from not knowing whether a given action will produce a given outcome, or what any of those outcomes are actually worth. Each type of not-knowing has its own structure, its own traps, and its own logic for what a good response looks like.
Not-knowing about actions and outcomes: We habitually treat actions and outcomes as simpler, more isolated, and better understood than they are. In reality, actions are enchained, partly invisible, and often poorly understood; outcomes are similarly complex and reveal their full consequences only over time (the Green Revolution is one example). The essay introduces a structured vocabulary for the properties of actions and outcomes, then identifies three distinct sources of not-knowing about actions and four about outcomes — each requiring a different strategy. Resolving the right type of not-knowing is what actually opens up new solutions and opportunities. 22 June 2023. Read a discussion summary of this article.
Causal not-knowing: Strategy assumes we know how actions produce outcomes. We almost never do. The essay identifies five distinct types of causal not-knowing — from inaccurately precise quantification to inconsistent causation — each requiring a different response. What makes this counterintuitive: causal not-knowing is not only a problem but a source of competitive advantage. A party that understands causation better than its rivals can conceal that understanding to prevent imitation, or simply act better under shared ignorance. Understanding which type of causal not-knowing you face is itself a strategic capability. 10 July 2023. Read a discussion summary of this article.
Not-knowing about value: Every intentional act rests on assumptions about value that are almost certainly wrong. Value is not stable, not agreed, and not knowable in advance: five distinct sources of not-knowing about value — ranging from differences in subjective opinion to changes in scarcity and social norms — complicate every theory of rational action. But the essay makes a constructive argument: understanding these five sources opens up entirely new ways of thinking about whose assessments of value should count, and creates space for new constituencies and new definitions of what success should mean. 15 August 2023. Read a discussion summary of this article.
The fog of time: Time does not simply add more not-knowing — it changes the type and structure of not-knowing through five distinct mechanisms. Exploration reveals new actions and outcomes; experimentation uncovers new causal pathways; imagination expands the scope of the possible; reallocation shifts what is scarce and therefore valuable; philosophy changes norms about what should be valued. Each mechanism can be actively influenced. The essay reframes futurity not as a source of dread but as a set of levers available for deliberate action — which transforms long-range planning from prediction into strategic intervention. 19 September 2023. Read a discussion summary of this article.
The fountain: The four types of not-knowing do not stay separate: they trigger each other unpredictably over time, like a fountain that keeps running once primed. A change in what actions are possible creates new outcomes, which reshapes causal understanding, which shifts what gets valued. This self-generating quality of not-knowing has a direct strategic implication: tools built on assumptions of stability — fixed causal models, stable goals, settled values — will keep failing. Organisations that are more comfortable with this self-generativity than their competitors can use it deliberately, as Moderna did with RNA therapeutics. 20 October 2023. Read a discussion summary of this article.
Part 4: Building a toolkit
The tools most organisations use for decision-making were designed for situations of formal risk. They are the wrong tools for genuine not-knowing — and using them produces the characteristic failure modes: over-optimisation, brittle strategy, and paralysis when things don’t go to plan. Dealing well with not-knowing requires changes at the level of how you think about intent, causation, and what counts as success.
A mindset for not-knowing: A mindset is a set of assumptions that shapes what you notice, how you interpret it, and what you then do. The risk mindset’s assumptions — that all unknowns are quantifiable — do not merely produce wrong answers: they prevent you from seeing non-risk unknowns at all. The WHO in early 2020 and banks before 2008 both fell into this trap. The fix is simpler than it sounds: use different words for different types of not-knowing. Distinct language forces distinct perception, and distinct perception is the prerequisite for choosing approaches and tools that actually fit the situation. 14 November 2023. Read a discussion summary of this article.
Broad approaches for situations of not-knowing: Standard decision-making tools — cost-benefit analysis, expected value calculations, risk assessments — all share a hidden assumption: that possible actions, outcomes, causal pathways, and value weightings are knowable in advance. A mindset for not-knowing eliminates this assumption, and with it eliminates most of the conventional toolkit. What remains are four distinct broad approaches, each matched to a specific type of not-knowing: portfolios of small experiments for action and outcome uncertainty, stacking the deck for unresolvable causal uncertainty, and negotiation or goal superordination for value uncertainty. None of these involves treating the situation as formally risky. 15 January 2024. Read a discussion summary of this article.
Building a toolkit for not-knowing: A toolkit for not-knowing needs four distinct compartments: diagnostic tools (to identify which type of not-knowing you face), action tools (matched to each type), capacity-building tools (to overcome the discomfort of working in genuine uncertainty), and update tools (to revise diagnoses and actions as situations evolve). Many relevant tools already exist — portfolio theory, scenario planning, hypothesis design, tradeoff elicitation — but are deployed without clarity about what type of not-knowing they are meant to address. That lack of clarity is precisely why they underperform. Knowing what a tool is for determines whether it works. 18 January 2024. Read a discussion summary of this article.
Further reading
Science is simultaneously the most powerful tool humans have ever developed for reducing not-knowing and a persistent source of false confidence that generates more of it. And even people who understand the theory of not-knowing clearly tend to repeat the same practical mistakes — mistakes that are predictable, diagnosable, and avoidable once you know what to look for.
What we keep getting wrong about not-knowing: Our risk-management machinery is not just useless for genuine uncertainty — it is actively harmful, because it creates an appearance of rigour while feeding fabricated inputs into analytical frameworks built for a different type of problem entirely. The four types of not-knowing (about actions, outcomes, causation, and value) each require distinct responses, yet organisations lump them together and run them through the same probability-and-optimisation tools. The real skill is accurate diagnosis of what type of not-knowing you face before reaching for any tool at all. 25 May 2025.
The paradox of Science and uncertainty: Science has given us longer lives, more control over the world, and a belief that control is always possible — and these three things combine to make genuine uncertainty feel more intolerable than ever. The longer future science buys us, the more uncertain future we must face. The causal understanding science reveals creates a myth of general control. And that myth makes the inherent uncertainty of the future feel like a failure of knowledge rather than a permanent feature of reality. 28 April 2024.