Not-knowing: An investigation

I’m working on a new book (maybe) about not-knowing and what we can do about it. My work in progress is posted further down on this page, but the best place to start is with this introduction to not-knowing

In conjunction with that, I’m running an InterIntellect online discussion series about not-knowing every 3rd Thursday, 8pm CET, starting 19 Jan 2023. The sessions will map broadly onto the structure of the book. This discussion series is open to everyone and you should join. I’ll post a summary note after each session. You can see all the session summaries here, and all the essays on not-knowing as they get published.

What people are saying about these conversations about not-knowing:

I found a lot of inspiration and food for thought, I will need a lot of time to digest it.”

This was the highest value 2 hours in recent memory.“

This conversation was really beautiful. I have wanted to have conversations like these for a long time, so it’s quite incredible to be in the company of people asking big questions along these lines.”

Genuinely enjoyed every bit of it, and I look forward to the future sessions.”

The content was off-the-charts engaging … There is so much depth that is both intellectually and emotionally satiating.“

Tickets for the entire series are available here. Tickets for individual sessions are available at the individual session pages linked below, a month in advance of each session. I have a few free tickets for each session; email me if you need one.

Why we really need to talk about not-knowing.

The way I see it, understanding not-knowing is motivated both by the carrot (it’s desirable) and the stick (we have no choice).

The carrot: Anything that is fully knowable and fully known becomes routine and doable by machines, so it is in situations of not-knowing that we become distinctively human. Thinking clearly about not-knowing is a way of understanding being human, which is essential in the context of the current landscape of machine learning and artificial intelligence.

The stick: Every day, either personally or professionally, we confront big and small situations where we must act with partial information. This means that we keep encountering situations of not-knowing in which we must still do something. The most important human work is navigating significant not-knowing. This is why we reward leaders who (in theory) lead their companies and countries through unpredictable situations, why we respect startup founders who work with not-yet-understood technology and markets to create new companies, and why we esteem researchers who work at the edge of the known to create new knowledge.

The last few years of a global pandemic, extreme weather, geopolitical insecurity, and economic disruption (among so many other things!) show that there are actually many different types of situations of not-knowing. And they are growing in number, scope, and impact. Every one of us will have to learn how to adapt to this.

Our experience of the last few years also shows that even those who are charged professionally with navigating these diverse situations of not-knowing often fuck it up. Most recently, the rapid ascent and even faster implosion of Effective Altruism, FTX, and Alameda Research is an object lesson in how poorly not-knowing is understood even by those nominally in the business of managing it — philosophers, venture capitalists, finance industry professionals, and financial journalists. To say nothing of politicians and policymakers.

Learning how to live on and live well in spite of not-knowing is a path to surviving and flourishing in an increasingly uncertain world. The problem is that we’re poorly prepared to even understand not-knowing, let alone know how to respond to it. We urgently need better tools for thinking and action in situations of not-knowing. This is what the book will explore.

One of my underlying assumptions is that the same framework for thinking clearly about not-knowing applies in both personal and professional life. What changes is the context to which that framework is applied, not the framework itself.

My own experience over the last 15 years suggests that some personal and professional questions to which the same framework for clear thinking about not-knowing may be relevant include:

An outline of the discussion series

The discussion series has 14 sessions in 5 parts. My plan for the moment is that each session will map broadly onto one (or maybe two) book chapters.

Part 1 : Why Bother?

Here we set the stage for the rest of the series by considering why the world we inhabit is becoming more uncertain, how that forces us to think more clearly about not-knowing, and why doing so is a path to being happier and more successful.

Session 1 (19/1/2023): Introduction (Session 1 summary). Why the world is becoming increasingly uncertain and unknowable as it becomes ever more complex and more interdependent. How this causes people to make poor decisions, have inaccurate expectations, and generally be less happy and successful than they could be. Thinking clearly about not-knowing is one path to happiness. Why this series is about a general way to think about and act in situations of not-knowing. Why it is not a series only about risk, uncertainty, complexity, chaos, or ambiguity (which all feed into not-knowing). Draft chapter.

Session 2 (16/2/2023): Not-knowing and happiness (Session 2 summary). Why thinking clearly about not-knowing makes us 1) more curious, 2) more free, 3) more effective, and 4) more content — and thus happier and more successful in an increasingly uncertain world. Draft chapter: Part 1 and Part 2.

Part 2: Clearing the Ground

To begin, we explore three obstacles to thinking clearly about not-knowing. First, our fear of the unknown (both from evolution and from social dynamics) which prevents us from looking closely at situations of not-knowing. Second, our long-standing habit of colloquially calling everything risk. Finally, the growing practice among computer scientists and technologists to claim that they have methods for dealing with true uncertainty (spoiler: they usually don’t).

Session 3 (16/3/2023): Fear is the mindkiller (Session 3 summary). Humans viscerally fear the unknown in a way that usually prevents us from dwelling on it properly. We have an evolved, physiological aversion to the unknown that was very useful for survival thousands of years ago but is maladapted to the modern environment. On top of that, social norms and dynamics push us to pretend to be certain even when we’re not, because as social animals we’re deeply afraid of losing face in front of others. Draft chapter.

Session 4 (20/4/2023): Misnaming the beasts (Session 4 summary). We use the same word risk” to describe many different situations of not-knowing, so we always mix them up without meaning to. Things end up being called risk” (even when they aren’t) — usually this is to make them seem more knowable than they really are. This session will clarify the differences between several of the most common situations that are all called risk,” and think through why they are important to distinguish from each other. Draft chapter.

Session 5 (18/5/2023): False advertising (Session 5 summary). At the same time, many approaches advertised as designed to deal with uncertainty actually deal with either very specific types of uncertainty or risk — usually this is to make them seem better at dealing with difficult problems of uncertainty than they actually are. I call this misuse a form of appropriation.” By the end of this session, we might have a very healthy skepticism of technology that claims to deal properly with uncertainty. Draft chapters on overloading and appropriation and on falsely advertising capacity for dealing with true uncertainty.

Part 3: Laying the Foundations

After sweeping away the obstacles, we can start thinking clearly about the different kinds of not-knowing we face. Each of the next four sessions focuses on a different source of not-knowing. We’ll unpack each source’s implications to understand why and how to respond to it differently from the others. Each of the four sessions also asks what it means to take successful and effective action given that particular source of not-knowing.

Session 6 (22/6/2023): Actions and results (Session 6 summary). The most easily understood type is simply not knowing what actions you can take and what outcomes are possible. As technology advances, we have more and more tools at our disposal. This means that we often don’t know the full range of actions that we could take in a given situation. As the world becomes more complex and interconnected, we also are unaware of many possible outcomes — in particular those outcomes that are created by conditions beyond our direct control. Draft chapter.

Session 7 (25/7/2023): Connecting actions and their results (More information and tickets). Increasingly frequently, we also don’t know exactly how much one or more outcomes is connected to a given action. There are in fact at least four distinct sources of not-knowing in causation.

  1. Quantification obsession and legibility obsession means we assign precise probabilities (or probability ranges) to outcomes without strong basis — this has become habitual and taken for granted.
  2. When actions are unknown, we can’t know how likely they are to produce particular outcomes.
  3. When outcomes are unknown, we cannot know when particular actions will produce them.
  4. We are often only partly aware of the causal connections between an action and a range of simultaneously produced outcomes.

Session 8 (17/8/2023): What results are worth. Not knowing how much particular outcomes are worth to you. This is the most under-explored, despite being the basis of expected value theory (which drives so much long-termist decision-making, including much of the Effective Altruism movement). Without valuing outcomes, it’s impossible to compare them and choose actions. Yet how we value outcomes often changes over time, for at least two reasons:

  1. We learn about new outcomes to pursue which we value more, or which change how much we value existing outcomes.
  2. We learn about new actions we can take, which change the cost of achieving particular outcomes.

Session 9 (21/9/2023): The fog of time. The length of the time horizon also exacerbates not-knowing. Present actions affect future outcomes, but the environment in which actions are taken changes over time. The longer the time horizon, the more likely it is that we discover new actions and new outcomes, or that the environment changes so that causation and/or valuation changes.

Part 4: A Not-knowing Toolkit

We’ve become used to decision-making approaches and tools that are ultimately only appropriate for situations of formal risk. These approaches and tools are sharp-edged, legible, explicable, and easily put into books. Tooling up for other kinds of not-knowing is particularly hard because it requires fundamental changes in how we think about why we act (intent), how we think about taking action (causation), and how we understand success (values). The appropriate approaches and tools for other kinds of not-knowing look and feel different: They are more amorphous, less legible, and resist being grasped quickly. They rely a lot on tacit knowledge that only develops with practice and repetition

Session 10 (19/10/2023): Intent, causation, and values. Not-knowing reverberates back and forth between actions, outcomes, causation, and valuation: A change in one of them can affect the others. This means that the approaches and tools of not-knowing must do their best to accommodate how a change in one of the four affects how you think about the rest. This requires a fundamentally different way of thinking about the world and how to take action in it; in other words, a mindset change.

Session 11 (16/11/2023): Mindset change. An appropriate mindset is founded in clearly recognising different non-risk forms of not-knowing. Because of this it privileges thinking and acting in ways which increase degrees of freedom to act (a strategy of tactics) by exploring the interrelationships between actions, outcomes, causation, and values.

Session 12 (21/12/2023): Broad approaches. Broad principles for dealing with not-knowing:

Session 13 (18/1/2024): Tools for thought and action. Concrete tools for use in situations of not-knowing:

Part 5: The Path Forward

Session 14 (15/2/2024): A clear view. In the final session, we briefly review the arc of the salon series and then return to explore in more informed detail how clear thinking about not-knowing can be a vector for happiness.

Work in progress

The book and the discussion series are both works in progress. Each bullet below is a topic that may become a component of the book. I publish a topic writeup draft when it seems baked enough even if not perfect.

I would love comments on pre-publication drafts. You can find all drafts open for public comment here and linked below.

Warning: The ideas here are likely to still be half-baked.

🕳️ Drafts open for comments

🖨 Published

💬 Discussion series summary notes

🤞 Coming soon

🤔 Possible future topics

My first book was about uncertainty and how organizations can design themselves better around it. You can find out more about The Uncertainty Mindset here.