When a news app says there is a 73 percent chance a candidate will win the election, you see just one number. But there are actually two numbers you are processing. The first is the probability they’re stating. The second is how much you trust that probability. These are two separate uncertainties you are processing, and it’s good decision process to deal with both explicitly.

But we tend to merge these into one number in our head. So the number we actually end up using as an input into our decision has often been adjusted unconsciously. Our brain lies to us about probability.

We develop this over time through repeated interactions with sources. We learn naturally that not everything we consume deserves equal weight. The first time a news channel gives you a percentage, you might take it at face value. By the fourth or fifth election cycle, you change the number they give you. You make an adjustment based on how that channel has performed before.

You build a kind of credibility discount system. You multiply the data given to you by a ‘credibility weight’. So the final number in your head is something like source data X source credibility. A source with a strong track record might just have a weight close to 1 (i.e. you take the data at face value). A source you’ve learned to distrust might be 0.5 (effectively halving how believable it is).

In other words, there are always two probabilities being processed:

1. The first, a probability prediction of the event (e.g. there is 80% chance of rain today).

2. The second, the probability of that prediction (i.e. what is the predictions credibility?).

The problem is we often do this second part implicitly when we would be better off making it explicit.

It is interesting to think about how this works when we interact with AI.

First, we have a bias towards believing fluency correlates with reliability. A well-structured argument is generally thought to be believable. Humans commonly mistake how certain something is for how likely it is to be right. But AI will produce confident output regardless of the strength of the underlying answer. This means the signal we would normally use to detect uncertainty is largely absent (e.g. hesitation, qualification, hedging, etc.).

Second, the track record is not tied to a single source (e.g. person or an institution). In fact, it isn’t obvious what it would be tied to. The model provider (e.g. OpenAI / Anthropic)? The specific model (e.g. Opus 4.8)? Even if you can resolve that, the technology itself is essentially multiple sources presenting as one. It is probabilistic by design so it isn’t obvious what the measurement criteria should be. All of this makes it difficult to build a stable track record.

Regardless of what data source you’re interacting with, it is good decision process to make the second number explicit.

Two things to think about when making a credibility estimate explicit:

First, the track record mentioned above. Check the evidence for how accurate this source has been over time. Remember to make sure you are evaluating the track record in the specific domain. When a news site gives you a weather forecast, you should care about how accurate that news site has been for predicting the weather.1

Second, like most decisions, your effort should scale with the cost of being wrong. None of this really matters for small decisions. It is not worth spending the cognitive effort to do this when the stakes are low. When the decision is important, that’s when you should always be making the source credibility explicit.

This week, pick one information source you interact with regularly. The next time you get some information from it, spend some time explicitly thinking about the credibility question: how much do I usually weight this? How much should I weight this?

Unfortunately, there are few things that are true decision-making ‘hacks’. Meaningful improvements to your decision-making takes deliberate effort (hence the encouragement here for people to work on Decidership). But there is one principle that gets quite close.

In my general decision framework, Goals are the context for your decision-making. Goals structure and filter your perception. You notice and pay attention to things connected to what you value and care about. Once an opportunity or problem is noticed, then you can start making a decision.

You can’t include something you haven’t noticed into your decision process. We have all made a decision and then found out afterwards that we had an incomplete picture. We would have chosen differently if we knew more.

Getting the right information in front of you at the right time is often the hardest part of a decision. Once you have it, the decision calculation part becomes simple. So the hard part is often organisational: arranging your life/business so the relevant info is available to you at the right time. Anything that increases the chance of getting that information early is very valuable.

Share your goals and your values with the people around you is one way to do this.1

Think about what happens after a good friend tells you they’re looking for a new job. You start noticing new things. You stop scrolling when you see a relevant job advertisement. You overhear different conversations at the office. The way you interact with the world changes.

You can scale this. Every person you know is moving through the world absorbing information. They are all potential ‘perception nodes’ for you. But without knowing what you are trying to achieve they can never pass the information back to you.

I think about it as Distributed Perception: the practice of distributing your perception by turning those around you into perception nodes.

In personal life, this is just being open with those close to you about what you’re working towards. Like the job example above.

In business, this is a strong argument for sharing your corporate values. It’s easy to become cynical about internal messaging. For employees, it can seem like corporate speak, disconnected from your day-to-day job. For management, it’s easy to forget the point of company values. They can become something for optics rather than for behaviour change.

But they can encourage behaviour change. People who understand what the company wants to achieve will notice new things and act accordingly. Sales can structure deals according to your revenue ambitions. Marketing will notice relevant risks in your advertising. Teams will hire people that match the cultural values. Let people understand the destination and they will find routes without being told.

Distributed Perception requires a little structure. Here are a few principles that help it work:

• Two-way: Someone else helping you with your Goals is a privilege. So, the process should also be reciprocal. You have to be willing to make it mutual. Ask the people around you what they’re working towards.

• Repeatable: You shouldn’t expect your needs to stay at the top of anyone else’s mind. You should find a way to create a repeatable process where goals and values are shared.

• Low friction: Distributed Perception only works if the information reaches who needs it. If people spot something relevant but don’t know who to tell then the information will get lost.

• Actioned: Distributed Perception only works if people trust that their contribution matters. If they raise things and nothing happens, then they will stop. What is seen has to be acted on (or at least acknowledged).

So this week, tell someone your goals. Ask them what they value. The more people around you who know what to look for, the more likely they are to help you. And the more you know about what they need, the more useful you become to them.

Everyone’s lives get a little bit easier to manage.

Some beliefs are so embedded in our thinking that they become the assumed conditions of doing things. A bit like how we never think about gravity. We just operate within its constraints without a second thought.

This week I want to talk about one of these beliefs that underpins almost all our decisions. When two things you want are in tension, one of them has to give. More of A means less of B. There is always a trade-off.

We frame all decisions as a negotiation between gains and losses. “Which part of what I want am I willing to sacrifice?”. But this should be a conclusion you’ve reasoned to rather than an assumed starting point. This assumption prevents you from properly Deliberating about a problem because you've narrowed your option set before you've done the work of exploring it with an open mind.

The assumption of trade-offs is consistently reinforced from different domains. It’s been foundational to much of economic theory. We see it in nature by looking at the different ways animals have evolved by trading off features to survive. And we are constantly reminded of it at a cultural/societal level when “there’s no such thing as a free lunch” is stated with the authority of a natural law. So I think it is entirely understandable how even excellent decision-makers accept the framing that you can’t have it both ways.

However, this is a case where the perceived wisdom is not entirely true. Trade-offs are not a fundamental law of decision-making. Sometimes, trade-offs are avoidable. Sometimes, you can have it both ways.

Australian economist wrote an article exploring how economics is often framed as the study of trade-offs. It is so central that it’s rarely questioned. This causes better solutions to stay hidden. Gruen references Toyota disproving the assumed trade-off between cost and quality. Toyota designed production systems that got things right the first time. Traditional manufacturers focused on fixing errors after the fact, which was both more expensive and less effective. Toyota improved quality and cut costs simultaneously.

IKEA might be another example. A long-standing industry assumption that cost and aesthetics were trading off against each other. Cheap furniture must be ugly. Perhaps IKEA showed this was less of a trade-off and more of an independent design problem that could be solved.

So a lot of the work decisions we face today are removed from natural conditions. They’re not bound by the same zero-sum logic. The modern corporate world, in particular, is not “real” in the same way the natural world is. The business decision space is more “constructed” rather than “determined”. It is a positive-sum game. The constraints of many business decisions can be changed or circumvented in ways that natural constraints cannot.

Even the binary framing (trade-offs being between two things) is a simplification. In practice, trade-offs rarely present as a simple sliding scale between two things. More often, you’re dealing with a web of variables all pulling in different directions at once. Time, cost, quality, convenience, etc.

So this week I encourage you to be skeptical of every trade-off you encounter. Question whether it is real. Spend a moment to think about whether there is a way to solve the problem with less sacrifice.

One technique you might want to try is what’s sometimes called “constraint relaxation”. This is asking what you would do if the constraint simply didn’t exist. What would the ideal solution look like if you couldn’t fail? Sometimes a better solution can be found in an idealised scenario. Or at least the thought exercise might point you in the right direction.

Most of the time, you’ll indeed confirm that the trade-off is real, and proceed accordingly. No harm done by checking. But sometimes, you might just get a free lunch. The only way to find out is to stop assuming you can’t.

There is a Chinese idiom “鸡毛蒜皮”. It translates approximately to ‘trivial things like chicken feathers and garlic skins’. The origin is from a story about two neighbours. One raises chickens; the other sells garlic. When the wind shifts, chicken feathers drift into the garlic seller’s garden. When the wind shifts back, garlic skins blow the other way. A dispute breaks out. It escalated until it landed in front of a judge, who dismissed the case as too petty to bother with.

The judge did what neither neighbour could: recognised that some problems aren’t worth solving.

Many of us are more like those neighbours than we’d like to admit. It can be embarrassing to admit just how much of our mental energy goes to exactly this kind of thing. The trivial, low-stakes, mildly irritating decisions that pile up like garlic skins on the wrong side of the fence.

To admit this fully, we also have to acknowledge the fantasy version of our lives: if we could just clear away all the petty small stuff, we’d have unlimited bandwidth for the decisions that actually matter. We would be full of vitality ready to make progress on what we want.

It’s a lovely fantasy. It’s also wrong in at least three ways:

First, the trivial decisions never go away. They’re infinite. You solve one petty thing and two more appear. That’s just nature.

Second, sometimes the small stuff is the real stuff. Small things left unsolved can often compound into bigger issues. Or sometimes the small stuff is necessarily linked to the bigger picture. If you can’t sell your garlic because it’s covered in chicken feathers, then it might not be so trivial.

Third, those big decisions you’re longing to get to? You’ll complain about them too. They only look appealing at a distance because the small decisions are so immediately annoying. The grass isn’t always greener on the other side of your to-do list.

So, how should we deal with small decisions?

I came across an idea from called personal policies.1 It is a pre-determined list of things he does or doesn’t do. Adam gets asked to do lots of things: blurbs, advice, introductions, collaborations. Individually, these requests are reasonable. Collectively, they could consume his entire life. Personal policies allow him to have a rule that applies when he receives these one-off requests. For example, has he been given two months’ notice? If not, then the answer is no.

I find the logic behind this useful. By converting a small decision into a personal policy, he gets to stop spending energy on it. I stole the skeleton of that idea and applied it more broadly.

You should start creating Tiny Rules.

A Tiny Rule is a pre-made decision. A small, practical instruction to your future self: when this type of thing comes up, do this.

Here are what some might look like:

• Reject caffeine after 5pm.

• Always take the stairs if it’s three floors or fewer.

• Never accept a meeting without an agenda.

• If a task takes less than two minutes, do it now.

• Always sleep on a purchase above a certain threshold.

Share

When I share this idea I often get pushback: isn’t it bad process to deliberately build a system you know will produce suboptimal decisions? I struggled with a response to this for quite some time. It is a fair and intuitive point. I have two main arguments for the benefit of Tiny Rules:

First, your decision-making life is a portfolio. You make thousands of choices competing for the same finite pool of attention, time, and energy. It is wrong to evaluate each decision solely as a self-contained event. Accepting an individual suboptimal outcome to get more bandwidth elsewhere is a positive expected value move. Tiny Rules produce better outcomes cross your whole decision-making life.2

Second, everyone has to start somewhere. For certain people, Tiny Rules are a good entry point for improving decision-making. Get Tiny Rules in place first. Once you have more structure to your decision-making you can revisit Tiny Rules and make them more optimal. I have had success in getting people started on a behaviour change journey through using Tiny Rules.

You don’t need many. A handful of well-considered Tiny Rules, applied consistently, can do a surprising amount of work. Here is a rough framework for identifying when you might want to use a Tiny Rule:

• Is this a recurring decision? Creating a Tiny Rule makes sense for things that come up repeatedly.

• Is there limited downside? Tiny Rules should apply to low risk situations.

• Does this deserve my time? Tiny Rules gain the most value when applied to decisions that should be made quickly.

• Is the area well understood? Tiny Rules work best where you have enough experience to know what a good default looks like.

• Does it strongly relate to emotion? It might make sense to let Tiny Rules take over if emotion is likely to make you choose poorly (e.g. you’re tired, hungry, or angry).

If a decision passes most of those tests, you’re probably safe to make it a tiny rule. I would suggest writing it down because there will be a temptation to re-examine each decision when it comes up. Resist that!

You can start spending more time on what is important to you by using Tiny Rules.

was one of the world’s top poker players for two decades, earning over $4 million in tournament winnings. Annie is the author of Thinking in Bets, How to Decide, Quit, and the co-founder of the Alliance for Decision Education. She also runs an excellent decision-making course available on Maven (I know because i’ve taken it!).

I had the privilege of talking to Annie about what it actually means to get better at making decisions.

You can listen to the whole conversation above or read the transcript below.

Here’s what you’ll learn if you read or listen:

• Why no decision is truly a “one-off”.

• How to practice decision-making (fantasy sports, poker, and why the feedback loop matters more than the activity).

• Can you think probabilistically even in situations of radical uncertainty?

• Why decision journaling usually fails.

• The one thing Annie recommends people do to start taking responsibility for the quality of their decisions.

Subscribe now

I recommend you listen to the conversation, but here is a transcript, with minor edits for clarity, brevity, and flow.

The Only Two Things That Determine How Your Life Turns Out

Stephen: Annie, welcome.

Annie: Thanks for having me. Excited to be here.

Stephen: Good to see you again. I think a good place to start might actually be how I discovered your work and why it resonated with me. I originally did my bachelor’s and my master’s in judgment and decision making, but in the moral realm — all about moral decision making. It had a big focus on Jonathan Haidt’s social intuitionist model. I was basically trying to answer the question of, if there was such a thing as moral expertise, what might that look like? Then, fast forward to 2019, when I was searching about different kinds of thinking styles, I came across Thinking in Bets. What I really took away from it was that it connected the dots for me on decision making being its own learnable skill. I’m wondering if you could talk to me a little bit about improving decision skills and why you’re so confident that’s actually something we can do.

Annie: So let me just start with this. There are only two things that determine how your life turns out: luck, which you have no control over, and the decisions that you make. What luck sets into place is the distribution of the outcomes that are available to you at the beginning of your life. If I had been born in 1600, that’s a very different set of outcomes than being born when I was, particularly as a woman. Being born in America versus a third world country. Who are your parents? How much do they care about education? Are you tall? That’s going to determine your probability of being in the NBA. You don’t have any control over that stuff. So that leaves the decisions that you make. The decisions you make are actually affecting the probability that you observe certain outcomes in your life. If you’re good at decision making, it’s going to increase the probability that you end up with better outcomes than you otherwise would. It’s not going to guarantee it, but it’s going to push you toward the right tail. In terms of whether this is a skill that can be taught — the history of the science has been to have this very big focus on the ways that our decision making goes wrong. For people who’ve read Thinking, Fast and Slow, that’s really cataloguing a lot of the types of errors in thinking we can succumb to. Cognitive bias, heuristics, sunk cost fallacy, overconfidence, confirmation bias. Those ideas were actually quite controversial in the seventies. Starting with Kahneman and Tversky, Richard Thaler, and others, they really showed that these errors are real. That’s where the focus was for a long time. What’s shifted more recently is people asking: okay, but what do you do about it? You can’t just know about it and think it’s going to go away. Knowing you’re subject to confirmation bias alone is not enough. People like Phil Tetlock and Barb Mellers, my own work, Katie Milkman, have been thinking about how we actually create processes and structures that reduce the impact of these errors. And it turns out there are a lot of scientifically backed interventions that can improve decision making. That’s a lot of the more recent work from the last decade, decade and a half. I just really believe in getting that message out and giving people the tools to become better decision makers.

From Pessimism to Practice: How Decision Science Changed

Stephen: I think that’s also one of my focuses here — stemming from what I studied, I think what we do and how to do it is essentially the fundamental human problem. That’s why I push it as far as saying we have a responsibility to ourselves to work on it, because it’s the only thing we have control over that can determine everything else. I like that you’ve picked up on this swing that’s happened in the field, from what I see as quite a pessimistic outlook — it’s all about what goes wrong, we’re fallible, and a lot of that material makes people feel quite nihilistic. But now we’ve swung more toward forward-looking things we can actually do to build structures around making better decisions.

Annie: Yeah. What was happening, particularly in economics for a long time, was this idea that we act in our own best self-interest — rational actor theory. When Kahneman, Tversky, and then Thaler came along, they were very poo-pooed. So much of the focus was on how systematic and predictable these errors are, rather than how to fix them, because there was so much convincing that needed to be done. I think your point is well taken. I would add something: Kahneman is famously pessimistic — famously so. When you’re diving into how pervasive and durable these errors are, you can become very nihilistic. But for lay people interacting with this material, I actually find the opposite — they think that because they know about it, they’re fine and won’t do it. They can list off sunk cost fallacy. If I buy a stock at 40 and it’s trading at 30, I should be neutral to whether I’m in gains or losses — the fundamentals and forward-looking ROI are what matter. But people say “yeah, but I know about that, so I don’t do it.” The scientists were saying these things are so durable and hardwired that we’re just going to make errors. The general public hearing about the material had the opposite response: now I know, that’s a competitive advantage, I’m not going to do it anymore. That’s where the convergence comes in. The research, which is very pessimistic, has become more optimistic. And people who were way too optimistic need to become more pessimistic — to realise knowing about it is not enough, and they need to be using these structures in their own lives.

How to Get Your Reps In

Stephen: And one thing I find quite interesting when that reality hits people — and we talk about this as a skill — is what does practice look like? I’m curious to talk about how you get your reps in. When I did your course last year, we spoke about how people were trying to integrate this into their daily lives. For me, I chose fantasy sports about seven or eight years ago, solely as a good way to get reps in. It’s got enough hidden information, a good blend of luck and skill. That’s been incredibly helpful for disciplining a whole range of habits. I’m wondering if you could talk a little bit about that. What I’m particularly interested in is how do we do it well? With decision journaling, it didn’t work for me because I needed something tangible where I wasn’t so tempted to lie to myself. It’s easy to write it down and say “yeah, I would take that bet” — but when there’s nothing on the line, no skin in the game, I really struggled with that. Fantasy sports worked because you’ve got to make a decision, the deadline comes, it’s out there, it’s competitive. What are your thoughts on good ways to get your reps in?

Annie: I’m not a big fan of decision journaling. I think it’s kind of mushy. Some of the things I teach might look like journaling, but it’s actually not — it’s creating structure around your decision making and making sure you’re quantifying your opinions pretty precisely, in a way that allows you to close your feedback loops. When people talk about journaling, they’ll say “I did this because blah blah blah” — I don’t really know what that means, and I particularly don’t know because, as you know from fantasy sports, what does one outcome mean? You go back to your journal and you’re going to say you got unlucky if it’s bad and exactly right if it’s good. I want a rubric. I want forecasts. I want things on a Likert scale — be precise — and then do that over lots of decisions. Things you’ll see in a decision journal: “I think there’s a really good chance that...” — what does that mean? Those are words describing some probability and you haven’t said what you think the probability is. Obviously I came from poker, so I was repping this all the time in that environment. To get your reps in its fullest form, you need an environment with hidden information and a lot of short-term influence of luck. In fantasy sports or poker, there isn’t a lot of long-term influence of luck, but in the short run there is. If I flip a coin, twenty-five percent of the time I’ll get two heads in a row. That doesn’t mean heads is a hundred percent. Twelve and a half percent of the time I’ll get three heads in a row. Six and a quarter percent of the time, four heads in a row — that’s actually quite often. But if I flip it ten thousand times, I’ve taken the luck element out and I’ll see it’s fifty-fifty. Fantasy sports, poker, options trading — similar in that sense. To become a good decision maker, you need both hidden information and a strong influence of short-term luck. Chess will help you think in decision trees — if-thens. If I move here, here are the possible moves they could make, and I can think about the probability of each. That’s valuable. But not a super strong influence of luck in the short run, and not much hidden information. If you really want it: play cards, do fantasy sports. But also — and this is the interesting thing — most things we decide about in life are actually very similar to that. For important decisions, start structuring your decisions in a way that allows you to close the feedback loop and understand what happened. That takes more than a journal.

Share

Feeling Twenty-Five Percent in Your Bones

Stephen: The way I always think about it — and what I’ve found from using it as a decision-making exercise — is there really isn’t a substitute for having that kind of experience, particularly for understanding short-term versus long-term. When I first started playing seven or eight years ago, I was so caught up in the near-term information every single week. The best way I can describe how I play now is: in the first week of the season, I’m simultaneously in the last week. There’s almost no difference to me now, because I’ve had that experience of knowing what information reveals itself over the short term and the long term.

Annie: The way I describe it with poker — when people ask me what poker really taught me — is: I can feel twenty-five percent in my bones. If I’m a three-to-one favourite, seventy-five percent of the time I’ll have the good outcome and twenty-five percent of the time the bad one. That’s actually quite noisy. It means twenty-five percent of the time I will observe a bad outcome, but I don’t know if I will observe it on this occasion. What happens is when people observe that twenty-five percent — there’s a bad card, they lose the pot — they just think they got so unlucky. And when the seventy-five percent occurs, they think that was just supposed to happen. No — it was supposed to happen seventy-five percent of the time. When you play that game as long as I played it, and you get into all these situations, you start to really feel the probability differently. The world is probabilistic, but everything settles to one or zero. On a poker hand, I can only win or lose. You can watch an NFL game and track a team’s win probability toggling around — 35%, then they score and jump to 52% — but in the end, they either win or don’t. If I get in my car and I’m 0.5% likely to have an accident, on that drive I either have one or I don’t. The clearest example in everyday life is a weather forecast saying there’s a 75% chance of rain. It either rains or it doesn’t. And we all know that when it rains we say the forecast was right, and when it doesn’t rain we say it was wrong. No — on that particular day it was going to settle to one or zero. This is a really big problem for how we interpret the world and our decisions. Poker solved that for me. Repping that over and over got me out of the trap of putting too much weight on whether it happens or doesn’t on a given try.

Stephen: Yes. And it’s getting those outcome reps in high volume, which is the difficulty when we think we can reliably practise decision-making skills just as we go about our lives generally. You only get that feedback on moving house maybe a couple of times in your life. The feedback loops aren’t there.

The One-Off Fallacy

Annie: Yeah. Although it’s interesting — people think, well, I’m only going to get married once, or hopefully once. Or how many times am I actually going to move house? Does this really apply? How could I know what the probability is? How could I close that feedback loop? And then I talk to people in early-stage venture capital and they’ll say, I’m not even going to find out how it turns out until ten or fifteen years from now. The feedback loop is just too long. And they’ll cite power law — you might have a portfolio with forty companies, three of them are going to be the fund drivers. How could you tell whether the other ones were poor decisions? I don’t accept any of that. On moving house: other people have moved house before. Go look at them. You’re not the only person who has ever moved house, and you’re not that different from other people. You can always find other reps that will inform the decision you make. Go find data. If you’re going to buy a house and live there forever — go look at the average length of time someone owns a house. That informs all sorts of decisions you make around it. Know that your income is going to increase and you’re willing to pay more than you probably should on your mortgage — go look up the data on how that works out for people. There’s all sorts of things you can do, even for a one-off, if you go look at the base rates. Second: there is no decision that’s really isolated in the arc of your life. You make thousands of decisions every day, and in terms of really consequential decisions, you make a lot of those — what college, who to marry, where to live, what house to buy. You should think about it as a portfolio of decisions. Going to venture capital: the feedback loop isn’t ten to fifteen years. What are you even talking about? You invest and then all sorts of things happen — does it raise another round, is revenue growing, are they making key hires, what does churn look like? Why are you accepting that the feedback loop is that long? That’s true for a house too. Do you like the neighbourhood? Were there unexpected repairs? Is it financially taxing? You know all of those things in between. Treat it that way. And in terms of luck in the short run — that’s why you have portfolio theory, to smooth it out. Treat your decisions the same way. On average, is the way I think about these things predicting better outcomes or worse outcomes? Good decision making is work. And people throw their hands up and say, well, it’s a one-off, so decision frameworks don’t apply. Of course they do. Or they say, it’s a one-off, so I’ll just go with my gut. Please don’t do that. Your gut might be really good, it might be really bad — we’ll never know because you left it in your gut.

Radical Uncertainty

Stephen: And I think one other thing people struggle with around underlying probabilities — and just thinking probabilistically — is this tendency to feel that some decisions fall into a bucket of “radical uncertainty,” where the probabilities can’t be determined in any meaningful sense, and that by thinking probabilistically you’re generating a fantasy rather than anything useful. What are your thoughts on that?

Annie: There is no decision you have ever made that doesn’t involve a forecast. Period. A forecast is some educated guess about what the set of outcomes might be given any choice you’re making, and what the probability associated with those outcomes is — even in cases of radical uncertainty. The reason I know that is: even when you’re in radical uncertainty, if you choose A over B, what’s implied is that you think A has a higher ROI than B. And I don’t mean just money — it could be happiness, fulfilment. This is where we get into expected value, where value means what you’re trying to get out of whatever you’re choosing.

Stephen: There’s always a caricature made of probabilistic thinking.

Annie: Yes, that it’s somehow heartless and doesn’t take love into the picture.

Stephen: Yes, and people zone in on that word “value.” Even if you want to make it as emotionally laden as you like — I always think of Emily Falk’s question, which I steal all the time: would you rather cuddle a puppy for ten minutes or have ten dollars? What I take from it is that we put things on a common value scale already. And when you ask it, people are so quick and sure in having an answer. What’s more interesting than the answer is the fact that they can answer it at all.

Annie: Because we translate these things across all the time. This idea that the uncertainty can be so radical that you shouldn’t think probabilistically — but you are thinking probabilistically. That’s the point. No matter how uncertain the world is, you’re making choices. One choice people make is to not change anything. Implied in that is that the path you’re already on is going to get you to the best place. There’s all sorts of biases that push you toward sticking with your current path independent of whether it’s better for you, but let’s ignore that for a second. Even if you’ve got a choice between two foods you’ve never tried — I guess we’re in radical uncertainty — except that you’re thinking, maybe this tastes like chicken. You’re making educated guesses any time you choose one option over another. Is it going to be right in the sense that two plus two equals four? Of course not. That’s the whole point of uncertainty. The less uncertainty there is, the more you get into two plus two equals four territory. Like if I’m an inch from another car, my forecast of the probability that I’ll hit it if I put my foot on the gas is going to be close to certain. But you’re still making a forecast, because the last time I checked you’re still acting, still deciding, still choosing. And implied in that choice is that you have made a forecast. So I’m saying: make it explicit. Not in the sense that you’ve found the right model or that you’ve fallen for false precision — I don’t want that at all. But make the forecast explicit and make clear why you’re forecasting that. Why are you choosing this? So it can be examined like an object, and other people can examine it and give their input. And it is most important when you’re in radical uncertainty. Because when you accept that there is no forecast to be made and say instead, no — I’m going to make one anyway, so I might as well make it explicit — that’s actually a huge competitive advantage. Everybody’s pretty okay when you’re an inch from a bumper. The more uncertainty there is, the bigger your advantage from getting disciplined about not accepting “I’m just going to wing it.” Part of it is accepting that you don’t know, and so you’re writing down your best guess with your rationale and then getting other people to help you with it.

Stephen: And I think that winging it point is also important — even if someone is feeling unconvinced by all of this in cases of radical uncertainty, what’s the alternative? What’s the better alternative? That’s another way of thinking about it.

Annie: Right. People are so uncertain about the future of AI — what impact will it have, what jobs will it take away, what will happen to electricity bills? It’s a new technology and we don’t really know what the future looks like. Not just in terms of AI’s capabilities or how businesses will respond with their workforce, but also whether governments will regulate it, and how. It’s all reflected in the valuations — there’s a huge fat right tail of returns, but also a lot of these things end up being duds, and we don’t know where we are yet. But if you are going to college and choosing a major right now, among all this uncertainty around AI, I assume you’re choosing the major that will give you the most value, whatever that might be. If you’re thinking about employability, you’re making a guess at what you think the future of AI is — as uncertain as that might be. Make those guesses explicit. Go talk to other people who might have different points of view. Don’t tell them what you think first. Compare those points of view. Explore that. And don’t just throw your hands up and say, well, we don’t know the answer. You don’t know the answer about anything. And I know that because fifteen years ago, everybody became a coder because that was a sure thing.

Closing Thoughts

Stephen: Thank you so much. One final question for you. If someone wanted to take more responsibility for the quality of their decisions and wanted to start doing that, what’s the one small thing you would recommend?

Annie: We’ve talked about a bunch of them, so listen to what we’ve discussed — that will help. But the biggest thing, particularly for anyone making group decisions — and again, everything is a group decision — is: stop telling people what you think when you’re trying to find out what they think. That’s the biggest thing you can do. Within the clients I consult for, we put them into an asynchronous independent model of eliciting feedback. We’re not all yelling in a room what we think. We’re writing it all down first, where I don’t know what Stephen thinks when I’m writing down what I think. But you can do this in really simple ways. If I’m thinking about what major to choose, I could have a set of structured questions I send to Stephen and have him answer them. Or in a conversation, I can say: so what do you think? What do you think the impact of AI is? But notice I didn’t tell you what I think. Surely I have an opinion. But I need to keep it away from you if I really want good information from you. What people do is not that. They’ll say, okay, I’ve been thinking about this... and it’s very hard socially. And the first response they get is often: well, what do you think? Sometimes it’s easier to just send a survey — takes some of the social awkwardness away. But I put everybody into that model because so much of the way we’re biased has to do with contamination from other people’s opinions. And I might suffer from confirmation bias, and you might too — but the things we’re trying to confirm are different. So that’s really helpful. But if I tell you my belief, I’ve now infected you with it. That’s actually going to cause a lot of problems in finding out that you have a very different viewpoint. Everybody should maximise the amount of disagreement they see. And you can’t do that if you’re talking at the same time.

Stephen: I would encourage everyone to do the same. You’ll catch yourself doing it all the time once you start thinking about it more explicitly. As I said, it’s really hard.

Annie: All the time. It’s the natural thing to do. “I love that movie. What did you think of it?” That’s a low-stakes decision, so I don’t really care if you make the error there. But that’s how we talk about everything.

Stephen: You have a brilliant course on Maven. And I believe you have a new book coming out towards the end of this year. Anything else you would like to plug?

Annie: Yeah, the new cohort has opened on Maven — that would be cohort eleven. People can go to Maven and sign up. I think it’s a fun class. I like teaching it. The new book — I’ll be done turning it into the publisher this summer, so I’m guessing Q1 2027. The book is about how to become a better consumer of information — to take responsibility for the way you are interpreting the data you’re coming across. There’s a lot of focus right now on misinformation, but misinformation isn’t actually very prevalent online. Most of what people tell you is true. The problem is that it’s an incomplete picture — not necessarily because they’re malign actors, but because that’s the information they have. So I’m trying to give people the tools and the desire to take responsibility for the way they’re consuming information, and the questions they’re asking of that information before drawing conclusions that will inform their decisions. A simple example: go scroll Instagram, and most of the information you see is before-and-after data. I took a supplement, my cold went away in a week. But what about people who didn’t take the supplement? In the business world: sales were here, I did a marketing campaign, then sales went up — and I’ll show you someone who doesn’t think the campaign caused the increase. It may have. But that data does not tell you that. What if I said sales were here, Ted Lasso premiered, and then sales went up? Nobody falls for it then — but the data tells you just as much in both cases. One is just narratively satisfying. So I’m trying to give people simple tools to stop falling for these statistical illusions. I’m also not a pessimist. Most of the errors are just bonus videos. Most of my course is actually: here are practical things you can do to make your decision making better. The Alliance for Decision Education would also be great for people to check out — it’s a nonprofit I co-founded with my husband. We’re trying to bring decision-making skills into K through 12. Why did you have to wait until adulthood to start learning these skills? We never tell people what to think — the focus is never on that. Two people can look at the exact same information, have great decision processes, and reach a totally different conclusion, because what you value really matters. We would never tell someone what they’re supposed to value or decide. We’re just trying to help them understand how you might go about it.

Stephen: Annie, it’s been a pleasure.

Annie: It’s really nice seeing you again. This has been a super fun conversation. Thank you.

If you enjoyed this course you may wish to check out Annie’s excellent decision-making course available on Maven.

If you want to think more carefully about how you make decisions, this newsletter is free. Every Monday morning, something short to help you start the week deciding a little better.

Subscribe now

If you are interested in understanding about the decision process more generally, I recommend reading my decision framework.

Read The Framework

Share this post with a friend or colleague who wants to make better decisions.

Share

AI models are growing more capable and more decisions are being outsourced to them. Efficiency and ease of use make it appealing. Why spend thirty minutes thinking when you can get an answer in thirty seconds?

AI is great at processing large amounts of information quickly. It can pull together research, structure data, and model outcomes. These are useful tools to have available as inputs to human decision-making.

But there is one thing it cannot (and should not) do.

It can’t tell you what you want. It can’t tell you why something matters to you. It lacks the ability to understand human motives, deep inner context, and the nuances of human needs. It doesn’t know whether the output actually leads somewhere you care about going.

The typical AI-assisted decision looks something like this. Someone has a problem. They describe it to an AI. The AI gives them options, analysis, and a recommendation. They act on it.

But decisions are not calculations. A decision is an attempt to move you from where you are to where you want to be. The quality of any decision (good or bad) can only be measured against your goals. And your goals are yours.

Getting clear about your goals and values involves comparing things AI can’t. Stability against ambition, relationships against career, your appetite for risk, what feels right.

It means bringing your entire self to a decision.

Good decisions require the full integration of thinking and feeling. The thinky stuff (your knowledge, your logic, your rationality) and the feely stuff (what you value, what you care about, what you’re afraid of).

AI can be a great input for lots of the thinky stuff.

AI can’t help with the feely stuff.

What’s dangerous is even though it can’t do the feely stuff, it will still give you an answer. A confident, well-structured, plausible-sounding answer. It just won’t be an answer that can substitute for your own answer. Even if you give the AI enough context, it is not you.

Only you can say if that output is actually useful for moving you towards your goals. You are the one who has to make the decision.

People who use AI well understand this. They don’t ask AI what to decide. They bring it in to improve thinking they’ve already done. To help them with how to decide. The AI comes in after you’ve already done some human work.

The human work is understanding your goals and values clearly enough that you can determine what will make your life better. This is the one skill AI can’t replace. I think the best way to build this is through reflection.

Every decision you make without clarity about your goals is essentially a guess. You might get lucky and end up where you want. But you have no reliable way to know whether the choice you are making is actually moving you toward your goals.

Deliberate reflection can help. The more you reflect on your goals and your choices, the better you become at thinking critically about what you want and estimating the likelihood something will get you there. The more you do this, the more you increase the chances that your next decision will get you closer to where you want.

AI will be significantly more useful to you if it is working in that context.

Here is a 3-step process for how I suggest using AI to help with a decision:

Step 1: Understand What You’re Actually After

The process needs to begin with and maintain human agency throughout. You have the power to choose what you want. Without clarity on your goals, AI output is more likely to reinforce your biases that pull you towards the wrong things.

It helps to separate your goals into two types.

Ultimate Goals are the things valuable in themselves and are the point of everything else. You don’t want them because they get you somewhere. They are the somewhere you want to be.

Enabling Goals are the things that move you toward other goals (all the way up to your Ultimate Goals).

AI cannot decide what you should value. The decision you are contemplating is essentially an attempt to move toward goals that reflect your values (and that remains uniquely human!).

So before you bring AI anywhere near a significant decision, reflect on the question: what am I actually trying to achieve here, and why does it matter to me?

Step 2: Separate Enabling Goals from Ultimate Goals

Once you have outlined your goals, look carefully at which category they fall into. Then try to get specific about how they relate to each other.

Most people go to AI with something like: “I have a performance review, how should I approach it to get a promotion?” That is a reasonable ask but it misses context that would improve the output for you.

A promotion is an Enabling Goal. But what exactly is it enabling? A greater sense of financial stability? The feeling of being seen? The ability to do more meaningful work? More control over your schedule? Each of these points towards a different Ultimate Goal and each would produce a different answer if the AI had that context.

The practical exercise of this step will look something like this: take the decision and write down your goal in one sentence. Then ask why that goal matters. Keep doing that until you reach something that doesn’t need further justification. That is your Ultimate Goal. Everything smaller that helps you get there is Enabling. Now you know what you are actually trying to achieve, and you know what context to give AI.

When you know which Enabling Goals serve which Ultimate Goals, two things happen:

• First, you can check whether you are actually chasing the right thing. You should be open to the possibility that the choice you are considering doesn’t move you toward the life you want.

• Second, you can give the AI the context it needs to be useful. The better the AI understands your Ultimate Goals the better it can work with you to actually get you there.

For example, long-term job security may be your ultimate focus because you want to support your family. This clarification will change the output you get about your performance review. The actionable advice will move towards achieving that (e.g. using the review to demonstrate you’re worth investing in long-term rather than pushing for immediate bonuses).

This better advice could come about because you directly prompt AI in that direction or because AI can naturally infer that from the context of your goals. Either way, you don’t get it without first having clarified your goals.

Step 3: Use AI to help you understand how to decide (not what to decide)

Once you have done the reflective work, this is the right time to bring AI into the process. Work with it by challenging your assumptions and the thinking you’ve done so far:

• Generate alternatives you might not have considered.

• Get it to argue for every option (especially those you don’t like).

• What information might be missing?

• What could go wrong with your preferred option?

• What might the people affected by this decision think?

• What might someone with different values do and why?

And when using the AI in this way, make sure you have taken some time to educate yourself on the system’s limitations. For example, does it have the tendency to agree with you and increase the risk your existing assumptions are reinforced?

Grapple with it with great skepticism. You are not asking it what to decide. You are using it as a thinking partner to help you work out how to decide. You still have to do the work of weighing up the AI output against your own goals and appetite for risk.

The people who make the best decisions using AI are those who know themselves enough to use it so it increases the probability they will reach their goals.

That starts with knowing what you want.

Modern decision science has taught us a lot about how humans make decisions. Most of it is pessimistic. The research focuses on what goes wrong: the biases, the shortcuts, the mistakes, the overconfidence. Despite this, decision-making is a skill anyone can get better at.

The only thing you can control is the quality of your decisions. The process that takes you from perceiving a problem to doing something about it. Most bad decisions are a failure of this process. To make better decisions you need a better process, but you can’t fix what you don’t understand.

This page is my attempt at some organisation. Having a basic decision framework can help protect you against the many ways decisions go wrong. What follows is my personal framework for thinking about decisions. It blends ancient Chinese philosophy, which I studied during my years in China, with behavioural science, which I later applied in my career. It is the most useful way I have found to think about the decision process. I have had great success using it myself and using it with other people. The more your decisions follow this kind of process, the better your outcomes are likely to be.

The short version:

• Heart-Mind: Everything you think and do is a combination of your heart (the feely stuff) and your mind (the thinky stuff).

• Goals: Every decision happens within the context frame of your goals.

• Perception: A specific decision process begins the moment you notice a problem or opportunity.

• Evaluation: You then define the decision by finding out what you know and what you value.

• Deliberation: You generate options and work out which one gives you the best chance of reaching your Goals.

• Action: You commit to the best option through taking action.

• Reflection: You become a better decision-maker by consistently asking yourself questions about your Goals and decisions.

The full version:

Heart-Mind

One of the biggest things I took from my time in China was how differently emotion and reason are treated in Classical Chinese thought. In ancient Chinese, xin (心) is best translated as heart-mind, a single faculty that felt, thought, and valued all at once. Knowing something and caring about something are two parts of the same thing.

This seems to me like an accurate description of what it means to be human.

I think the human brain is best understood as a general purpose tool. And decision-making is a general purpose activity. You have the ability to think about and decide between inherently incomparable inputs by converting them into a single internal value scale. For example, you can decide if you would rather have an extra 500 dollars per week or an extra hour with your family per week.

We don’t have multiple systems of reason and emotion that work independently. Rational thinking never happens in isolation. It is always shaped by your emotions, motivations, and relationships. It is a mistake to draw such a sharp line between thinking and feeling, and treating emotion as the problem side. Humans are emotion and reason, not error and reason.

Humans make decisions. Any useful decision-making framework has to match how humans actually work. To be human is to be, at all times, shaped by both your heart and your mind. How you feel about something is actually relevant input in the decision process. Emotions are not obstacles to clear thinking. They are what make clear thinking possible in the first place. Our emotions set the frame for how you think and solve problems. They determine which questions you even think to ask, and which goals feel worth pursuing. A good decision requires both your head and your heart to be part of the process.

So emotions give you real, useful data about what you value and how your decisions are going. But like any data source, they can also mislead you. Good decision-making comes from properly integrating your emotions with rational thought. Keep asking yourself: are my thoughts and feelings both playing an equal role here? Things tend to go wrong when one dominates the other.

Goals

Every decision is an attempt to move you from where you are to where you want to be. Everyone has goals: a fulfilling career, security, good health, strong relationships. From the perspective of decision-making skill, your decisions are only “good” or “bad” based on how much they move you towards your goals. Your goals are the context frame for all decisions.

Because goals feel like a destination, it’s tempting to think of them as separate from the process. You think, you decide, you act, then you ask: did I hit the goal? This is intuitive. It is also mostly wrong.

Goals sit above and around the entire decision process, filtering what seems worth noticing (Perception), what information seems relevant (Evaluation), which options seem worth pursuing (Deliberation), and what you are willing to commit to (Action). You are always deciding within the context frame of your goals.

It is useful to distinguish between two kinds:

• Ultimate Goals: these are things valuable in themselves. Meaning, security, happiness, connection. They are the point of everything else. You do not want them because they get you somewhere; they are the somewhere you want to be.

• Enabling Goals: these are the things that move you towards your Ultimate Goals. Money is the obvious example. Nobody really wants money for its own sake. They want what money makes possible: security, freedom, options.

You should be able to state your goal in one sentence. The goal should also be specific and tied to Action. This is so you can know when they have been achieved. Vague goals produce vague decisions which produce vague outcomes.1 At that point you are just lost.

You should also be able to clearly differentiate between your Ultimate Goals and Enabling Goals. This includes being able to articulate how they relate to each other causally. Get clear on what you actually value most and try to build a coherent order of goals from that. When your different motivations know their relative place in that structure they will be more likely to work together rather than against each other.

A surprising amount of bad decisions come down to mixing goals up. It is possible to spend lots of time and effort chasing an Enabling Goal you have mistaken for an Ultimate Goal. Think about the last time you achieved something you desperately wanted and felt underwhelmed.

Reflecting carefully about your goals is one of the most important parts of making good decisions. It is the main tool for making your goals explicit and you are more likely to achieve explicit goals than subconscious ones. When you get clear on your long-term foals, it gives your emotions something coherent to appear around. Without that clarity, your emotions are more likely to reinforce your biases that pull you towards the wrong things.

Working out what you truly want isn’t as simple as following your gut or listing your preferences. You attend to everything that matters to you: the feely stuff and the thinky stuff (the heart-mind). A good decision-maker isn’t simply someone who makes a lot of good individual choices. It is someone who has a coherent sense of who they are, where they are trying to go, and makes decisions that move them closer toward that.

1. Perception

There is no decision to make until you have first noticed something that requires one. Perception is the start of the process for a specific decision.

You cannot possibly attend to everything, so you attend to the things connected to what you value and care about. What you notice is the product of who you are, what you have reflected on, and what you have cultivated over time. In other words: your Goals structure your perception.2

Guided by your goals, you will notice something that kicks off the process for a specific decision. That something is either a problem (something potentially bad) or an opportunity (something potentially good). I will refer to these as Something throughout. Partly for convenience, and partly because the difference is can be unclear at first. The same Something can be either, depending on your goals and how you look at it.

Perceiving Something can happen:

• actively (you go looking for a problem to solve or an opportunity to take advantage of); or

• passively (Something arrives and demands your attention without any effort on your part).

Humans are not neutral receivers of information. We perceive things with implications already attached. When you hear a knock at the door, you don’t hear an isolated sound.3 Your experience is that the sound came from somewhere and requires a response. Who is it? Do I answer? Cause, effect, feeling, and intended action are already built in. In this sense, we are always deciding.

Once you’ve noticed Something, a second question follows: how much attention does it actually deserve? You have to decide how much mental energy to give it.

This can be done automatically and unconsciously if the Something has a very obvious interaction with your goals. Your emotions carry real information worth paying attention to. Negative emotions can be a signal that something is pulling you away from your goals. Positive emotions can be a signal that you are moving in the right direction. The decisions that matter most for your life will often start from emotions, not abstract rationalising about what you want.

If done consciously, then this allocation is itself a kind of mini-decision, and like all decisions, it can be made well or poorly. Treating everything as important leaves you stretched too thin. Treating nothing as important means you miss everything.

In summary, Perception is where you notice Something and decide whether it’s worth moving further through the decision process.

Subscribe now

2. Evaluation

Once you have perceived Something, you then need to define the decision to be made. This involves clarifying the scope of what is in your control and removing irrelevant factors. This kind of Evaluation ensures you are answering the right question. It is better to make a mediocre decision about the right question than a good one about the wrong question.

Two questions structure this stage.

What do I value? (Heart): This requires looking inward and can’t be outsourced.4 You must be honest about who you are and what matters most. What is important to you? This is largely determined by your goals. Similar to what was said in Perception, pay attention to the signal your emotions are giving you.

What do I know? (Mind): This is largely an external question, about facts, data, and the nature of the situation. Look at a range of unbiased sources. It is crucial to keep in mind that this also includes what you do not know. Answering this question can surface things that improve your chances of a good decision: missing information, people you should consult, or the recognition that what is in front of you is a symptom of a deeper problem.

Your biases will make fact-finding hard and your values can often pull you in opposite directions. With your emotions in particular, they are crucial data points but like any data source they can also mislead you.

Your best weapon against this is training good cognitive habits and reflecting on the situation5: testing your knowledge against what you do not yet know (what remains uncertain and what assumptions am I making?) and aligning your values with your goals (does this actually move me toward where I want to be?).

In summary, Evaluation is where you start defining the decision by reflecting on what you value (heart) and what you know (mind).

3. Deliberation

Everything else has been preparatory: perceiving Something, calibrating how much attention it deserves, establishing the relevant facts, clarifying your values. Deliberation is where you use this to work out the best way to meet your goals.

Deliberation has two steps:

Exploration: the work of generating options.

Judgement: the work of analysing which option is best.

Try to generate multiple (minimum 3) options that are sufficiently varied. You should make sure all options can actually be executed (more on this in the Action stage below). You may find exploration forces you to dip back into the Evaluation stage to gather more information.

Try to do as little judgement as possible during exploration. There is a temptation to judge the quality of each option as you generate it, but this produces worse outcomes. You end up with a shrunken option set and you construct a case for one of the first plausible options rather than the best one available.6

Sometimes judging the best option is simple enough that you don't need much more than a cost-benefit analysis. You are choosing between pasta and a cheeseburger and you don’t like pasta. You win a raffle and can take either $10 cash or a $150 theme park ticket. The value difference is enough so you take the ticket and move on.

But most real decisions rarely offer us that kind of simplicity. Perhaps you are scared of heights. Who would accompany you to the theme park? Would you go alone? Perhaps the theme park is hours away and you don’t own a car. The best choice often becomes something that needs to be weighed against circumstance, values, your relationships, your time, and how you actually feel about spending half a day standing in queues.

Humans are gifted in that they can think about vastly different things on a common value scale. Returning to the question: Would you rather have five dollars or spend ten minutes cuddling a puppy? Your answer to that question is less interesting than the fact you can answer it. These things share no common unit. There is no puppy-to-money conversion rate. And yet you had a response, probably within a few seconds.7

Therefore, Deliberation can feel less like a calculation and more like a conversation between the available options and your whole self. The best option is the one that gives you the best chance of reaching your goals. And your goals are your goals.8

When it comes to selecting your choice, it helps to get specific about how confident you are. Rather than thinking "this is the right choice," try asking "how likely is this choice to get me where I want to be?" Putting a rough number on it, even just "more likely than not" or "maybe one in three", forces clearer thinking. It also makes it easier to reflect well afterward, because you've made your expectations explicit before the outcome happens.

In summary, Deliberation is where you use all your previous work to generate viable options and analyse the best one.

Subscribe now

4. Action

The transition from Deliberation into Action is really about the moment of commitment. One of the misleading quirks of everyday language is that we say we have "made a decision" when what we mean is that we have finished Deliberating. But making judgement and deciding are not the same thing. Judging produces a preferred option. The preferred option does not become real until you take some action in the world.

I always find the etymology of the word decide helpful to keep in mind here. It comes from the Latin dēcīdere, meaning "to cut off." A decision is to commit to a specific path and sever all others. Notice how often people announce “we have decided to…” without any accompanying action. You can intend to do something all you like, but you are still in the Deliberation stage until you take action.

A useful test for whether Deliberation is complete: you should be able to give a coherent account of the process so far. This includes the options generated, the alternatives considered, the values weighed, the probabilities estimated, and why specifically the chosen option has the best chance of meeting your goals. You must also be willing to act. Until you act, you are still deliberating. There is a difference between extending deliberation because you have noticed a genuine issue, and procrastinating because you are not ready to commit.

The willingness to commit is what adds complication to this otherwise simple stage. We don't just do the thing we judged is best because we don't have full control of ourselves. Deliberation alone does not produce action. There is other stuff at play.

Motivation is the most obvious. You can identify the best option with perfect clarity and still do nothing. Motivation is not a rational output. It cannot be calculated into existence or willed on demand. Courage might be another, when the decision involves real risk. Or self-control, when something appealing seduces you away from your ultimate goals.

Your situation also has immense influence over what you do, often more than your intentions. Research in decision science shows that people behave differently depending on the environment they are in. The same person will act differently at home, at work, under stress, or around other people. Motivation and willpower are not always enough. You also need to shape your environment so the choice which takes you towards your goals becomes the easy choice. Think about what makes that easier or harder, then make small changes that help it become more likely.

Overall, your ability to act depends on how well you have reflected on your goals. The more you have thought through your goals, the more you believe in them, the more internal contradictions you have resolved, how you might best achieve them, the more successful your actions become.

Reflection

Reflection is the act of thinking in questions. If done correctly, it has the greatest long-run impact on the quality of your decision-making. Reflection is the most valuable tool you have to improve your decision-making skill because it matches how we actually experience problems. When we’re stuck, we don’t immediately reach for a rule. We wonder. And we do it more or less constantly.

Think about what your inner life sounds like on any given day. You wake up and the narration has already begun. Planning, worrying, replaying. You can’t find your phone and you will literally ask yourself where did I put it?. Your daily life is an endless question-and-answer broadcast. Reflection is about leaning into this and using it to improve your decision-making.

There are three kinds of reflection worth distinguishing.

The first is reflection on your life and goals more generally. Given how important goals are to the decision process, this kind of reflection is fundamental. Your goals are not fixed. They get discovered, revised, and abandoned through the ongoing process of reflecting. Without reflection, the context for your entire decision landscape will be wrong.

The second is reflection during each specific stage of this decision process. These are the questions you ask yourself before, during, and after, to check whether you are handling each stage well enough. If you want to learn from a decision, you have to be examining it before the outcome has had a chance to edit your memory of the process. Keeping a record of your reasoning before the outcome is known is helpful.

The third is reflection after the decision. You examine what the outcome produced and try to extract learning for future decisions. This is where the record of your decision process gives you immense value. If we reflect on that, we obtain understanding. If we do not, we will not.

This third type is the hardest, because outcomes in the real world are messy. The goal is to distinguish between outcomes that reflect a genuine error in your process and outcomes that are the product of luck.

Not everything that goes wrong means the decision was poor.

Not everything that goes right means the decision was good.

Sometimes the process was as good as it could have been and the outcome was bad because that is what probability looks like in an uncertain world. You can't control luck.

Good reflection might be asking questions such as:

• Did I perceive the right Something?

• Did I give it the right amount of attention?

• Did I identify the relevant facts?

• Did I weigh my values honestly?

• Did I generate enough options?

• Did I judge the best option correctly?

• Did I ask the right questions?

• Did I adequately consider other people?9

Reflection is also heavily neglected. Either because the outcome was good and we assume there is nothing to examine, or because the outcome was bad and we would rather not revisit it.

When done consistently, reflection will directly improve your future decision-making. Reflection gives you a distillation of what works. Every future decision you make can start from a more advanced position if you have updated your beliefs from reflection.

Reflection is how you take every decision and use it to become a better decision-maker than you were yesterday.

Thank you for reading.

If you want to think more carefully about how you make decisions, this newsletter is free. Every Monday morning, something short to help you start the week deciding a little better.

Subscribe now

I have been working on decision-making for over a decade now. I spent years at university researching it and chose my career because of my passion for it. A few years into that journey, I wanted to move beyond theory and try something I could actively experiment with. A real competitive environment where I could test whether these ideas actually worked. In 2019, that search led me to Fantasy Premier League (FPL).

Now, seven years later. The results have been great (or very lucky!). Over the past four years, I have the best rank out of 10+ million people who play.

There is a tendency to assume you do well in fantasy sports by knowing more about sports and data. I don’t think that is the case. I didn’t really know anything about football when I started. I believe what matters most is the quality of your decisions.

I have written about decision-making more generally (e.g. personal life and work) in my framework, which you can read here. This article is a more personal account, an exploration of what seven years of competitive fantasy sports have taught me about decision-making.

Fantasy sports are, on the surface, an incredibly silly hobby. But it is a very instructive laboratory for decision-making. If you have never played fantasy sports, don’t worry. This piece is about what it reveals about how we make decisions.

Incomplete Information

It is difficult for us to shake the assumption that good outcomes mean good decisions. In systems where information is complete and luck plays little role (e.g. chess), that is true. But activities like fantasy sports, poker, investing, real life, are not like that.

You never have all the information. You don’t know which player will get injured, which team will get a penalty, which player will get a red card. Playing fantasy sports is similar to picking stocks: you are always deciding with incomplete information, and remote events entirely outside your control will regularly impact your outcomes.

This gap between decision quality and outcome is where you learn interesting decision-skills. People who improve are the ones who understand this gap and develop a process of reflection that works around it. They spend a significant amount of time evaluating their choices independent of their results. They learn to ask not "did it work out?" but "was that a good decision given what I knew and what I valued at the time?"

Thousands of Decisions

Over a full season, fantasy sports gives you thousands of decision points. You have limited resources each week and dozens of different options available to you. Most areas of life offer relatively fewer decisions, with consequences spaced out over a longer time period. Fantasy sports compresses this. In a single season, you encounter an enormous range of decision types and get the opportunity to observe your own tendencies across all of them.

This allows you to create a "decision archive". If you document your reasoning before outcomes are known, you can start to review patterns in where you go right and wrong, and use each decision as a small data point for refining the next one. The habit of structured reflection is, in my view, the single biggest tool for improving decision quality.

Expose Your Biases

We all carry cognitive biases we are not aware of. Fantasy sports makes them visible because the structure of the game creates conditions in which they flourish. Here are a few:

• Sunk-cost bias: irrational choices because of costs that have already been incurred. This shows up when you hold on to a player you are too invested in to let go of, even when the rational move to let them go is obvious.

• Hyperbolic discounting: the pull toward immediate reward over better long-term outcomes. This appears every time you make a short-sighted move focusing only on this week rather than thinking long-term.

• Loss aversion: the tendency to feel losses more sharply than equivalent gains. You can only own 15 players so there will always be more negative events than positive events.

• Mere exposure effect: preferring what is familiar. You are reminded of this when you over-trust players you have had success with before, even when the evidence no longer justifies it.

What makes fantasy sports genuinely useful as a training environment is that these biases have immediate and legible consequences. You can trace an outcome back to a specific decision made under a specific emotional or cognitive condition. A knee-jerk decision made after a goal is scored or making choices when you are sleep-deprived. That traceability is not always available in real life. Experiencing this frequently allows you to learn a lot about yourself and develop strategies to counteract these tendencies.

Having a clear decision process you can stand behind is an antidote to a lot of this. It can give you ‘decision sanity’. When you have documented your reasoning in advance and you understand what was luck versus what was your own judgment, a bad outcome loses much of its power over you. You can look at what happened and assess it rationally rather than react to it emotionally.

Thinking in Probabilities

Fantasy sports force you to make decisions under uncertainty. You can't find certainty by delaying, procrastinating, or hedging. Every week you have to commit to a choice before the deadline.

This trains a particular kind of thinking that is enormously helpful in life: thinking probabilistically. You have to get comfortable assigning rough probabilities to uncertain outcomes. This forces you to be explicit about what you actually believe and how confident you are. What is the likelihood something will happen and how big will that impact be?

Fantasy punishes thinking in binary terms. The best decision-makers are those who learn to deal with uncertainty properly. To think in distributions rather than outcomes. To make decisions that are right in expectation even when they won’t always pay off individually.

Having good predictive data sets can sometimes give us a false sense of certainty. You still have to think about your decision. Learning to think probabilistically about your decisions is an entirely separate skill from having good data models.

Experience Is Not Enough

One of the strongest pieces of evidence that fantasy sports is primarily a game of decision-skills is that those who have played for a long time are not meaningfully better than those who start later. Experience alone does not produce improvement.

The reason is that without honest reflection, experience teaches you the wrong things. You credit wins to good thinking and losses to bad luck. You adjust your strategy based on outcomes rather than the quality of your reasoning. Experience without examining your process isn’t all that useful.

What produces genuine improvement is deliberately analysing your decisions separately from their outcomes. That means being honest enough to grapple with things you don’t want to see. Being able to label a decision as poor even when the outcome was good. When you ignore a poor decision, you give up the opportunity to extract its lesson.

Without failure there is no learning and without learning there is no improvement.

Thank you for reading.

If you want to think more carefully about how you make decisions, this newsletter is free. Every Monday morning, something short to help you start the week deciding a little better.

Subscribe now

If you are interested in understanding about the decision process more generally, I recommend reading my decision framework.

Read The Framework

This is the resource list I wish I’d had earlier.

I have organised into three levels: beginner, intermediate, advanced.

Each level has a book and a podcast so you can learn in whatever format suits you.

Work through in order or jump straight to the level suitable for you.

Beginner

Book: Thinking in Bets by Annie Duke

This is always my first recommendation to everyone. Excellent introduction to thinking probabilistically and understanding decisions as a prediction of an uncertain future.

Podcast: Choiceology

Probably the most accessible podcast on decision-making. Does a great job of blending the science of decision-making with engaging stories and good production.

Intermediate

Book: Thinking, Fast and Slow by Daniel Kahneman

Most of the popular modern psychology books were inspired by this. It is not the best reading experience (long, dense, and challenging to finish) but there is a lot of wisdom about human behaviour in this book.

Podcast: The Decision Education Podcast

This is a good step up from Choiceology. The website is useful as it provides a transcript and show notes for each episode, including links to resources relevant to that episode’s content.

Advanced

Book: Judgment in Managerial Decision Making by Max H. Bazerman and Don A. Moore

Probably the highest value per word resource you can get on decision-making. There is a reason this is used so widely in universities and business schools. It is essentially a textbook so it isn’t bedtime reading (includes decision exercises).

Podcast: The Decision Corner

In-depth interviews with academics and industry experts. Similar to the Decision Education podcast but slightly less approachable. Unfortunately the podcast is no longer active but there is still 40+ hours of content to listen to.

Thank you for reading.

If you want to think more carefully about how you make decisions, this newsletter is free. Every Monday morning, something short to help you start the week deciding a little better.

Subscribe now

If you are interested in understanding about the decision process more generally, I recommend reading my decision framework.

Read The Framework