The Ledger
Every statement that passed quotation and attribution checks. Mix any filter with any other: certainty 1/5, debate potential 5/5, or both at once.
why aren't all 103 resolved? a statement only gets an assessment when the public record can support or contradict it. opinions and what-ifs never can, and 0 checkable ones are still open, waiting for their date. predictions held up or didn't; assertions are supported or contradicted. on every card: ▮▮▮▮▮ certainty · ▮▮▮▮▮ debate potential. speakers are clickable
Duke: Measure decision impact by the time horizon of a bad outcome
“The longer the time horizon in which the bad outcome is going to have actually had an impact. On whether you are happy, whether you are advancing toward your goals or not, that is going to tell you something about the impact of the decision.”
Duke: People Spend 6–7 Work Weeks Yearly Choosing Food, Clothes, and Shows
“The interesting thing about what to eat in a restaurant, what to wear, what to watch on Netflix, is that I think if you add just those three things together, I think it's something like between six and seven work weeks that we spend per year on those decisions…”
Duke: Rapid execution on repeating choices speeds predictive model accuracy
“So the more often that you're making a similar type decision, you want to go fast because it speeds up your ability to build models that will actually help your decision making going forward in terms of making those things more accurate.”
Duke: High-impact decisions can be made faster when easily reversible
“But even for higher impact decisions, if it's easy for you to quit and reverse course and get back to another option, then you can actually speed your decision making up simply because there isn't as big a penalty for being wrong.”
Duke: Parsing close-call options is impossible and a waste of time
“All we've determined is that they're close, and even if we could determine it, it's a bad use of our time. Because they are so close that you're probably going to be kind of about the same happiness either way, at least in expected value, and so you should pro…”
Duke: Additional information rarely flips leading decision choices
“Is there a piece of information that I can find out That would actually make it so that some other option became clearly better than the option that I'm considering. And you'd be surprised. Most of the time the answer is no. In those rare cases when the answer…”
Duke: People underestimate luck when imagining their own success
“And we know that when we think about like illusion of control, for example, we think when we're imagining our own success that the role of luck will be less than it actually is.”
Duke: Pre-planning downside responses enables rational action during crises
“So by imagining in advance what your plan of action is, if there's a downturn, if the downside realizes, you're actually going to tend to be acting more rationally. You're going to be prepared, and that's going to allow you to be more nimble rather than to be …”
Duke: Decision-makers perform worst when facing immediate choices due to rationalization
“So when we're actually facing a choice, generally that's when we're at our decision-making worst, because that's when the most rationalization can happen.”
Duke: Investors should pair strict category rules with a secondary updating process
“Is make a category decision, hold that strongly, but then have a secondary process that's going on where you're really trying to think about the world and incorporate information in a rational way that will allow you to maybe alter those categories in the long…”
Duke: Investors should not stray outside core competency without new models or hires
“If you have competencies in a particular type of investing, just straying out of that category is just a really bad idea. If you specialize in REITs, you probably shouldn't be straying out of that unless you have a secondary stream, right? Unless you like hire…”
Duke: Qualitative probability terms create false consensus and hide substantive disagreement
“This is just one of these little ways where when we speak to each other, we may think we're agreeing with each other, because we may both agree that the term real possibility applies to the likelihood of a particular future occurring, but it turns out that we …”
Duke: Group Deliberations Waste Time Lingering on Consensus Points
“When you're in a group meeting, a lot of the time is spent on agreement. Oh, you know, Ted, you say something, and then I say, oh, I totally agree with Ted, and let me now say it in my own words, why that's such a smart idea. And then the next person says, yea…”
Duke: Decision-Makers Should Be Grateful for Dissenters and Skeptics
“If the person is in the role of the sort of person who's pushing back on it, You should kind of be grateful that that person exists because it forces you to be thinking about things from different perspectives. And like I said, it forces you to actually be abl…”
Annie Duke: Leaving academia for poker was a huge mistake
“I think it was a huge mistake for me to leave academia and become a poker player.”
Duke: Extra data past a tipping point inflates confidence, not decision quality
“Because it's very rare that you're going to get to that much certainty and there's diminishing returns as you collect more data. And very often what happens is once you get to kind of a tipping point, it's not so much the quality of your decision is going to g…”
Duke: Decision-makers must prioritize speed over certainty in exponential crises
“What we know is that we should be favoring time over certainty, so we should be trying to act and putting like a lot of hedges in place and doing things that are a little bit more extreme when we don't have a lot of data, because the one thing that we do know …”
Duke: Acting in complex systems creates data that modeling cannot predict
“So it actually behooves you to actually make lots of decisions and do lots of actions In order to get the world to give you the data back so that you actually can find out how the world would respond if you did certain things.”
Duke: Ramping COVID testing early was a near-free hedge relative to GDP
“At that time, in January, it's very uncertain what the course is going to be here. Is it going to get here? Is there going to be community transfer? There's just all sorts of, like, the range of outcomes is super, super, super wide at this point, so we're real…”
Duke: US schools pushed probability math from kindergarten to eighth grade
“So when you look at the curriculum for most K through 12 in 1980, starting in kindergarten, you were learning probability. And now the first point at which probability is introduced in the math curriculum in the United States is in eighth grade.”
Duke: People wrongly treat outcome quality as a proxy for decision quality
“We don't like randomness as a species, and so once we know the outcome, it's like we can't go back and kind of reconstruct
the
The decision tree in any kind of real way, and so what ends up happening is we use the outcome quality as too perfect a signal for …”
Duke: Every decision allocating limited resources under uncertainty is a bet
“We can think about any decision we make as a bet. My book is called Thinking in Bets, and so this is obviously the premise of it. So people don't tend to think about their decisions as bets, because they think about bets as confined, like they think about it i…”
Duke: Leaders must withhold their beliefs to get unbiased team advice
“When you're trying to get high fidelity advice from your team, it's really important actually that you keep your beliefs to yourself as you're trying to work a decision and allow them to sort of speak freely.”
Duke: Premortems redefine teamwork to include exploring failure modes
“So I think this idea of doing these premortems are really important because it shifts what it means to be a team player into something where you're willing to explore the way that things can go wrong.”
Duke: Mapping uncertainty yields better long-term results than focusing on outcomes
“What matters to us is that we foresee the landscape much more than that we get a good outcome, right? Because we assume that if we foresee the landscape, and we go through these exercises, and we're willing to do this together, that we're going to end up on th…”
Duke: Underforecasting upside harms capital allocation as much as missing downside
“If your model is forecasting low, that's as big a problem for your ability to actually allocate your capital in a really smart way as if your forecast is coming in wrong the other way.”
Duke: Cognitive psychology directly improves decision-making under uncertainty
“What I'd been doing in graduate school actually had been incredibly helpful for me in poker and could be helpful in general to understanding how do you make decisions, you know, particularly under conditions of uncertainty.”
Duke: All human decisions are fundamentally bets allocating scarce resources
“So first of all, I say it's really easy because all decisions are actually bets. It's just a matter of making that explicit.”
Duke: Humans believe information upon hearing it and vet it later
“So what Dan Gilbert showed in the nineties was, no, no, no, no, no. It's just like perceptual belief. You hear it, you lodge it as true, and then maybe later on you vet it. So now we believe what we hear.”
Duke: Humans evaluate evidence to confirm beliefs and discredit contradictions
“The problem is that there's all sorts of work that shows that you just don't do step three, and that's where we really get into trouble, is that we hear it, we believe it, and then, like, maybe if we have the timer inclination later, we might do some vetting o…”
Duke: Expressing outcomes with probabilities makes leaders more believable than projecting certainty
“So not only do you end up with a richer conversation, but you're more believable as a communicator. So I think it's a little counterintuitive, but our intuition isn't always good. So, I mean, our intuition is that when I hear something, I vet it before I decid…”
Duke: Framing decisions as bets transforms dissenting information into helpful calibration
“What a bet does is it shifts your idea of what winning means. So winning is now actually having the most accurate representation in the world. And that really comes through the accountability that this idea of betting is, because when you challenged your daugh…”
Duke: Decision-makers must form accountability groups because others spot their biases better
“Other people are actually pretty good at spotting our biases. So in the same way that you sort of naturally recruit people into the process with you when you express things probabilistically, why not do that intentionally and just purposely recruit people into…”
Duke: Effective decision-making groups require only two dissenters and one referee
“You actually only need three. You need three because you need two to disagree and one to referee.”
Duke: Effective decision-making requires overcoming temporal discounting via mental time travel
“But what you really Learn when you start thinking about how do you become a really good decision maker is that the best thing that could happen is for past Marty to run into future Marty and have a little discussion, have a little, you know, chit chat about ho…”
Duke: Investors mismanage risk by overestimating their edge and underestimating volatility
“We're bad at it for a couple of reasons. One is that we generally overestimate our edge. Let's be honest. We think we're better than we are. We also underestimate the vol. I mean, we tend to. So we generally think we need a lot less money than we do, and then …”
Duke: Accurate behavioral predictions require combining physical cues with historical baselines
“So what's good to do is sort of take these kind of physical cues and merge them with the story that they've kind of told you in the past as you've gotten a feel for how they behave in different situations, and that now you can merge together to make some sort …”
Duke: Reversible two-way door decisions provide greater margin of error
“Having a two way door decision, one that you can reverse, or let's call it quit, makes it so that you just have more margin of error
On the initial choice.”
Duke: Most mountaineering fatalities occur during the descent
“And most of the dangers are on the descent. That's when most people die is coming back down where you're tired and you don't have as much adrenaline and obviously your oxygen is low.”
Duke: The option to quit enables people to start new endeavors
“It's the option to say, that date wasn't very good. I don't want to go out on another date with you. That allows us to go out on dates, right? And it's not just dates. It's like we can take a job because maybe if we don't like it, or it turns out the boss is t…”
Duke: Loss aversion stops starting, while sure loss aversion stops quitting
“Loss aversion is not wanting to start things for fear of the losses you might incur later. So let's think about loss aversion as something that stops you from starting. Sure loss aversion, as Kahneman says, talks about, is not wanting to turn a loss on paper i…”
Duke: Sears once represented one percent of total US GNP
“At one point, they represented one percent of the total GNP of the United States.”
Duke: Outside observers evaluate failing projects better without sunk cost bias
“Basically, it's like, I don't carry with me all the stuff that you carry with you, all the cognitive debris that you carry with you, when you've already made that allocation decision to a particular division. Because I didn't make the original allocation decis…”
Duke: Harvard Study Showed Hand-Raising Artificially Created Supermajorities
“What they found with the group that was raising their hands in class was that super majorities formed very quickly. In other words, consensus formed. Once people sort of saw consensus start to form, people were sort of going with the consensus. I think partly …”
Duke: Surveyed probability interpretations for 'real possibility' ranged from 16% to 81%
“The biggest spread I ever got was on real possibility, where the range was 16% to 81%, and it wasn't like Three people in the group were like up toward 81 person was at 16. It was actually spread out across them.”
Duke: Simple card games like Go Fish teach kids probability and strategy
“If you play Go Fish, you can teach them about probability, and you can teach them game theory and strategy through, like, really simple games like that”
Duke: Preexisting beliefs drive information processing rather than facts shaping beliefs
“It's that we form beliefs in a relatively haphazard way, and then those beliefs actually drive the way we process information. So what ends up happening is kind of twofold. One is that we notice information that confirms the beliefs that we already have. So yo…”
Duke: Future-focused questions prompt honest post-mortems without defensiveness
“Now, notice that in order for you to actually reasonably answer that question, you must go back and look at where you went wrong in the first place, because you can't answer that question without doing that. But I'm not the one making the challenge to you. I'm…”