- [Derek] Friday, the 17th of July, 1992.<br>Amidst the chaos of the trading floor<br>at Singapore's international<br>stock exchange,<br>one of the junior traders<br>makes an expensive mistake.<br>Instead of buying 20 futures<br>contracts for a client,<br>she sold them instead,<br>costing Barings Bank nearly $40,000.<br>To save her job, Her boss, Nick Leeson,<br>a young trader keen to make his mark,<br>decides to hide the loss.<br>He puts it in an obscure error account,<br>that's account 88888.<br>It's used by banks to solve<br>small discrepancies and trades.<br>It's a dangerous move<br>and it should have been picked up<br>by the central office immediately,<br>but when nobody from Barings<br>notices, he gains confidence,<br>convinced he can win back the loss<br>and get the team out of trouble.<br>- I'm operating in the belief<br>that I can get it back.<br>You know, maybe it goes back<br>to the confidence thing.<br>I don't know.<br>I mean, I'm confident I can<br>get it back at that stage.<br>- But Leeson's staggering overconfidence<br>would create a debt of<br>gargantuan proportions<br>and lead to one of the biggest<br>collapses in banking history.<br>- It's easy for me to make<br>the case at overconfidence<br>as the most dangerous of the human biases.<br>Overconfidence gets us<br>into all sorts of trouble.<br>It leads us to take<br>risks, make commitments,<br>enter contests, try things<br>that will ultimately fail,<br>sometimes in costly,<br>embarrassing, and dangerous ways.<br>- Overconfidence has been implicated<br>in almost every big disaster,<br>from the sinking of the Titanic<br>to the Chernobyl nuclear disaster,<br>to the loss of the space<br>shuttle Challenger.<br>But overconfidence isn't reserved<br>for just a few reckless individuals.<br>We can all fall victim to it.<br>For example, 93% of us<br>think we are better drivers<br>than the median, which<br>is, of course, impossible.<br>The scale of the problem was identified<br>in some now-classic research<br>with a set of simple quiz questions.<br>True or false, Australia<br>is wider than the moon.<br>- Oh!<br>I bet they're about the same size.<br>- It looks small on a map, but it's huge.<br>- Are there more or less<br>stars in the Milky Way<br>than trees on Earth?<br>- More stars.<br>- Than trees?<br>- More stars.<br>- One teaspoon of pure olive oil,<br>- Right.<br>- Four and a half grams,<br>contains more chemical potential energy<br>than an equivalent four and<br>a half gram amount of TNT.<br>- I'm gonna say true.<br>- Hmm.<br>Oh no.<br>(Derek laughs)<br>- This feels kind of<br>like a trick question.<br>- The energy density of butter<br>is 10 times that of<br>lithium batteries, right?<br>I mean, lithium batteries are rubbish.<br>Everyone's going on about them.<br>(Derek laughs)<br>Just run your car on butter, I say.<br>- True or false, plants are more efficient<br>than the average solar panel<br>at absorbing the sun's energy.<br>- Yeah.<br>- They must be.<br>- Now, these aren't the exact questions<br>from the original research,<br>more on those later,<br>but the specifics don't really matter.<br>The interesting bit is<br>what they asked next,<br>how confident are you that you are right?<br>- 100% confident.<br>(interviewee speaks faintly)<br>- 100% confident?<br>- 100, yes.<br>- I think it's true.<br>So, I've also gotta go 90%.<br>- 100%.<br>- 100%.<br>- 100%?<br>- Yeah.<br>- I don't think we actually<br>had any expectations.<br>This was very early on in the research<br>on confidence assessment,<br>but we didn't really<br>know what we would find.<br>- The results revealed<br>a surprising disparity<br>between how accurate<br>someone thinks they are<br>and how accurate they actually are.<br>When people are 90%<br>certain, they are right,<br>only 75% of the time.<br>In fact, we ran our own<br>version of this online<br>and found an even more<br>extreme discrepancy.<br>We asked the Veritasium community a bunch<br>of science questions and then asked<br>how confident they were in their answer.<br>Our results showed that those<br>who were the most confident<br>describing themselves<br>as 91 to 100% sure were only<br>correct 51% of the time.<br>Since the original research,<br>these results have been<br>replicated again and again<br>in all areas from general<br>knowledge to motor skills,<br>and it affects everyone, even experts.<br>- I have a paper where<br>we analyze the survey<br>of professional forecasters.<br>These are chief economists<br>at various corporations<br>and banks who are invited<br>to forecast the state<br>of the economy on a quarterly basis.<br>We find that they are, on average,<br>too sure that they know<br>what's going to happen,<br>so they, on average,<br>say there's something like 53% sure<br>they have correctly predicted<br>what's gonna happen to<br>inflation, for instance,<br>and they are right, like, 23% of the time.<br>- How well what you think you know matches<br>with what you actually know<br>is called your calibration.<br>So, if you're perfectly calibrated<br>and, say, 80% confident,<br>you should be right 80% of the time,<br>but most of us are not well calibrated.<br>100%<br>- 100%?<br>- Yeah.<br>- There are 200 billion stars<br>in the Milky Way,<br>but there are 3 trillion trees on Earth.<br>- [Group] Oh!<br>(interviewee groans)<br>- 100% confident?<br>- 100, yes.<br>False! (laughs)<br>(interviewee mimics mocking tune)<br>(group laughs)<br>- Most of the time, being<br>overconfident isn't a huge issue,<br>but for Nick Leeson, the<br>stakes soon became very high.<br>Leeson's plan to recover his team's loss<br>was to bet that the Japanese<br>stock market would go up,<br>so he went long on the top<br>225 companies in Japan,<br>the Nikkei 225.<br>Ever since the Japanese asset price bubble<br>peaked two years previously,<br>the Nikkei had been dropping<br>from 38,000 to 16,000,<br>and he was confident the market<br>would soon bottom out and start rising,<br>but it continued to fall.<br>Over the next few weeks, it<br>dropped to a low of 14,000,<br>and the whole time, Nick<br>was betting the wrong way<br>with ever bigger and riskier bets,<br>he would double down,<br>betting double his losses,<br>so the next win would<br>bring him back above water.<br>This strategy should work<br>when the next win comes along,<br>but the market collapse was unrelenting,<br>and his losses ballooned from<br>$40,000 to around $3 million.<br>In spite of this, Leeson<br>remained confident,<br>and eventually, his luck turned around.<br>In the spring of 1993, the<br>market rebounded up to 20,000,<br>and by the summer, that<br>account was back in credit.<br>Leeson goes out to celebrate that weekend,<br>drinking and dancing on tables,<br>finally free of the hole<br>he had dug himself into.<br>But on Monday morning,<br>he made another trading<br>error on the futures market,<br>a damaging loss that Leeson<br>didn't want to admit,<br>so he put the loss back<br>into account 88888,<br>confident that if he'd got<br>outta the previous one,<br>he could do it again.<br>He began making risky trades<br>to try to recover that initial loss,<br>and as the losses mounted<br>in the five eights account,<br>it became harder and<br>harder to make new trades<br>to cover the last loss.<br>By the end of 1993, his<br>losses exceeded $30 million.<br>Leeson is obviously an extreme example,<br>but this tendency to<br>overestimate our abilities<br>is something we all share.<br>So, the question is why?<br>- The obvious explanation that<br>a lot of people default to<br>that is we wanna feel good about ourselves<br>and pretend like we're well-informed,<br>that we enjoy enormous satisfaction<br>from being able to say,<br>"I know,"<br>or even better,<br>"I told you so."<br>And so, we pretend to ourselves and others<br>that we know when we don't.<br>That is the motivated egocentric version<br>of an explanation for excessive<br>faith in our own judgment.<br>- [Narrator] Reichelt<br>had faith than the idea,<br>if nothing else.<br>- But what if it's something<br>more uncomfortable, stupidity?<br>(lively retro music)<br>Franz Reichelt was a tailor<br>with no formal scientific<br>or engineering training,<br>but he was obsessed<br>with solving the problem<br>of aviation safety.<br>He spent months designing<br>a parachute suit,<br>which repeatedly failed during testing,<br>but he became convinced the issue<br>was that he didn't have<br>enough height for it to work.<br>So, in February, 1912, he<br>took it to the Eiffel Tower,<br>and rather than using a dummy<br>to test it, he jumped himself.<br>Tragic, but also really dumb.<br>(soft electro music)<br>Do you recognize this graph?<br>- The, like, Dunning-Kruger graph?<br>- Are you confident that this<br>is the actual Dunning-Kruger graph?<br>- Not at all, in fact, I think<br>that this is not the Dunning-Kruger graph.<br>I think that this is a bit<br>of a communication trick.<br>- This is what people call<br>the Mount Stupid Curve,<br>which will pop up if you search<br>for the Dunning-Kruger<br>effect into Google images.<br>- As far as I understand<br>it, it's an effect<br>that says that when you<br>know very little about it,<br>your confidence is high,<br>and as you know more about it,<br>your confidence sinks until<br>you become a master at it.<br>- But I think that the actual result<br>is, like, a little bit<br>more fuzzy than this.<br>- It was.<br>This graph doesn't actually<br>show the Dunning-Kruger effect.<br>It's just a meme that<br>resonates with a lot of people,<br>and it became conflated with the effect.<br>In their original research,<br>Dunning and Kruger tested people<br>on tasks like grammar, logic and humor,<br>and then they asked<br>participants to estimate<br>how well they performed.<br>That produced this curve.<br>You can see that those who performed worse<br>had the largest mismatch<br>between their confidence and performance.<br>They were the most overconfident.<br>Those who performed the best<br>were actually slightly underconfident.<br>This suggests that<br>overconfidence may be linked<br>to how much we know.<br>Those who know less think that<br>they know more than they do,<br>but to me, this effect looks a little<br>like a statistical artifact<br>because while I was doing my PhD,<br>I asked students some physics questions<br>and I also asked for their<br>confidence in their answers.<br>And when I looked at the results,<br>accuracy spanned the full range<br>from nearly 0% to 100%,<br>but confidence varied over<br>a much narrower range.<br>All the scores were roughly in the middle.<br>This meant that poor performers<br>were the most overconfident<br>and the highest performers<br>were actually slightly under confident.<br>This is exactly what<br>Dunning and Kruger found,<br>so maybe overconfidence isn't entirely<br>due to how much we know,<br>but also because most of us express<br>at least kind of<br>middle-of-the-road confidence.<br>But there is another factor at play here,<br>which is how much information our brains<br>are capable of processing.<br>Monday, the 27th of January, 1986,<br>it's 5:45 PM at the Kennedy<br>Space Center in Cape Canaveral,<br>and the engineers from Morton Thiokol,<br>who made the Challenger space<br>shuttle's rocket boosters<br>make an emergency conference call.<br>They've just seen the weather forecast,<br>predicting temperatures<br>overnight will plummet<br>to 25 degrees Fahrenheit,<br>far colder than any<br>previous shuttle launch.<br>They know the rubber<br>O-rings that seal joints<br>in the boosters become less flexible<br>in the cold, but have just<br>a few hours to gather data,<br>create charts, and present their case.<br>Over the next six hours,<br>the engineers present 13 charts with data<br>on O-ring temperature, hot gas erosion,<br>joint rotation, and more.<br>But the data is scattered, incomplete,<br>and not synthesized<br>into a clear narrative.<br>They'd never tested below<br>53 degrees Fahrenheit,<br>so NASA managers were<br>trying simultaneously<br>to track historical O-ring<br>data, erosion patterns,<br>joint dynamics, seal resiliency,<br>pressure differentials, and more.<br>No single chart told the whole story.<br>So, overwhelmed by<br>seemingly contradictory data<br>and confident that their<br>rocket boosters were safe,<br>Thiokol management overruled<br>their own engineers<br>and approved the launch.<br>(fire roars)<br>At 11:38 the next morning,<br>73 seconds after launch,<br>(explosion booms)<br>all seven crew members were killed.<br>(chiming mysterious music)<br>- Calibrating your<br>certainty requires thinking<br>of all the ways that you could be wrong,<br>and that's hard for finite,<br>fallible agents like us<br>if it means considering<br>everything that we don't know.<br>- How many chunks of novel<br>information you can hold<br>in your head at one time is<br>your short-term memory capacity.<br>In 2008, Hansson, Juslin, and Windmann<br>investigated how<br>short-term memory capacity<br>was linked to accuracy and overconfidence.<br>Participants were asked to give ranges<br>for factual questions<br>like the length of a river<br>or the population of a city.<br>People's ranges were<br>consistently too narrow,<br>which effectively meant they<br>were being overconfident,<br>and those who had worse short-term memory<br>were more often wrong and more<br>likely to be overconfident.<br>Another study conducted by<br>Conte in 2023 asked participants<br>to keep sequences of letters in their mind<br>while they judged their own performance.<br>As the memory load increased,<br>confidence estimates became less accurate,<br>even for participants with<br>higher working memory capacities.<br>Together, these studies suggest<br>that assessing your accuracy<br>is a mentally taxing task.<br>So, overconfidence isn't<br>necessarily about arrogance.<br>It's your brain working at the limits<br>of what it can track and hold.<br>And because of this, your brain<br>can start using shortcuts.<br>Psychologist Daniel Kahneman describes<br>these mental shortcuts as<br>heuristics, that together,<br>create systemic errors<br>called cognitive biases.<br>One shortcut we use a lot is<br>substituting hard questions<br>with easier related ones.<br>Researchers have tested<br>this by asking students<br>how happy they were in their life<br>and how many dates they'd<br>had in the last month.<br>Unsurprisingly, these two<br>questions did not correlate.<br>There's a lot more to happiness<br>than matches on Tinder,<br>at least for most people.<br>However, if you switch it around<br>and ask about their dates first,<br>suddenly, the correlation jumps to 0.66.<br>Working out how happy you are in your life<br>is a difficult question.<br>You have to consider many<br>things and balance them all out.<br>So, when you're primed<br>with the information<br>about your dating life,<br>you're very likely to<br>substitute that hard question,<br>how happy am I in my life,<br>with, how many dates have I recently had?<br>Overconfidence from<br>misprocessing information<br>like this can be disastrous.<br>(low mysterious music)<br>So, why are our brains wired this way?<br>I mean, you might expect natural selection<br>to have wiped out the<br>confidently incorrect,<br>but there is evidence that overconfidence<br>can actually be advantageous.<br>Overconfidence can massively<br>improve your status.<br>In a scientific version<br>of "The Apprentice,"<br>scientists in 2012 compared<br>participants' own assessments<br>of their skill with objective measures.<br>They placed them in group<br>tasks to see who is chosen<br>as a leader and whose<br>ideas influenced the group,<br>tracking status over multiple sessions<br>and assessing whether the desire<br>for status led participants<br>to exaggerate their abilities.<br>The results were clear,<br>overconfident individuals<br>were more likely to lead,<br>assert themselves, and maintain influence,<br>even when their actual abilities were mid.<br>And the evidence certainly<br>shows people react better<br>to confident individuals.<br>Using an fMRI, researchers<br>at the University of Sussex<br>measured brain activity of people<br>after hearing unconfident<br>versus confident advice,<br>and those who listened<br>to the confident advice<br>had increased activity<br>in the ventromedial prefrontal cortex.<br>That's a brain region associated<br>with processing rewards<br>and expected satisfaction.<br>This means that human brains<br>are biologically tuned<br>to be influenced by confident individuals.<br>We literally feel better when<br>we hear confident people.<br>- But recognizing that dynamic<br>highlights a potentially<br>problematic incentive<br>for anyone who's contending<br>for positions that they want.<br>People who express more confidence<br>in employment interviews<br>or political campaigns, for instance,<br>do earn the confidence of<br>interviewers and potential voters,<br>even if they're being overconfident.<br>They can't actually deliver.<br>They don't actually know the answer,<br>but they can talk a good<br>line to impress others,<br>even if they can't ultimately deliver<br>on those grand assertions.<br>They know the audience<br>will place more faith in<br>them, if they express...<br>Well, they should express<br>maximal confidence.<br>- [Derek] Leeson knew this<br>and he took advantage of it.<br>While he was secretly racking up<br>huge losses in the infamous<br>five eights account,<br>he was publicly posting huge profits<br>and everyone believed the illusion.<br>- They come in and they<br>don't test any records,<br>so I can't be happier.<br>They didn't test one record.<br>- [Derek] As Leeson's<br>hidden losses mounted,<br>he requested huge sums from head office<br>to keep doubling down up<br>to $5 million at a time.<br>Barings management, barely<br>understanding futures<br>and believing in their star trader,<br>they granted his requests<br>and continued to grant future requests.<br>- Every day that I make<br>one of these requests<br>for additional money, I<br>never expect it to come.<br>I expect somebody to say,<br>"Look, mate, what the hell's going on?"<br>- By Autumn 1994,<br>the account was nearly<br>$260 million in the red.<br>To mask this, Leeson began<br>buying futures from himself,<br>inflating his perceived profits further.<br>At SIMEX's 10th anniversary in September,<br>Barings received two awards,<br>largely credited to Leeson.<br>In December, he was flown to New York,<br>seemingly responsible for $44 million<br>in Barings turnover that year.<br>Some traders raised doubts<br>that these kinds of profits could be made<br>from the low risk trading<br>he was supposed to be doing.<br>But management's faith in<br>Leeson's talent was unshakeable,<br>and they dismissed these concerns.<br>Barings' overconfidence in Leeson<br>and Leeson's overconfidence in himself<br>would be the bank's downfall.<br>The figures are so big,<br>it seems like the<br>ultimate confidence trick,<br>but there's a partial explanation<br>for the delusion on both sides.<br>Leeson's and Barings'<br>overconfidence was amplified<br>by the complexity and<br>unpredictability of the market<br>and the trades he was making.<br>It's sort of an issue of feedback.<br>In a controlled environment<br>where there are clear rules<br>and guaranteed outcomes<br>like in a chess match,<br>there is clear feedback<br>on whether decisions are good or bad.<br>With reliable feedback,<br>professional chess players<br>are able to learn to make better decisions<br>with more accurate confidence.<br>But in a noisy environment<br>where there is rarely consistent<br>or timely consequences for predictions,<br>this feedback is unreliable.<br>Whether our confidence<br>is accurate or misplaced<br>is increasingly hard to<br>judge, even amongst experts.<br>It's especially problematic<br>for political pundits.<br>- This is going to be a landslide.<br>I think Romney's gonna win by quite a bit.<br>- So, right now, we have<br>Hillary's about a 75<br>or an 80% favorite.<br>- [Derek] For example, prior to 2024,<br>political analyst Allan Lichtman<br>had accurately predicted<br>the winner of nine<br>of the 10 previous US<br>presidential elections,<br>using his 13 keys to<br>the White House method.<br>Using the same strategy in<br>2024, he predicted that-<br>- Kamala Harris will be a<br>precedent-breaking president.<br>- And look what happened.<br>Is this crazy?<br>(crowd cheering)<br>- He attributed his<br>miscalculation to the spread<br>of disinformation that<br>misled the electorate.<br>This noisy environment made it difficult<br>to discern key issues like the<br>actual state of the economy.<br>Similarly, the feedback Leeson<br>had received was inconsistent.<br>He was making some bad trades,<br>but he had also won it all back before,<br>and this significantly<br>clouded his judgment,<br>amplifying his overconfidence.<br>By 1995, Leeson's losses were<br>in the hundreds of millions,<br>and Barings had unwittingly<br>sent him $1 billion.<br>For a bank with around $700<br>million in capital base,<br>they were legally only allowed to lend<br>around a quarter of that,<br>but no one questioned it.<br>They were all blinded<br>by his apparent success.<br>At this point, his positions were so big,<br>he estimates that he was probably half<br>of the entire Nikkei futures market,<br>so all he could do was<br>buy himself some time<br>and hope that the market went his way.<br>And for a while, it worked.<br>The economy was stable and<br>he couldn't see anything<br>on the horizon that would change this.<br>But then disaster struck.<br>- [Reporter] Japan is tonight<br>in a state of mourning<br>and of shock.<br>(siren wailing)<br>- On the 17th of January, 1995,<br>the Great Hanshin Earthquake struck Japan<br>20 kilometers from the city of Kobe.<br>With the magnitude of 6.9,<br>it devastated the city,<br>which was one of Japan's key ports.<br>This devastation spread<br>to the stock market.<br>The Nikkei index plunged 1055 points.<br>Leeson, attempting to double down again,<br>risked even more money.<br>He bet heavily that the Nikkei<br>would make a rapid recovery,<br>but it didn't.<br>And in the end, in today's money,<br>Leeson had lost $2.8 billion.<br>On the 23rd of February,<br>Leeson went on the run.<br>And three days later, Barings,<br>one of the oldest and most trusted banks<br>in the world, collapsed.<br>Overconfidence was at<br>least partially to blame<br>for its downfall.<br>Realizing the walls were closing in,<br>Leeson fled to Malaysia and then Thailand,<br>but his escape was short-lived.<br>He was eventually arrested in Germany<br>and extradited to face justice,<br>marking the end of a<br>spectacularly destructive gamble.<br>He was just 28 years old.<br>Now, we don't all bring down banks,<br>but we are all vulnerable<br>to overconfidence.<br>In a complex world with<br>unclear, noisy feedback<br>where our brains are overwhelmed,<br>a set of simplistic biases can take over.<br>And we all too often end up thinking<br>we know more than we do.<br>So, what can we actually do about this?<br>- I try to get better at<br>calibrating my confidence judgments<br>by keeping track and keeping score.<br>So, when a colleague asks me,<br>"How long is it going to take you<br>to get me comments on this<br>paper draft we're working on?"<br>I don't promise I'll do it by Friday.<br>I'm much more likely<br>to say something like,<br>"I think there's a 60% chance<br>I can get you comments by Friday."<br>They'll often react with a<br>quizzical look or a laugh.<br>- Well, practicing and being<br>aware of our calibration<br>is the obvious way to improve,<br>but so is being intellectually humble.<br>- I think that the best<br>medicine for overconfidence<br>is not so much information<br>as feedback, (chuckles)<br>and I get plenty of that. (laughs)<br>(Derek laughs)<br>Though also, I think people are right,<br>that sometimes I do have a<br>little bit of overconfidence.<br>- If we wanna become more accurate,<br>we should capitalize on<br>the wisdom of the crowd<br>by listening more to others.<br>In particular, we should listen<br>to people who disagree with us.<br>Understanding the best<br>arguments of your critics,<br>understanding what information<br>those who disagree with<br>you have that you lack<br>is very helpful for<br>making better decisions.<br>- The best calibrated people<br>aren't those who know the most.<br>It's those who know what they don't know.<br>So, true wisdom lies not in being certain,<br>but in knowing the limits<br>of your own certainty.<br>And that's an idea that's<br>inspired our latest project.<br>All those questions I asked-<br>- These are good questions. (laughs)<br>- Aw!<br>(Hannah laughs)<br>- They come from a new<br>board game that we've made.<br>It's called "Elements of Truth."<br>The game contains over 800 fascinating<br>science trivia questions with a twist.<br>The number of points<br>you win on each question<br>depends on how confident you are.<br>You can bid any number from 1 to 10.<br>If you're not sure, you<br>can play a low number<br>or you can try to follow<br>the lead of someone else<br>who you think should know the answer.<br>We've tested this game with scientists,<br>teachers, and students,<br>and what we've found<br>is that it regularly leads to discussions<br>that go way beyond the initial questions.<br>The core game comes with 200 questions<br>that cover all aspects of science,<br>and there are five additional<br>packs on specific topics<br>like physics, technology,<br>engineering, and astronomy.<br>Plus, there's a Veritasium<br>pack on concepts covered<br>in many of our most popular videos.<br>We're launching the<br>game through Kickstarter<br>to give you the opportunity<br>to shape it with us.<br>Over the next month, you'll<br>be able to submit questions<br>for a special community pack.<br>This is your chance to etch your name<br>into Veritasium history<br>with a credited question in the game.<br>To reserve your copy and get involved,<br>use this QR code or the<br>link in the description<br>to head over to Kickstarter.<br>It is only because of<br>you that I've been able<br>to make this channel and this game,<br>so as always, I wanna<br>thank you for your support<br>and thank you for watching. Friday the 17th of July 1992. Amidst the<br>chaos of the trading floor at<br>Singapore's International Stock<br>Exchange, one of the junior traders<br>makes an expensive mistake.<br>Instead of buying 20 futures contracts<br>for a client, she sold them instead,<br>costing Beerings Bank nearly $40,000.<br>To save her job, her boss, Nick Leon, a<br>young trader keen to make his mark,<br>decides to hide the loss. He puts it in<br>an obscure error account. That's account<br>88888<br>It's used by banks to solve small<br>discrepancies in trades. It's a<br>dangerous move and it should have been<br>picked up by the central office<br>immediately. But when nobody from<br>bearings notices, he gains confidence,<br>convinced he can win back the loss and<br>get the team out of trouble.<br>>> I'm operating in the belief that I can<br>get it back.<br>You know, maybe it goes back to the<br>confidence thing. I don't know. I mean,<br>I'm confident I can get it back at that<br>stage. But Leon's staggering<br>overconfidence would create a debt of<br>gargantuan proportions and lead to one<br>of the biggest collapses in banking<br>history.<br>It's easy for me to make the case that<br>overconfidence is the most dangerous of<br>the human biases. Overconfidence gets us<br>into all sorts of trouble.<br>It leads us to take risks, make<br>commitments, enter contests, try things<br>that will ultimately fail, sometimes in<br>costly, embarrassing, and dangerous<br>ways. Overconfidence has been implicated<br>in almost every big disaster, from the<br>sinking of the Titanic to the Chernobyl<br>nuclear disaster to the loss of the<br>space shuttle Challenger. But<br>overconfidence isn't reserved for just a<br>few reckless individuals. We can all<br>fall victim to it. For example, 93% of<br>us think we are better drivers than the<br>median, which is of course impossible.<br>The scale of the problem was identified<br>in some now classic research with a set<br>of simple quiz questions. True or false?<br>Australia is wider than the moon. Oh, I<br>bet they're about the same size.<br>>> It's just it looks small on a map, but<br>it's huge.<br>>> Are there more or less stars in the<br>Milky Way than trees on Earth?<br>>> More stars. more stars.<br>>> One teaspoon of pure olive oil,<br>>> 4 1/2 g, contains more chemical<br>potential energy than an equivalent 4<br>1/2 g amount of TNT.<br>>> I'm going to say true.<br>>> Oh no,<br>>> this feels kind of like a trick<br>question.<br>>> The energy density of butter is 10 times<br>that of lithium batteries, right? I<br>mean, lithium batteries are rubbish.<br>Everyone's going on about them. Just run<br>your car on butter, I say.<br>>> True or false? Plants are more efficient<br>than the average solar panel at<br>absorbing the sun's energy.<br>>> Yeah, they must be.<br>>> Now, these aren't the exact questions<br>from the original research. More on<br>those later, but the specifics don't<br>really matter. The interesting bit is<br>what they asked next. How confident are<br>you that you are right?<br>>> 100% confident.<br>>> 100% confident.<br>>> 100. Yes.<br>>> I think it's true. Um, so I've also got<br>to go 90%.<br>>> 100%<br>>> 100%<br>>> 100%.<br>>> Yeah.<br>>> I don't think we actually had any<br>expectations. This was very early on the<br>re in in the research on on confidence<br>assessment, but we didn't really know<br>what we what what we would find.<br>>> The results revealed a surprising<br>disparity between how accurate someone<br>thinks they are and how accurate they<br>actually are. When people are 90%<br>certain, they are right only 75% of the<br>time. In fact, we ran our own version of<br>this online and found an even more<br>extreme discrepancy. We asked the<br>Veritassin community a bunch of science<br>questions and then asked how confident<br>they were in their answer. Our results<br>showed that those who were the most<br>confident, describing themselves as 91<br>to 100% sure, were only correct 51% of<br>the time. Since the original research,<br>these results have been replicated again<br>and again in all areas from general<br>knowledge to motor skills and it affects<br>everyone even experts. I have a paper<br>where we analyze the survey of<br>professional forecasters. These are<br>chief economists at various corporations<br>and banks who are invited to forecast<br>the state of the economy on a quarterly<br>basis. We find that they are on average<br>too sure that they know what's going to<br>happen. So they on average say they're<br>something like 53% sure they have<br>correctly predicted what's going to<br>happen to inflation for instance and<br>they are right like 23% of the time. How<br>well what you think you know matches<br>with what you actually know is called<br>your calibration. So if you're perfectly<br>calibrated and say 80% confident you<br>should be right 80% of the time. But<br>most of us are not well calibrated<br>>> 100%<br>>> 100%.<br>>> Yeah. There are 200 billion stars in the<br>Milky Way, but there are three trillion<br>trees on Earth.<br>>> 100% confident.<br>>> 100. Yes.<br>>> False.<br>>> Most of the time, being overconfident<br>isn't a huge issue. But for Nick Leon,<br>the stakes soon became very high. Leon's<br>plan to recover his team's loss was to<br>bet that the Japanese stock market would<br>go up. So he went long on the top 225<br>companies in Japan, the Nikkeay 225.<br>Ever since the Japanese asset price<br>bubble peaked 2 years previously, the<br>Nikkeay had been dropping from 38,000 to<br>16,000. And he was confident the market<br>would soon bottom out and start rising,<br>but it continued to fall. Over the next<br>few weeks, it dropped to a low of<br>14,000. And the whole time, Nick was<br>betting the wrong way with ever bigger<br>and riskier bets. He would double down,<br>betting double his losses so the next<br>win would bring him back above water.<br>This strategy should work when the next<br>win comes along, but the market collapse<br>was unrelenting and his losses ballooned<br>from $40,000 to around $3 million. In<br>spite of this, Leon remained confident<br>and eventually his luck turned around.<br>In the spring of 1993, the market<br>rebounded up to 20,000, and by the<br>summer, that account was back in credit.<br>Leon goes out to celebrate that weekend,<br>drinking and dancing on tables, finally<br>free of the hole he had dug himself<br>into.<br>But on Monday morning, he made another<br>trading error on the futures market. A<br>damaging loss that Leon didn't want to<br>admit. So he put the loss back into<br>account 88888.<br>Confident that if he got out of the<br>previous one, he could do it again. He<br>began making risky trades to try to<br>recover that initial loss. And as the<br>losses mounted in the 58 account, it<br>became harder and harder to make new<br>trades to cover the last loss. By the<br>end of 1993, his losses exceeded $30<br>million.<br>Leon is obviously an extreme example,<br>but this tendency to overestimate our<br>abilities is something we all share. So<br>the question is why? The obvious<br>explanation that a lot of people default<br>to that is we want to feel good about<br>ourselves and pretend like we're<br>wellinformed that we enjoy enormous<br>satisfaction from being able to say I<br>know or even better I told you so. And<br>so<br>uh we pretend to ourselves and others<br>that we know when we don't. That is the<br>the motivated egocentric version of an<br>explanation for excessive faith in her<br>own judgment.<br>>> Reichel had faith and the idea if<br>nothing else.<br>>> But what if it's something more<br>uncomfortable? Stupidity.<br>France. Reichelt was a tailor with no<br>formal scientific or engineering<br>training. But he was obsessed with<br>solving the problem of aviation safety.<br>He spent months designing a parachute<br>suit which repeatedly failed during<br>testing. But he became convinced the<br>issue was that he didn't have enough<br>height for it to work. So in February<br>1912, he took it to the Eiffel Tower.<br>And rather than using a dummy to test<br>it, he jumped himself.<br>Tragic, but also really dumb.<br>>> Do you recognize this graph?<br>>> Uh the like Dunning Krueger graph.<br>>> Are you confident that this is the<br>actual Dunning Krueger graph?<br>>> Not at all. I in fact, I think that this<br>is not the Dunning Krueger graph. I<br>think that this is a bit of a<br>communication trick. This is what people<br>call the Mount Stupid curve, which will<br>pop up if you search for the Dunning<br>Krueger effect into Google images.<br>>> As far as I understand it, um it's a<br>it's an effect that says that when you<br>know very little about it, your<br>confidence is high. And as you know more<br>about it, your confidence sinks until<br>you become a master at it.<br>>> But I think that the actual result was<br>like a little bit more fuzzy than this.<br>It was this graph doesn't actually show<br>the Dunning Krueger effect. It's just a<br>meme that resonates with a lot of people<br>and it became conflated with the effect.<br>In their original research, Dunning and<br>Krueger tested people on tasks like<br>grammar, logic, and humor and then they<br>asked participants to estimate how well<br>they performed. That produced this<br>curve. You can see that those who<br>performed worse had the largest mismatch<br>between their confidence and<br>performance. They were the most<br>overconfident. Those who performed the<br>best were actually slightly<br>underconfident. This suggests that<br>overconfidence may be linked to how much<br>we know. Those who know less think that<br>they know more than they do. But to me,<br>this effect looks a little like a<br>statistical artifact. Because while I<br>was doing my PhD, I asked students some<br>physics questions and I also asked for<br>their confidence in their answers. And<br>when I looked at the results, accuracy<br>spanned the full range from nearly 0% to<br>100%. But confidence varied over a much<br>narrower range. All the scores were<br>roughly in the middle. This meant that<br>poor performers were the most<br>overconfident and the highest performers<br>were actually slightly underconfident.<br>This is exactly what Dunning and Krueger<br>found. So maybe overconfidence isn't<br>entirely due to how much we know, but<br>also because most of us express at least<br>kind of middle-of the road confidence.<br>But there is another factor at play<br>here, which is how much information our<br>brains are capable of processing.<br>Monday, the 27th of January, 1986. It's<br>5:45 p.m. at the Kennedy Space Center in<br>Cape Canaveral, and the engineers from<br>Morton Theacle, who made the Challenger<br>Space Shuttle's rocket boosters, make an<br>emergency conference call. They've just<br>seen the weather forecast, predicting<br>temperatures overnight will plummet to<br>25° F, far colder than any previous<br>shuttle launch. They know the rubber<br>O-rings that seal joints in the boosters<br>become less flexible in the cold, but<br>have just a few hours to gather data,<br>create charts, and present their case.<br>Over the next 6 hours, the engineers<br>present 13 charts with data on O-ring<br>temperature, hot gas erosion, joint<br>rotation, and more. But the data is<br>scattered, incomplete, and not<br>synthesized into a clear narrative.<br>They'd never tested below 53° F. So NASA<br>managers were trying simultaneously to<br>track historical O-ring data, erosion<br>patterns, joint dynamics, seal<br>resiliency, pressure differentials, and<br>more. No single chart told the whole<br>story. So overwhelmed by seemingly<br>contradictory data and confident that<br>their rocket boosters were safe, theole<br>management overruled their own engineers<br>and approved the launch.<br>At 11:38 the next morning, 73 seconds<br>after launch,<br>all seven crew members were killed.<br>Calibrating your certainty requires<br>thinking of all the ways that you could<br>be wrong. And that's hard for finite<br>fallible agents like us. If it means<br>considering everything that we don't<br>know,<br>>> how many chunks of novel information you<br>can hold in your head at one time is<br>your short-term memory capacity. In<br>2008, Hansen, Jesus, and Winman<br>investigated how short-term memory<br>capacity was linked to accuracy and<br>overconfidence. Participants were asked<br>to give ranges for factual questions<br>like the length of a river or the<br>population of a city. People's ranges<br>were consistently too narrow, which<br>effectively meant they were being<br>overconfident. And those who had worse<br>short-term memory were more often wrong<br>and more likely to be overconfident.<br>Another study conducted by Conti in 2023<br>asked participants to keep sequences of<br>letters in their mind while they judged<br>their own performance. As the memory<br>load increased, confidence estimates<br>became less accurate, even for<br>participants with higher working memory<br>capacities. Together, these studies<br>suggest that assessing your accuracy is<br>a mentally taxing task. So<br>overconfidence isn't necessarily about<br>arrogance. It's your brain working at<br>the limits of what it can track and<br>hold. And because of this, your brain<br>can start using shortcuts.<br>Psychologist Daniel Conorman describes<br>these mental shortcuts as heruristics<br>that together create systemic errors<br>called cognitive biases. One shortcut we<br>use a lot is substituting hard questions<br>with easier related ones. Researchers<br>have tested this by asking students how<br>happy they were in their life and how<br>many dates they'd had in the last month.<br>Unsurprisingly, these two questions did<br>not correlate. There's a lot more to<br>happiness than matches on Tinder, at<br>least for most people. However, if you<br>switch it around and ask about their<br>dates first, suddenly the correlation<br>jumps to 66. Working out how happy you<br>are in your life is a difficult<br>question. You have to consider many<br>things and balance them all out. So,<br>when you're primed with the information<br>about your dating life, you're very<br>likely to substitute that hard question<br>like, "How happy am I in my life?" with,<br>"How many dates have I recently had?"<br>Overconfidence from misprocessing<br>information like this can be disastrous.<br>So, why are our brains wired this way? I<br>mean, you might expect natural selection<br>to have wiped out the confidently<br>incorrect, but there is evidence that<br>overconfidence can actually be<br>advantageous.<br>Overconfidence<br>can massively improve your status. In a<br>scientific version of the apprentice,<br>scientists in 2012 compared participants<br>own assessments of their skill with<br>objective measures. They placed them in<br>group tasks to see who was chosen as a<br>leader and whose ideas influenced the<br>group, tracking status over multiple<br>sessions and assessing whether the<br>desire for status led participants to<br>exaggerate their abilities. The results<br>were clear. Overconfident individuals<br>were more likely to lead, assert<br>themselves, and maintain influence even<br>when their actual abilities were mid.<br>And the evidence certainly shows people<br>react better to confident individuals.<br>Using an fMRI, researchers at the<br>University of Sussex measured brain<br>activity of people after hearing<br>unconfident versus confident advice.<br>Those who listened to the confident<br>advice had increased activity in the<br>ventromedial prefrontal cortex. That's a<br>brain region associated with processing<br>rewards and expected satisfaction. This<br>means that human brains are biologically<br>tuned to be influenced by confident<br>individuals. We literally feel better<br>when we hear confident people. But<br>recognizing that dynamic<br>highlights a potentially problematic<br>incentive for anyone who's contending<br>for positions that they want. People who<br>express more confidence in employment<br>interviews or political campaigns, for<br>instance, do earn the confidence of<br>interviewers and potential voters even<br>if they're being overconfident. They<br>can't actually deliver. They don't<br>actually know the answer, but they can<br>talk a good line to impress others even<br>if they can't ultimately deliver on<br>those grand assertions. They know the<br>audience will place more faith in them<br>if they express conf. Well, they should<br>express maximal confidence.<br>>> Leon knew this and he took advantage of<br>it. While he was secretly racking up<br>huge losses in the infamous 58 account,<br>he was publicly posting huge profits and<br>everyone believed the illusion.<br>>> They come in and they don't test any<br>records.<br>>> So, I can't be happier. They didn't test<br>one record. As Leon's hidden losses<br>mounted, he requested huge sums from<br>head office to keep doubling down, up to<br>$5 million at a time. Bearing<br>management, barely understanding futures<br>and believing in their star trader, they<br>granted his requests and continued to<br>grant future requests. Every day that I<br>make one of these requests for<br>additional money, I never expected to<br>come. I expect somebody to say, "Look,<br>what the hell's going on?"<br>>> By autumn 1994, the account was nearly<br>$260 million in the red. To mask this,<br>Leon began buying futures from himself,<br>inflating his perceived profits further.<br>At Syax's 10th anniversary in September,<br>Beerings received two awards, largely<br>credited to Leon. In December, he was<br>flown to New York, seemingly responsible<br>for $44 million in Bearings turnover<br>that year. Some traders raised doubts<br>that these kinds of profits could be<br>made from the low-risisk trading he was<br>supposed to be doing. But management's<br>faith in Leon's talent was unshakable,<br>and they dismissed these concerns.<br>Beering's overconfidence in Leon and<br>Leon's overconfidence in himself would<br>be the bank's downfall. The figures are<br>so big it seems like the ultimate<br>confidence trick. But there's a partial<br>explanation for the delusion on both<br>sides. Leon's and bearings<br>overconfidence was amplified by the<br>complexity and unpredictability of the<br>market and the trades he was making.<br>It's sort of an issue of feedback.<br>In a controlled environment where there<br>are clear rules and guaranteed outcomes,<br>like in a chess match, there is clear<br>feedback on whether decisions are good<br>or bad. With reliable feedback,<br>professional chess players are able to<br>learn to make better decisions with more<br>accurate confidence. But in a noisy<br>environment where there is rarely<br>consistent or timely consequences for<br>predictions, this feedback is<br>unreliable. Whether our confidence is<br>accurate or misplaced is increasingly<br>hard to judge even amongst experts. It's<br>especially problematic for political<br>pundits.<br>>> This is going to be a landslide. I think<br>Romney is going to win by quite a bit.<br>>> So, right now, we have Hillary's about a<br>75 or an 80% favorite.<br>>> For example, prior to 2024, political<br>analyst Alan Lickman had accurately<br>predicted the winner of nine of the 10<br>previous US presidential elections using<br>his 13 keys to the White House method.<br>Using the same strategy in 2024, he<br>predicted that<br>>> Kla Harris will be a president-breaking<br>president.<br>>> Look what happened. Is this crazy?<br>>> He attributed his miscalculation to the<br>spread of disinformation that misled the<br>electorate. This noisy environment made<br>it difficult to discern key issues like<br>the actual state of the economy.<br>Similarly, the feedback Leon had<br>received was inconsistent. He was making<br>some bad trades, but he had also won it<br>all back before. and this significantly<br>clouded his judgment, amplifying his<br>overconfidence. By 1995, Leon's losses<br>were in the hundreds of millions, and<br>Bearings had unwittingly sent him $1<br>billion. For a bank with around $700<br>million in capital base, they were<br>legally only allowed to lend around a<br>quarter of that. But no one questioned<br>it. They were all blinded by his<br>apparent success. At this point, his<br>positions were so big, he estimates that<br>he was probably half of the entire Nikay<br>futures market. So all he could do was<br>buy himself some time and hope that the<br>market went his way. And for a while it<br>worked. The economy was stable and he<br>couldn't see anything on the horizon<br>that would change this. But then<br>disaster struck. Japan is tonight in a<br>state of mourning and of shock.<br>>> On the 17th of January 1995, the great<br>Henshin earthquake struck Japan. 20 km<br>from the city of Coobe. With a magnitude<br>of 6.9, it devastated the city which was<br>one of Japan's key ports. This<br>devastation spread to the stock market.<br>The NIK index plunged 1,055 points. Leon<br>attempting to double down again risked<br>even more money. He bet heavily that the<br>Nikay would make a rapid recovery, but<br>it didn't. And in the end, in today's<br>money, Leon had lost $2.8 billion. On<br>the 23rd of February, Leon went on the<br>run. And 3 days later, Bearings, one of<br>the oldest and most trusted banks in the<br>world, collapsed. overconfidence was at<br>least partially to blame for its<br>downfall. Realizing the walls were<br>closing in, Leon fled to Malaysia and<br>then Thailand, but his escape was<br>short-lived. He was eventually arrested<br>in Germany and extradited to face<br>justice, marking the end of a<br>spectacularly destructive gamble. He was<br>just 28 years old.<br>Now, we don't all bring down banks, but<br>we are all vulnerable to overconfidence.<br>In a complex world with unclear, noisy<br>feedback where our brains are<br>overwhelmed, a set of simplistic biases<br>can take over and we all too often end<br>up thinking we know more than we do. So,<br>what can we actually do about this? I<br>try to get better at calibrating my<br>confidence judgments by keeping track<br>and keeping score. So, uh, when a<br>colleague asks me, "How long is it going<br>to take you to get me comments on this<br>paper draft we're working on?" I don't<br>promise I'll do it by Friday. I'm much<br>more likely to say something like, "I<br>think there's a 60% chance I can get you<br>comments by Friday." They'll often uh<br>react with a quizzical look or uh a<br>laugh. Well, practicing and being aware<br>of our calibration is the obvious way to<br>improve, but so is being intellectually<br>humble.<br>>> I think that uh the best medicine for<br>overconfidence is not so much<br>information as feedback. Um and I get<br>plenty of that<br>though. I also I I think people are<br>right that sometimes I do have a little<br>bit of overconfidence.<br>>> If we want to become more accurate, we<br>should capitalize on the wisdom of the<br>crowd by listening more to others. In<br>particular, we should listen to people<br>who disagree with us. Understanding the<br>best arguments of your critics,<br>understanding what information those who<br>disagree with you have that you lack is<br>very helpful for making better<br>decisions. The best calibrated people<br>aren't those who know the most. It's<br>those who know what they don't know. So,<br>true wisdom lies not in being certain,<br>but in knowing the limits of your own<br>certainty. And that's an idea that's<br>inspired our latest project. All those<br>questions I asked. These are good<br>questions. A<br>>> well,<br>>> they come from a new board game that<br>we've made. It's called Elements of<br>Truth. The game contains over 800<br>fascinating science trivia questions<br>with a twist. The number of points you<br>win on each question depends on how<br>confident you are. You can bid any<br>number from 1 to 10. If you're not sure,<br>you can play a low number, or you can<br>try to follow the lead of someone else<br>who you think should know the answer.<br>We've tested this game with scientists,<br>teachers, and students. And what we<br>found is that it regularly leads to<br>discussions that go way beyond the<br>initial questions. The core game comes<br>with 200 questions that cover all<br>aspects of science, and there are five<br>additional packs on specific topics like<br>physics, technology, engineering, and<br>astronomy. Plus, there's a Veritassium<br>pack on concepts covered in many of our<br>most popular videos.<br>We're launching the game through<br>Kickstarter to give you the opportunity<br>to shape it with us. Over the next<br>month, you'll be able to submit<br>questions for a special community pack.<br>This is your chance to etch your name<br>into Veritasium History with a credited<br>question in the game. To reserve your<br>copy and get involved, use this QR code<br>or the link in the description to head<br>over to Kickstarter. It is only because<br>of you that I've been able to make this<br>channel and this game. So, as always, I<br>want to thank you for your support and<br>thank you for watching.