Tag: Daniel Kahneman

Complexity Bias: Why We Prefer Complicated to Simple

Complexity bias is a logical fallacy that leads us to give undue credence to complex concepts.

Faced with two competing hypotheses, we are likely to choose the most complex one. That’s usually the option with the most assumptions and regressions. As a result, when we need to solve a problem, we may ignore simple solutions — thinking “that will never work” — and instead favor complex ones.

To understand complexity bias, we need first to establish the meaning of three key terms associated with it: complexity, simplicity, and chaos.

Complexity, like pornography, is hard to define when we’re put on the spot, although most of us recognize it when we see it. The Cambridge Dictionary defines complexity as “the state of having many parts and being difficult to understand or find an answer to.” The definition of simplicity is the inverse: “something [that] is easy to understand or do.” Chaos is defined as “a state of total confusion with no order.”

“Life is really simple, but we insist on making it complicated.”

— Confucius

Complex systems contain individual parts that combine to form a collective that often can’t be predicted from its components. Consider humans. We are complex systems. We’re made of about 100 trillion cells and yet we are so much more than the aggregation of our cells. You’d never predict what we’re like or who we are from looking at our cells.

Complexity bias is our tendency to look at something that is easy to understand, or look at it when we are in a state of confusion, and view it as having many parts that are difficult to understand.

We often find it easier to face a complex problem than a simple one.

A person who feels tired all the time might insist that their doctor check their iron levels while ignoring the fact that they are unambiguously sleep deprived. Someone experiencing financial difficulties may stress over the technicalities of their telephone bill while ignoring the large sums of money they spend on cocktails.

Marketers make frequent use of complexity bias.

They do this by incorporating confusing language or insignificant details into product packaging or sales copy. Most people who buy “ammonia-free” hair dye, or a face cream which “contains peptides,” don’t fully understand the claims. Terms like these often mean very little, but we see them and imagine that they signify a product that’s superior to alternatives.

How many of you know what probiotics really are and how they interact with gut flora?

Meanwhile, we may also see complexity where only chaos exists. This tendency manifests in many forms, such as conspiracy theories, superstition, folklore, and logical fallacies. The distinction between complexity and chaos is not a semantic one. When we imagine that something chaotic is in fact complex, we are seeing it as having an order and more predictability than is warranted. In fact, there is no real order, and prediction is incredibly difficult at best.

Complexity bias is interesting because the majority of cognitive biases occur in order to save mental energy. For example, confirmation bias enables us to avoid the effort associated with updating our beliefs. We stick to our existing opinions and ignore information that contradicts them. Availability bias is a means of avoiding the effort of considering everything we know about a topic. It may seem like the opposite is true, but complexity bias is, in fact, another cognitive shortcut. By opting for impenetrable solutions, we sidestep the need to understand. Of the fight-or-flight responses, complexity bias is the flight response. It is a means of turning away from a problem or concept and labeling it as too confusing. If you think something is harder than it is, you surrender your responsibility to understand it.

“Most geniuses—especially those who lead others—prosper not by deconstructing intricate complexities but by exploiting unrecognized simplicities.”

— Andy Benoit

Faced with too much information on a particular topic or task, we see it as more complex than it is. Often, understanding the fundamentals will get us most of the way there. Software developers often find that 90% of the code for a project takes about half the allocated time. The remaining 10% takes the other half. Writing — and any other sort of creative work — is much the same. When we succumb to complexity bias, we are focusing too hard on the tricky 10% and ignoring the easy 90%.

Research has revealed our inherent bias towards complexity.

In a 1989 paper entitled “Sensible reasoning in two tasks: Rule discovery and hypothesis evaluation,” Hilary F. Farris and Russell Revlin evaluated the topic. In one study, participants were asked to establish an arithmetic rule. They received a set of three numbers (such as 2, 4, 6) and tried to generate a hypothesis by asking the experimenter if other number sequences conformed to the rule. Farris and Revlin wrote, “This task is analogous to one faced by scientists, with the seed triple functioning as an initiating observation, and the act of generating the triple is equivalent to performing an experiment.”

The actual rule was simple: list any three ascending numbers.

The participants could have said anything from “1, 2, 3” to “3, 7, 99” and been correct. It should have been easy for the participants to guess this, but most of them didn’t. Instead, they came up with complex rules for the sequences. (Also see Falsification of Your Best Loved Ideas.)

A paper by Helena Matute looked at how intermittent reinforcement leads people to see complexity in chaos. Three groups of participants were placed in rooms and told that a loud noise would play from time to time. The volume, length, and pattern of the sound were identical for each group. Group 1 (Control) was told to sit and listen to the noises. Group 2 (Escape) was told that there was a specific action they could take to stop the noises. Group 3 (Yoked) was told the same as Group 2, but in their case, there was actually nothing they could do.

Matute wrote:

Yoked participants received the same pattern and duration of tones that had been produced by their counterparts in the Escape group. The amount of noise received by Yoked and Control subjects depends only on the ability of the Escape subjects to terminate the tones. The critical factor is that Yoked subjects do not have control over reinforcement (noise termination) whereas Escape subjects do, and Control subjects are presumably not affected by this variable.

The result? Not one member of the Yoked group realized that they had no control over the sounds. Many members came to repeat particular patterns of “superstitious” behavior. Indeed, the Yoked and Escape groups had very similar perceptions of task controllability. Faced with randomness, the participants saw complexity.

Does that mean the participants were stupid? Not at all. We all exhibit the same superstitious behavior when we believe we can influence chaotic or simple systems.

Funnily enough, animal studies have revealed much the same. In particular, consider B.F. Skinner’s well-known research on the effects of random rewards on pigeons. Skinner placed hungry pigeons in cages equipped with a random-food-delivery mechanism. Over time, the pigeons came to believe that their behavior affected the food delivery. Skinner described this as a form of superstition. One bird spun in counterclockwise circles. Another butted its head against a corner of the cage. Other birds swung or bobbed their heads in specific ways. Although there is some debate as to whether “superstition” is an appropriate term to apply to birds, Skinner’s research shed light on the human tendency to see things as being more complex than they actually are.

Skinner wrote (in “‘Superstition’ in the Pigeon,” Journal of Experimental Psychology, 38):

The bird behaves as if there were a causal relation between its behavior and the presentation of food, although such a relation is lacking. There are many analogies in human behavior. Rituals for changing one's fortune at cards are good examples. A few accidental connections between a ritual and favorable consequences suffice to set up and maintain the behavior in spite of many unreinforced instances. The bowler who has released a ball down the alley but continues to behave as if he were controlling it by twisting and turning his arm and shoulder is another case in point. These behaviors have, of course, no real effect upon one's luck or upon a ball half way down an alley, just as in the present case the food would appear as often if the pigeon did nothing—or, more strictly speaking, did something else.

The world around us is a chaotic, entropic place. But it is rare for us to see it that way.

In Living with Complexity, Donald A. Norman offers a perspective on why we need complexity:

We seek rich, satisfying lives, and richness goes along with complexity. Our favorite songs, stories, games, and books are rich, satisfying, and complex. We need complexity even while we crave simplicity… Some complexity is desirable. When things are too simple, they are also viewed as dull and uneventful. Psychologists have demonstrated that people prefer a middle level of complexity: too simple and we are bored, too complex and we are confused. Moreover, the ideal level of complexity is a moving target, because the more expert we become at any subject, the more complexity we prefer. This holds true whether the subject is music or art, detective stories or historical novels, hobbies or movies.

As an example, Norman asks readers to contemplate the complexity we attach to tea and coffee. Most people in most cultures drink tea or coffee each day. Both are simple beverages, made from water and coffee beans or tea leaves. Yet we choose to attach complex rituals to them. Even those of us who would not consider ourselves to be connoisseurs have preferences. Offer to make coffee for a room full of people, and we can be sure that each person will want it made in a different way.

Coffee and tea start off as simple beans or leaves, which must be dried or roasted, ground and infused with water to produce the end result. In principle, it should be easy to make a cup of coffee or tea. Simply let the ground beans or tea leaves [steep] in hot water for a while, then separate the grounds and tea leaves from the brew and drink. But to the coffee or tea connoisseur, the quest for the perfect taste is long-standing. What beans? What tea leaves? What temperature water and for how long? And what is the proper ratio of water to leaves or coffee?

The quest for the perfect coffee or tea maker has been around as long as the drinks themselves. Tea ceremonies are particularly complex, sometimes requiring years of study to master the intricacies. For both tea and coffee, there has been a continuing battle between those who seek convenience and those who seek perfection.

Complexity, in this way, can enhance our enjoyment of a cup of tea or coffee. It’s one thing to throw some instant coffee in hot water. It’s different to select the perfect beans, grind them ourselves, calculate how much water is required, and use a fancy device. The question of whether this ritual makes the coffee taste better or not is irrelevant. The point is the elaborate surrounding ritual. Once again, we see complexity as superior.

“Simplicity is a great virtue but it requires hard work to achieve it and education to appreciate it. And to make matters worse: complexity sells better.”

— Edsger W. Dijkstra

The Problem with Complexity

Imagine a person who sits down one day and plans an elaborate morning routine. Motivated by the routines of famous writers they have read about, they lay out their ideal morning. They decide they will wake up at 5 a.m., meditate for 15 minutes, drink a liter of lemon water while writing in a journal, read 50 pages, and then prepare coffee before planning the rest of their day.

The next day, they launch into this complex routine. They try to keep at it for a while. Maybe they succeed at first, but entropy soon sets in and the routine gets derailed. Sometimes they wake up late and do not have time to read. Their perceived ideal routine has many different moving parts. Their actual behavior ends up being different each day, depending on random factors.

Now imagine that this person is actually a famous writer. A film crew asks to follow them around on a “typical day.” On the day of filming, they get up at 7 a.m., write some ideas, make coffee, cook eggs, read a few news articles, and so on. This is not really a routine; it is just a chaotic morning based on reactive behavior. When the film is posted online, people look at the morning and imagine they are seeing a well-planned routine rather than the randomness of life.

This hypothetical scenario illustrates the issue with complexity: it is unsustainable without effort.

The more individual constituent parts a system has, the greater the chance of its breaking down. Charlie Munger once said that “Where you have complexity, by nature you can have fraud and mistakes.” Any complex system — be it a morning routine, a business, or a military campaign — is difficult to manage. Addressing one of the constituent parts inevitably affects another (see the Butterfly Effect). Unintended and unexpected consequences are likely to occur.

As Daniel Kahneman and Amos Tversky wrote in 1974 (in Judgment Under Uncertainty: Heuristics and Biases): “A complex system, such as a nuclear reactor or the human body, will malfunction if any of its essential components fails. Even when the likelihood of failure in each component is slight, the probability of an overall failure can be high if many components are involved.”

This is why complexity is less common than we think. It is unsustainable without constant maintenance, self-organization, or adaptation. Chaos tends to disguise itself as complexity.

“Human beings are pattern-seeking animals. It's part of our DNA. That's why conspiracy theories and gods are so popular: we always look for the wider, bigger explanations for things.”

— Adrian McKinty, The Cold Cold Ground

Complexity Bias and Conspiracy Theories

A musician walks barefoot across a zebra-crossing on an album cover. People decide he died in a car crash and was replaced by a lookalike. A politician’s eyes look a bit odd in a blurry photograph. People conclude that he is a blood-sucking reptilian alien taking on a human form. A photograph shows an indistinct shape beneath the water of a Scottish lake. The area floods with tourists hoping to glimpse a surviving prehistoric creature. A new technology overwhelms people. So, they deduce that it is the product of a government mind-control program.

Conspiracy theories are the ultimate symptom of our desire to find complexity in the world. We don’t want to acknowledge that the world is entropic. Disasters happen and chaos is our natural state. The idea that hidden forces animate our lives is an appealing one. It seems rational. But as we know, we are all much less rational and logical than we think. Studies have shown that a high percentage of people believe in some sort of conspiracy. It’s not a fringe concept. According to research by Joseph E. Uscinski and Joseph M. Parent, about one-third of Americans believe the notion that Barack Obama’s birth certificate is fake. Similar numbers are convinced that 9/11 was an inside job orchestrated by George Bush. Beliefs such as these are present in all types of people, regardless of class, age, gender, race, socioeconomic status, occupation, or education level.

Conspiracy theories are invariably far more complex than reality. Although education does reduce the chances of someone’s believing in conspiracy theories, one in five Americans with postgraduate degrees still hold conspiratorial beliefs.

Uscinski and Parent found that, just as uncertainty led Skinner’s pigeons to see complexity where only randomness existed, a sense of losing control over the world around us increases the likelihood of our believing in conspiracy theories. Faced with natural disasters and political or economic instability, we are more likely to concoct elaborate explanations. In the face of horrific but chaotic events such as Hurricane Katrina, or the recent Grenfell Tower fire, many people decide that secret institutions are to blame.

Take the example of the “Paul McCartney is dead” conspiracy theory. Since the 1960s, a substantial number of people have believed that McCartney died in a car crash and was replaced by a lookalike, usually said to be a Scottish man named William Campbell. Of course, conspiracy theorists declare, The Beatles wanted their most loyal fans to know this, so they hid clues in songs and on album covers.

The beliefs surrounding the Abbey Road album are particularly illustrative of the desire to spot complexity in randomness and chaos. A police car is parked in the background — an homage to the officers who helped cover up the crash. A car’s license plate reads “LMW 28IF” — naturally, a reference to McCartney being 28 if he had lived (although he was 27) and to Linda McCartney (whom he had not met yet). Matters were further complicated once The Beatles heard about the theory and began to intentionally plant “clues” in their music. The song “I’m So Tired” does in fact feature backwards mumbling about McCartney’s supposed death. The 1960s were certainly a turbulent time, so is it any wonder that scores of people pored over album art or played records backwards, looking for evidence of a complex hidden conspiracy?

As Henry Louis Gates Jr. wrote, “Conspiracy theories are an irresistible labor-saving device in the face of complexity.”

Complexity Bias and Language

We have all, at some point, had a conversation with someone who speaks like philosopher Theodor Adorno wrote: using incessant jargon and technical terms even when simpler synonyms exist and would be perfectly appropriate. We have all heard people say things which we do not understand, but which we do not question for fear of sounding stupid.

Jargon is an example of how complexity bias affects our communication and language usage. When we use jargon, especially out of context, we are putting up unnecessary semantic barriers that reduce the chances of someone’s challenging or refuting us.

In an article for The Guardian, James Gingell describes his work translating scientific jargon into plain, understandable English:

It’s quite simple really. The first step is getting rid of the technical language. Whenever I start work on refining a rough-hewn chunk of raw science into something more pleasant I use David Dobbs’ (rather violent) aphorism as a guiding principle: “Hunt down jargon like a mercenary possessed, and kill it.” I eviscerate acronyms and euthanise decrepit Latin and Greek. I expunge the esoteric. I trim and clip and pare and hack and burn until only the barest, most easily understood elements remain.


Jargon…can be useful for people as a shortcut to communicating complex concepts. But it’s intrinsically limited: it only works when all parties involved know the code. That may be an obvious point but it’s worth emphasising — to communicate an idea to a broad, non-specialist audience, it doesn’t matter how good you are at embroidering your prose with evocative imagery and clever analogies, the jargon simply must go.”

Gingell writes that even the most intelligent scientists struggle to differentiate between thinking (and speaking and writing) like a scientist, and thinking like a person with minimal scientific knowledge.

Unnecessarily complex language is not just annoying. It's outright harmful. The use of jargon in areas such as politics and economics does real harm. People without the requisite knowledge to understand it feel alienated and removed from important conversations. It leads people to believe that they are not intelligent enough to understand politics, or not educated enough to comprehend economics. When a politician talks of fiscal charters or rolling four-quarter growth measurements in a public statement, they are sending a crystal clear message to large numbers of people whose lives will be shaped by their decisions: this is not about you.

Complexity bias is a serious issue in politics. For those in the public eye, complex language can be a means of minimizing the criticism of their actions. After all, it is hard to dispute something you don't really understand. Gingell considers jargon to be a threat to democracy:

If we can’t fully comprehend the decisions that are made for us and about us by the government then how we can we possibly revolt or react in an effective way? Yes, we have a responsibility to educate ourselves more on the big issues, but I also think it’s important that politicians and journalists meet us halfway.


Economics and economic decisions are more important than ever now, too. So we should implore our journalists and politicians to write and speak to us plainly. Our democracy depends on it.

In his essay “Politics and the English Language,” George Orwell wrote:

In our time, political speech and writing are largely the defence of the indefensible. … Thus, political language has to consist largely of euphemism, question-begging and sheer cloudy vagueness. Defenceless villages are bombarded from the air, the inhabitants driven out into the countryside, the cattle machine-gunned, the huts set on fire with incendiary bullets: this is called pacification. Millions of peasants are robbed of their farms and sent trudging along the roads with no more than they can carry: this is called transfer of population or rectification of frontiers. People are imprisoned for years without trial, or shot in the back of the neck or sent to die of scurvy in Arctic lumber camps: this is called elimination of unreliable elements.

An example of the problems with jargon is the Sokal affair. In 1996, Alan Sokal (a physics professor) submitted a fabricated scientific paper entitled “Transgressing the Boundaries: Towards a Transformative Hermeneutics of Quantum Gravity.” The paper had absolutely no relation to reality and argued that quantum gravity is a social and linguistic construct. Even so, the paper was published in a respected journal. Sokal’s paper consisted of convoluted, essentially meaningless claims, such as this paragraph:

Secondly, the postmodern sciences deconstruct and transcend the Cartesian metaphysical distinctions between humankind and Nature, observer and observed, Subject and Object. Already quantum mechanics, earlier in this century, shattered the ingenious Newtonian faith in an objective, pre-linguistic world of material objects “out there”; no longer could we ask, as Heisenberg put it, whether “particles exist in space and time objectively.”

(If you're wondering why no one called him out, or more specifically why we have a bias to not call BS out, check out pluralistic ignorance).

Jargon does have its place. In specific contexts, it is absolutely vital. But in everyday communication, its use is a sign that we wish to appear complex and therefore more intelligent. Great thinkers throughout the ages have stressed the crucial importance of using simple language to convey complex ideas. Many of the ancient thinkers whose work we still reference today — people like Plato, Marcus Aurelius, Seneca, and Buddha — were known for their straightforward communication and their ability to convey great wisdom in a few words.

“Any intelligent fool can make things bigger, more complex, and more violent. It takes a touch of genius — and a lot of courage — to move in the opposite direction.”

— Ernst F. Schumacher

How Can We Overcome Complexity Bias?

The most effective tool we have for overcoming complexity bias is Occam’s razor. Also known as the principle of parsimony, this is a problem-solving principle used to eliminate improbable options in a given situation. Occam’s razor suggests that the simplest solution or explanation is usually the correct one. When we don’t have enough empirical evidence to disprove a hypothesis, we should avoid making unfounded assumptions or adding unnecessary complexity so we can make quick decisions or establish truths.

An important point to note is that Occam’s razor does not state that the simplest hypothesis is the correct one, but states rather that it is the best option before the establishment of empirical evidence. It is also useful in situations where empirical data is difficult or impossible to collect. While complexity bias leads us towards intricate explanations and concepts, Occam’s razor can help us to trim away assumptions and look for foundational concepts.

Returning to Skinner’s pigeons, had they known of Occam’s razor, they would have realized that there were two main possibilities:

  • Their behavior affects the food delivery.


  • Their behavior is irrelevant because the food delivery is random or on a timed schedule.

Using Occam’s razor, the head-bobbing, circles-turning pigeons would have realized that the first hypothesis involves numerous assumptions, including:

  • There is a particular behavior they must enact to receive food.
  • The delivery mechanism can somehow sense when they enact this behavior.
  • The required behavior is different from behaviors that would normally give them access to food.
  • The delivery mechanism is consistent.

And so on. Occam’s razor would dictate that because the second hypothesis is the simplest, involving the fewest assumptions, it is most likely the correct one.

So many geniuses, are really good at eliminating unnecessary complexity. Einstein, for instance, was a master at sifting the essential from the non-essential. Steve Jobs was the same.

Comment on Facebook | Twitter

Do Algorithms Beat Us at Complex Decision Making?

Algorithms are all the rage these days. AI researchers are taking more and more ground from humans in areas like rules-based games, visual recognition, and medical diagnosis. However, the idea that algorithms make better predictive decisions than humans in many fields is a very old one.

In 1954, the psychologist Paul Meehl published a controversial book with a boring sounding name: Clinical vs. Statistical Prediction: A Theoretical Analysis and a Review of the Evidence.

The controversy? After reviewing the data, Meehl claimed that mechanical, data-driven algorithms could better predict human behavior than trained clinical psychologists — and with much simpler criteria. He was right.

The passing of time has not been friendly to humans in this game: Studies continue to show that the algorithms do a better job than experts in a range of fields. In Daniel Kahneman's Thinking Fast and Slow, he details a selection of fields which have demonstrated inferior human judgment compared to algorithms:

The range of predicted outcomes has expanded to cover medical variables such as the longevity of cancer patients, the length of hospital stays, the diagnosis of cardiac disease, and the susceptibility of babies to sudden infant death syndrome; economic measures such as the prospects of success for new businesses, the evaluation of credit risks by banks, and the future career satisfaction of workers; questions of interest to government agencies, including assessments of the suitability of foster parents, the odds of recidivism among juvenile offenders, and the likelihood of other forms of violent behavior; and miscellaneous outcomes such as the evaluation of scientific presentations, the winners of football games, and the future prices of Bordeaux wine.

The connection between them? Says Kahneman: “Each of these domains entails a significant degree of uncertainty and unpredictability.” He called them “low-validity environments”, and in those environments, simple algorithms matched or outplayed humans and their “complex” decision making criteria, essentially every time.


A typical case is described in Michael Lewis' book on the relationship between Daniel Kahneman and Amos Tversky, The Undoing Project. He writes of work done at the Oregon Research Institute on radiologists and their x-ray diagnoses:

The Oregon researchers began by creating, as a starting point, a very simple algorithm, in which the likelihood that an ulcer was malignant depended on the seven factors doctors had mentioned, equally weighted. The researchers then asked the doctors to judge the probability of cancer in ninety-six different individual stomach ulcers, on a seven-point scale from “definitely malignant” to “definitely benign.” Without telling the doctors what they were up to, they showed them each ulcer twice, mixing up the duplicates randomly in the pile so the doctors wouldn't notice they were being asked to diagnose the exact same ulcer they had already diagnosed. […] The researchers' goal was to see if they could create an algorithm that would mimic the decision making of doctors.

This simple first attempt, [Lewis] Goldberg assumed, was just a starting point. The algorithm would need to become more complex; it would require more advanced mathematics. It would need to account for the subtleties of the doctors' thinking about the cues. For instance, if an ulcer was particularly big, it might lead them to reconsider the meaning of the other six cues.

But then UCLA sent back the analyzed data, and the story became unsettling. (Goldberg described the results as “generally terrifying”.) In the first place, the simple model that the researchers had created as their starting point for understanding how doctors rendered their diagnoses proved to be extremely good at predicting the doctors' diagnoses. The doctors might want to believe that their thought processes were subtle and complicated, but a simple model captured these perfectly well. That did not mean that their thinking was necessarily simple, only that it could be captured by a simple model.

More surprisingly, the doctors' diagnoses were all over the map: The experts didn't agree with each other. Even more surprisingly, when presented with duplicates of the same ulcer, every doctor had contradicted himself and rendered more than one diagnosis: These doctors apparently could not even agree with themselves.


If you wanted to know whether you had cancer or not, you were better off using the algorithm that the researchers had created than you were asking the radiologist to study the X-ray. The simple algorithm had outperformed not merely the group of doctors; it had outperformed even the single best doctor.

The fact that doctors (and psychiatrists, and wine experts, and so forth) cannot even agree with themselves is a problem called decision making “noise”: Given the same set of data twice, we make two different decisions. Noise. Internal contradiction.

Algorithms win, at least partly, because they don't do this: The same inputs generate the same outputs every single time. They don't get distracted, they don't get bored, they don't get mad, they don't get annoyed. Basically, they don't have off days. And they don't fall prey to the litany of biases that humans do, like the representativeness heuristic.

The algorithm doesn't even have to be a complex one. As demonstrated above with radiology, simple rules work just as well as complex ones. Kahneman himself addresses this in Thinking, Fast and Slow when discussing Robyn Dawes's research on the superiority of simple algorithms using a few equally-weighted predictive variables:

The surprising success of equal-weighting schemes has an important practical implication: it is possible to develop useful algorithms without prior statistical research. Simple equally weight formulas based on existing statistics or on common sense are often very good predictors of significant outcomes. In a memorable example, Dawes showed that marital stability is well predicted by a formula: Frequency of lovemaking minus frequency of quarrels.

You don't want your result to be a negative number.

The important conclusion from this research is that an algorithm that is constructed on the back of an envelope is often good enough to compete with an optimally weighted formula, and certainly good enough to outdo expert judgment. This logic can be applied in many domains, ranging from the selection of stocks by portfolio managers to the choices of medical treatments by doctors or patients.

Stock selection, certainly a “low validity environment”, is an excellent example of the phenomenon.

As John Bogle pointed out to the world in the 1970's, a point which has only strengthened with time, the vast majority of human stock-pickers cannot outperform a simple S&P 500 index fund, an investment fund that operates on strict algorithmic rules about which companies to buy and sell and in what quantities. The rules of the index aren't complex, and many people have tried to improve on them with less success than might be imagined.


Another interesting area where this holds is interviewing and hiring, a notoriously difficult “low-validity” environment. Even elite firms often don't do it that well, as has been well documented.

Fortunately, if we take heed of the advice of the psychologists, operating in a low-validity environment has rules that can work very well. In Thinking Fast and Slow, Kahneman recommends fixing your hiring process by doing the following (or some close variant), in order to replicate the success of the algorithms:

Suppose you need to hire a sales representative for your firm. If you are serious about hiring the best possible person for the job, this is what you should do. First, select a few traits that are prerequisites for success in this position (technical proficiency, engaging personality, reliability, and so on). Don't overdo it — six dimensions is a good number. The traits you choose should be as independent as possible from each other, and you should feel that you can assess them reliably by asking a few factual questions. Next, make a list of questions for each trait and think about how you will score it, say on a 1-5 scale. You should have an idea of what you will call “very weak” or “very strong.”

These preparations should take you half an hour or so, a small investment that can make a significant difference in the quality of the people you hire. To avoid halo effects, you must collect the information one at a time, scoring each before you move on to the next one. Do not skip around. To evaluate each candidate, add up the six scores. […] Firmly resolve that you will hire the candidate whose final score is the highest, even if there is another one whom you like better–try to resit your wish to invent broken legs to change the ranking. A vast amount of research offers a promise: you are much more likely to find the best candidate if you use this procedure than if you do what people normally do in such situations, which is to go into the interview unprepared and to make choices by an overall intuitive judgment such as “I looked into his eyes and liked what I saw.”

In the battle of man vs algorithm, unfortunately, man often loses. The promise of Artificial Intelligence is just that. So if we're going to be smart humans, we must learn to be humble in situations where our intuitive judgment simply is not as good as a set of simple rules.

Blog Posts, Book Reviews, and Abstracts: On Shallowness

We’re quite glad that you read Farnam Street, and we hope we’re always offering you a massive amount of value. (If not, email us and tell us what we can do more effectively.)

But there’s a message all of our readers should appreciate: Blog posts are not enough to generate the deep fluency you need to truly understand or get better at something. We offer a starting point, not an end point.

This goes just as well for book reviews, abstracts, cliff's notes, and a good deal of short-form journalism.

This is a hard message for some who want a shortcut. They want the “gist” and the “high level takeaways”, without doing the work or eating any of the broccoli. They think that’s all it takes: Check out a 5-minute read, and instantly their decision making and understanding of the world will improve right-quick. Most blogs, of course, encourage this kind of shallowness. Because it makes you feel that the whole thing is pretty easy.

Here’s the problem: The world is more complex than that. It doesn’t actually work this way. The nuanced detail behind every “high level takeaway” gives you the context needed to use it in the real world. The exceptions, the edge cases, and the contradictions.

Let me give you an example.

A high-level takeaway from reading Kahneman’s Thinking Fast, and Slow would be that we are subject to something he and Amos Tversky call the Representativeness Heuristic. We create models of things in our head, and then fit our real-world experiences to the model, often over-fitting drastically. A very useful idea.

However, that’s not enough. There are so many follow-up questions. Where do we make the most mistakes? Why does our mind create these models? Where is this generally useful? What are the nuanced examples of where this tendency fails us? And so on. Just knowing about the Heuristic, knowing that it exists, won't perform any work for you.

Or take the rise of human species as laid out by Yuval Harari. It’s great to post on his theory; how myths laid the foundation for our success, how “natural” is probably a useless concept the way it’s typically used, and how biology is the great enabler.

But Harari’s book itself contains the relevant detail that fleshes all of this out. And further, his bibliography is full of resources that demand your attention to get even more backup. How did he develop that idea? You have to look to find out.

Why do all this? Because without the massive, relevant detail, your mind is built on a house of cards.

What Farnam Street and a lot of other great resources give you is something like a brief map of the territory.

Welcome to Colonial Williamsburg! Check out the re-enactors, the museum, and the theatre. Over there is the Revolutionary City. Gettysburg is 4 hours north. Washington D.C. is closer to 2.5 hours.

Great – now you have a lay of the land. Time to dig in and actually learn about the American Revolution. (This book is awesome, if you actually want to do that.)

Going back to Kahneman, one of his and Tversky’s great findings was the concept of the Availability Heuristic. Basically, the mind operates on what it has close at hand.

As Kahneman puts it, “An essential design feature of the associative machine is that it represents only activated ideas. Information that is not retrieved (even unconsciously) from memory might as well not exist. System 1 excels at constructing the best possible story that incorporates ideas currently activated, but it does not (cannot) allow for information it does not have.”

That means that in the moment of decision making, when you’re thinking hard on some complex problem you face, it’s unlikely that your mind is working all that successfully without the details. It doesn't have anything to draw on. It’d be like a chess player who read a book about great chess players, but who hadn’t actually studied all of their moves. Not very effective.

The great difficulty, of course, is that we lack the time to dig deep into everything. Opportunity costs and trade-offs are quite real.

That’s why you must develop excellent filters. What’s worth learning this deeply? We think it’s the first-principle style mental models. The great ideas from physical systems, biological systems, and human systems. The new-new thing you’re studying is probably either A. Wrong or B. Built on one of those great ideas anyways. Farnam Street, in a way, is just a giant filtering mechanism to get you started down the hill.

But don't stop there. Don't stop at the starting line. Resolve to increase your depth and stop thinking you can have it all in 5 minutes or less. Use our stuff, and whoever else's stuff you like, as an entrée to the real thing.

Daniel Kahneman on Human Gullibility

“The premise of this book is that it is easier to recognize other people's mistakes than our own.”


A simple article connecting two ideas from Daniel Kahneman's Thinking Fast and Slow on human gullibility and availability bias.

A reliable way to make people believe in falsehoods is frequent repetition, because familiarity is not easily distinguished from truth. Authoritarian institutions and marketers have always known this fact. But it was psychologists who discovered that you do not have to repeat the entire statement of a fact or idea to make it appear true. People who were repeatedly exposed to the phrase “the body temperature of a chicken” were more likely to accept as true the statement that “the body temperature of a chicken is 144°” (or any other arbitrary number). The familiarity of one phrase in the statement sufficed to make the whole statement feel familiar, and therefore true. If you cannot remember the source of a statement, and have no way to relate it to other things you know, you have no option but to go with the sense of cognitive ease.

This is due, in part, to the fact that repetition causes familiarity and familiarity distorts our thinking.

People tend to assess the relative importance of issues by the ease with which they are retrieved from memory—and this is largely determined by the extent of coverage in the media. Frequently mentioned topics populate the mind even as others slip away from awareness. In turn, what the media choose to report corresponds to their view of what is currently on the public's mind. It is no accident that authoritarian regimes exert substantial pressure on independent media. Because public interest is most easily aroused by dramatic events and by celebrities, media feeding frenzies are common. For several weeks after Michael Jackson's death, for example, it was virtually impossible to find a television channel reporting on another topic. In contrast, there is little coverage of critical but unexciting issues that provide less drama, such as declining educational standards or overinvestment of medical resources in the last year of life. (As I write this, I notice that my choice of “little-covered” examples was guided by availability. The topics I chose as examples are mentioned often; equally important issues that are less available did not come to my mind.)

Mental Model: Bias from Conjunction Fallacy

Daniel Kahneman and Amos Tversky spent decades in psychology research to disentangle patterns in errors of human reasoning. Over the course of their work they discovered a variety of logical fallacies that we tend to make, when facing information that appears vaguely familiar. These fallacies lead to bias – irrational behavior based on beliefs that are not always grounded in reality.

In his book Thinking Fast and Slow, which summarizes his and Tversky’s life work, Kahneman introduces biases that stem from the conjunction fallacy – the false belief that a conjunction of two events is more probable than one of the events on its own.

What is Probability?

Probability can be a difficult concept. Most of us have an intuitive understanding of what probability is, but there is little consensus on what it actually means. It is just as vague and subjective a concept as democracy, beauty or freedom. However, this is not always troublesome – we can still easily discuss the notion with others. Kahneman reflects:

In all the years I spent asking questions about the probability of events, no one ever raised a hand to ask me, “Sir, what do you mean by probability?” as they would have done if I had asked them to assess a strange concept such as globability.

Everyone acted as if they knew how to answer my questions, although we all understood that it would be unfair to ask them for an explanation of what the word means.

While logicians and statisticians might disagree, probability to most of us is simply a tool that describes our degree of belief. For instance, we know that the sun will rise tomorrow and we consider it near impossible that there will be two suns up in the sky instead of one. In addition to the extremes, there are also events which lie somewhere in the middle on the probability spectrum, such as the degree of belief that it will rain tomorrow.

Despite its vagueness, probability has its virtues. Assigning probabilities helps us make the degree of belief actionable and also communicable to others. If we believe that the probability it will rain tomorrow is 90%, we are likely to carry an umbrella and suggest our family do so as well.

Probability, Base Rates and Representativeness

Most of us are already familiar with representativeness and base rates. Consider the classic example of x number of black and y number of white colored marbles in a jar. It is a simple exercise to tell what the probabilities of drawing each color are if you know their base rates (proportion). Using base rates is the obvious approach for estimations when no other information is provided.

However, Kahneman managed to prove that we have a tendency to ignore base rates in light of specific descriptions. He calls this phenomenon the Representativeness Bias. To illustrate representativeness bias, consider the example of seeing a person reading The New York Times on the New York subway. Which do you think would be a better bet about the reading stranger?

1) She has a PhD.
2) She does not have a college degree.

Representativeness would tell you to bet on the PhD, but this is not necessarily a good idea. You should seriously consider the second alternative, because many more non-graduates than PhDs ride in New York subways. While a larger proportion of PhDs may read The New York Times, the total number of New York Times readers with only high school degrees is likely to be much larger, even if the proportion itself is very slim.

In a series of similar experiments, Kahneman’s subjects failed to recognize the base rates in light of individual information. This is unsurprising. Kahneman explains:

On most occasions, people who act friendly are in fact friendly. A professional athlete who is very tall and thin is much more likely to play basketball than football. People with a PhD are more likely to subscribe to The New York Times than people who ended their education after high school. Young men are more likely than elderly women to drive aggressively.

While following representativeness bias might improve your overall accuracy, it will not always be the statistically optimal approach.

Michael Lewis in his bestseller Moneyball tells a story of Oakland A’s baseball team coach, Billy Beane, who recognized this fallacy and used it to his advantage. When recruiting new players for the team, instead of relying on scouts he relied heavily on statistics of past performance. This approach allowed him to build a team of great players that were passed up by other teams because they did not look the part. Needless to say, the team achieved excellent results at a low cost.

Conjunction Fallacy

While representativeness bias occurs when we fail to account for low base rates, conjunction fallacy occurs when we assign a higher probability to an event of higher specificity. This violates the laws of probability.

Consider the following study:

Participants were asked to rank four possible outcomes of the next Wimbledon tournament from most to least probable. Björn Borg was the dominant tennis player of the day when the study was conducted. These were the outcomes:

A. Borg will win the match.
B. Borg will lose the first set.
C. Borg will lose the first set but win the match.
D. Borg will win the first set but lose the match.

How would you order them?

Kahneman was surprised to see that most subjects ordered the chances by directly contradicting the laws of logic and probability. He explains:

The critical items are B and C. B is the more inclusive event and its probability must be higher than that of an event it includes. Contrary to logic, but not to representativeness or plausibility, 72% assigned B a lower probability than C.

If you thought about the problem carefully you drew the following diagram in your head. Losing the first set will always, by definition, be a more probable event than losing the first set and winning the match.
Screen Shot 2016-08-05 at 6.28.30 PM

The Linda Problem

As discussed in our piece on the Narrative Fallacy, the best-known and most controversial of Kahneman and Tversky’s experiments involved a fictitious lady called Linda. The fictional character was created to illustrate the role heuristics play in our judgement and how it can be incompatible with logic. This is how they described Linda.

Linda is thirty-one years old, single, outspoken, and very bright. She majored in philosophy. As a student, she was deeply concerned with issues of discrimination and social justice, and also participated in antinuclear demonstrations.

Kahneman conducted a series of experiments, in which he showed that representativeness tends to cloud our judgements and that we ignore the base rates in light of stories. The Linda problem started off with the task to estimate the plausibility of 9 different scenarios that subjects were supposed to rank in order of likelihood.

Linda is a teacher in elementary school.
Linda works in a bookstore and takes yoga classes.
Linda is active in the feminist movement.
Linda is a psychiatric social worker.
Linda is a member of the League of Women Voters.
Linda is a bank teller.
Linda is an insurance salesperson.
Linda is a bank teller and is active in the feminist movement.

Kahneman was startled to see that his subjects judged the likelihood of Linda being a bank teller and a feminist more likely than her being just a bank teller. As explained earlier, doing so makes little sense. He went on to explore the phenomenon further:

In what we later described as “increasingly desperate” attempts to eliminate the error, we introduced large groups of people to Linda and asked them this simple question:

Which alternative is more probable?

Linda is a bank teller.
Linda is a bank teller and is active in the feminist movement.

This stark version of the problem made Linda famous in some circles, and it earned us years of controversy. About 85% to 90% of undergraduates at several major universities chose the second option, contrary to logic.

What is especially interesting about these results is that, even when aware of the biases in place, we do not discard them.

When I asked my large undergraduate class in some indignation, “Do you realize that you have violated an elementary logical rule?” someone in the back row shouted, “So what?” and a graduate student who made the same error explained herself by saying, “I thought you just asked for my opinion.”

The issue is not constrained to students and but also affects professionals.

The naturalist Stephen Jay Gould described his own struggle with the Linda problem. He knew the correct answer, of course, and yet, he wrote, “a little homunculus in my head continues to jump up and down, shouting at me—‘but she can’t just be a bank teller; read the description.”

Our brains simply seem to prefer consistency over logic.

The Role of Plausibility

Representativeness and conjunction fallacy occur, because we make the mental shortcut from our perceived plausibility of a scenario to its probability.

The most coherent stories are not necessarily the most probable, but they are plausible, and the notions of coherence, plausibility, and probability are easily confused by the unwary. Representativeness belongs to a cluster of closely related basic assessments that are likely to be generated together. The most representative outcomes combine with the personality description to produce the most coherent stories.

Kahneman warns us about the effects of these biases on our perception of expert opinion and forecasting. He explains that we are more likely to believe scenarios that are illustrative rather than probable.

The uncritical substitution of plausibility for probability has pernicious effects on judgments when scenarios are used as tools of forecasting. Consider these two scenarios, which were presented to different groups, with a request to evaluate their probability:

A massive flood somewhere in North America next year, in which more than 1,000 people drown

An earthquake in California sometime next year, causing a flood in which more than 1,000 people drown

The California earthquake scenario is more plausible than the North America scenario, although its probability is certainly smaller. As expected, probability judgments were higher for the richer and more detailed scenario, contrary to logic. This is a trap for forecasters and their clients: adding detail to scenarios makes them more persuasive, but less likely to come true.

In order to appreciate the role of plausibility, he suggests we have a look at an example without an accompanying explanation.

Which alternative is more probable?

Jane is a teacher.
Jane is a teacher and walks to work.

In this case, when evaluating plausibility and coherence there are no quick answers to the probability question and we can easily conclude that the first one is more likely. The rule goes that in the absence of a competing intuition, logic prevails.

Taming our intuition

The first lesson to thinking clearly is to question how you think. We should not simply believe whatever comes to our mind – our beliefs must be constrained by logic. You don’t have to become an expert in probability to tame your intuition, but having a grasp of simple concepts will help. There are two main rules that are worth repeating in light of representativeness bias:

1) All probabilities add up to 100%.

This means that if you believe that there’s a 90% chance it will rain tomorrow, there’s a 10% of chance that it will not rain tomorrow.

However, since you believe that there is only 90% chance that it will rain tomorrow, you cannot be 95% certain that it will rain tomorrow morning.

We typically make this type of error, when we mean to say that, if it rains, there’s 95% probability it will happen in the morning. That’s a different claim and the probability of raining tomorrow morning under such premises is 0.9*0.95=85.5%.

This also means the odds that, if it rains, it will not rain in the morning, are 90.0%-85.5% = 4.5%.

2) The second principle is called the Bayes rule.

 It allows us to correctly adjust our beliefs with the diagnosticity of the evidence. Bayes rule follows the formula:


In essence the formula states that the posterior odds are proportional to prior odds times the likelihood. Kahneman crystallizes two keys to disciplined Bayesian reasoning:

• Anchor your judgment of the probability of an outcome on a plausible base rate.
• Question the diagnosticity of your evidence.

Kahnmenan explains it with an example:

If you believe that 3% of graduate students are enrolled in computer science (the base rate), and you also believe that the description of Tom is 4 times more likely for a graduate student in computer science than in other fields, then Bayes’s rule says you must believe that the probability that Tom is a computer science student is now 11%.

Four times as likely means that we expect roughly 80% of all computer science students to resemble Tom. We use this proportion to obtain the adjusted odds. (The calculation goes as follows: 0.03*0.8/(0.03*0.8+((1-0.03)*(1-0.8)))=11%)

The easiest way to become better at making decisions is by making sure you question your assumptions and follow strong evidence. When evidence is anecdotal, adjust minimally and trust the base rates. Odds are, you will be pleasantly surprised.


Want More? Check out our ever-growing collection of mental models and biases and get to work.

Daniel Kahneman in Conversation with Michael Mauboussin on Intuition, Causality, Loss Aversion and More

Ever want to be the fly on the wall for a fascinating conversation. Well, here's your chance. Santa Fe Institute Board of Trustees Chair Michael Mauboussin interviews Nobel Prize winner Daniel Kahneman. The wide-ranging conversation talks about disciplined intuition, causality, base rates, loss aversion and so much more. You don't want to miss this.

Here's an excerpt from Kahneman I think you'll enjoy. You can read the entire transcript here.

The Sources of Power is a very eloquent book on expert intuition with magnificent examples, and so he is really quite hostile to my point of view, basically.

We spent years working on that, on the question of when can intuitions be trusted? What's the boundary between trustworthy and untrustworthy intuitions?

I would summarize the answer as saying there is one thing you should not do. People's confidence in their intuition is not a good guide to their validity. Confidence is something else entirely, and maybe we can talk about confidence separately later, but confidence is not it.

What there is, if you want to know whether you can trust intuition, it really is like deciding on a painting, whether it's genuine or not. You can look at the painting all you want, but asking about the provenance is usually the best guide about whether a painting is genuine or not.

Similarly for expertise and intuition, you have to ask not how happy the individual is with his or her own intuitions, but first of all, you have to ask about the domain. Is the domain one where there is enough regularity to support intuitions? That's true in some medical domains, it certainly is true in chess, it is probably not true in stock picking, and so there are domains in which intuition can develop and others in which it cannot. Then you have to ask whether, if it's a good domain, one in which there are regularities that can be picked up by the limited human learning machine. If there are regularities, did the individual have an opportunity to learn those regularities? That primarily has to do with the quality of the feedback.

Those are the questions that I think should be asked, so there is a wide domain where intuitions can be trusted, and they should be trusted, and in a way, we have no option but to trust them because most of the time, we have to rely on intuition because it takes too long to do anything else.

Then there is a wide domain where people have equal confidence but are not to be trusted, and that may be another essential point about expertise. People typically do not know the limits of their expertise, and that certainly is true in the domain of finances, of financial analysis and financial knowledge. There is no question that people who advise others about finances have expertise about finance that their advisees do not have. They know how to look at balance sheets, they understand what happens in conversations with analysts.

There is a great deal that they know, but they do not really know what is going to happen to a particular stock next year. They don't know that, that is one of the typical things about expert intuition in that we know domains where we have it, there are domains where we don't, but we feel the same confidence and we do not know the limits of our expertise, and that sometimes is quite dangerous.