1. You can’t reason based on track record in the same way you can with normal predictions.
2. The stakes are extremely high. Being wrong on normal predictions rarely matters as much.

Why?

Regarding point one, reasoning based on track record:

Normally, a type of prediction being wrong again and again will lead you to dismiss that type of prediction. For instance, if every year (for some reason), experts predict that your country will soon have the highest math scores in the world, and yet each year it is ranked 50th in such scores, eventually, you (rightly) ignore the experts.

However, with extinction risks, this kind of reasoning doesn’t quite work. In all possible universes, those who predict the extinction of their species will be wrong right up until extinction happens. The predictions can, at most, be right only once.

Consider two worlds: one where humans go extinct in 2030 and one where they don’t ever go extinct (or go extinct only much later). What would you observe in 2029 regarding past predictions of extinction in these two worlds?

Well, in both worlds you’d observe that all past extinction predictions had failed up until that point. (If anything, I’d anticipate having MORE past extinction predictions fail in the world where extinction happens in 2030 since there would be more evidence of potential extinction in that world, all else equal.)

Therefore, the reasoning that “we’ve had a lot of past extinction predictions and they’ve always failed, therefore extinction is unlikely” is not a good argument – you’d witness these failed predictions in both such worlds (and perhaps even more of them in the world where extinction happens soon).

This makes predictions of extinction a special class of prediction.

To dismiss arguments about extinction risk, it’s necessary to engage with the actual arguments themselves, as they can’t be dismissed as a group due to past failed predictions. While near misses can tell you about the probability of some extinction risks (e.g., times when nuclear war nearly broke out or asteroid near impacts), failed predictions are not very informative.

Regarding point two, the enormous stakes:

Extinction, most people will agree, would be incredibly bad. For that reason, they don’t have to be very likely to be worth taking very, very seriously.

In a world where there were millions of distinct, plausible extinction risks, the large number of them would suggest that each one is (a priori) not that likely to end the species, and in such a world, it might be silly to invest much in these kinds of concerns (unless a smaller number of much more likely ones could be identified).

But that’s not the world we live in. There are only around 13 plausible human extinction risks – and in this short list, some of them aren’t even really plausible (when considered as a potential cause of literally ALL humans dying out). Here’s the list, in no particular order (if I missed any, let me know):

1. Advanced AI technology
2. Nuclear war or the invention of new destructive weapons
3. Pathogens (e.g., human-engineered viruses)
4. Asteroids, extreme solar flares, supernovae, gamma-ray bursts, or other cosmological events
5. A mega volcano, mega earthquake, dramatic change in the earth’s magnetic field, or another major geological event
6. Advanced nanotech (e.g., grey goo) or synthetic biology
7. The second coming of various figures according to different religions, or god(s) or demons ending the world or terminating our species
8. Simulators (if we’re living in a simulation) ending the world or our species
9. Aliens from other planets
10. Runaway climate change/extreme climate shifts, or sudden ecosystem collapse
11. Physics experiments (e.g., related to vacuum stability) gone wrong or that were purposely carried out to end humanity
12. Universal environmental contaminates turning out to be deadly or to cause infertility
13. Extreme population decline until no reproduction takes place (e.g., following an event that greatly reduces the world population)

Obviously, some of these are much less probable than others. And maybe you think some of these are ridiculous. Okay, cross those out. What about the others?

Given the incredible stakes, the short size of the list, and humanity’s (in my view, bizarre and irrational) unwillingness to protect its own future, all of these are worth investing much more in than humanity currently does. Obviously, it could be stupid for humanity to invest so much in preventing extinction that it’s seriously impaired. But we invest so little, it’s almost absurd.

I don’t think that predictions of extinction can be easily dismissed, despite all prior such predictions being wrong – they don’t work like other predictions do and are much higher stakes.

This piece was first written on February 13, 2025, and first appeared on my website on April 9, 2025.

1. Music: we prefer songs that are similar to others we like but that feel novel. If a song is too similar to what we know, then it’s derivative or boring (like listening to the same music on loop), but if it’s too novel, it is usually unappealing or dissonant. Music from other cultures can be hard for us to appreciate until we’ve listened to enough of it.

2. Learning: we learn best when an idea connects to what we already understand while also adding something on top. If it’s too far removed from what we already understand, it’s confusing, or we don’t believe it. Learning often works best when bridges are built between a person’s old understanding and some new understanding.

3. Products: we like user interfaces to be familiar, but also like them to let us do something new (or do something we value in a new, better way). An unfamiliar interface feels like a struggle to learn and can leave us frustrated. One of the great strengths of the iPhone, when it was introduced, was that the interface somehow felt familiar even though we had never used something like it before.

4. Dating: most people want those they date to be familiar in most ways (e.g., similar culture, religion, political beliefs, life goals, attitudes) yet highly distinct with respect to a few key variables. The differences being sought differ for each person, but some fairly common ones sometimes sought include: a masculine person contrasting with feminine one, a submissive person contrasting with a dominant one, a breadwinner contrasting with a homemaker, a side-kick contrasting with a leader, a joker contrasting with someone who laughs often, a strong person contrasting with someone in need of protection, or someone successful at rule-breaking contrasting with someone successful at being rule-abiding.

5. Stories: we enjoy stories that follow the standard tropes (e.g., the Hero’s Journey or the “love found → love lost → love recovered” romantic comedy), yet we want them to feel novel in some way and not too derivative. Truly novel movies are rarely as popular as ones that are executed on a classic formula but strategically deviate from it in enough key ways to feel fresh. For instance, I think The Matrix did a good job of telling a very standard “hero’s journey” story but with a number of novel-feeling elements thrown in (an intriguing setting, “bullet time” filming, etc.).

So if you’re making something, and you want people to really enjoy it, you may want to consider how you can make it Familiar Yet Different.

This piece was first written on September 18, 2020, and first appeared on this site on June 10, 2022.

1. Parkinson’s Law: Work expands so as to fill the time available for its completion.
2. Hofstadter’s Law: It always takes longer than you expect, even when you take into account Hofstadter’s Law.
3. Gates’ Law: Most people overestimate what they can do in one year and underestimate what they can do in ten years.
4. Goodhart’s Law: When a measure becomes a target, it ceases to be a good measure.
5. Hanlon’s Razor: Never attribute to malice that which is adequately explained by stupidity (or, don’t invoke conspiracy when ignorance and incompetence will suffice, as conspiracy implies intelligence).
6. Acton’s Dictum: Power tends to corrupt, and absolute power corrupts absolutely.
7. Amara’s Law: We tend to overestimate the effect of a technology in the short run and underestimate the effect in the long run.
8. Benford’s Law: In a diverse collection of unrelated statistics, a given statistic has roughly a 30% chance of starting with the digit 1.
9. Betteridge’s Law: Any headline which ends in a question mark can be answered by the word ‘no’.
10. Brooks’ Law: Adding manpower to a late software project makes it later.
11. Chesterton’s Fence: Reforms should not be made until the reasoning behind the existing state of affairs is understood.
12. Claasen’s Law: Usefulness = log(technology).
13. Clarke’s First Law: When a distinguished elderly scientist states that something is possible, they are almost certainly right, but when they state something is impossible, they are probably wrong.
14. Cromwell’s Rule: Nothing but logical impossibilities have a prior probability of 0 or 1.
15. Cunningham’s Law: The best way to get the right answer on the Internet is not to ask a question, it’s to post the wrong answer.
16. Doctorow’s Law: When someone puts a lock on a thing you own, against your wishes, and doesn’t give you the key, they’re not doing it for your benefit.
17. Dunbar’s Number: Most people can’t maintain stable social relationships with more than 150 people.
18. Eroom’s Law: Drug discovery is becoming slower and more expensive over time, despite improvements in technology.
19. Gell-Mann Amnesia Effect: You’ll believe articles outside your area of expertise, even after acknowledging that neighboring articles in your area of expertise are completely wrong.
20. Gibson’s Law (or the Expert Witness Law): For each PhD (to use as an expert witness for one side) there’s an equal and opposite PhD.
21. Godwin’s Law: As an online discussion grows longer, the probability of a comparison involving Nazis or Hitler approaches one.
22. Morley-Souter’s Law (Rule 34): There is porn of it (no exceptions).
23. Greenspun’s Tenth Rule: Any sufficiently complicated C program contains an ad hoc, informally specified, bug-ridden, slow implementation of half of Common Lisp.
24. Hebb’s Law: Neurons that fire together wire together.
25. Hubble’s Law: Galaxies recede from an observer at a rate proportional to their distance to that observer.
26. Hume’s Guillotine (Is-Ought Problem): Normative statements (about what’s moral/immoral/right/wrong) cannot be deduced exclusively from descriptive statements.
27. Humphrey’s Law: Conscious attention to a task normally performed automatically can impair its performance.
28. Kranzberg’s Law: Technology is neither good nor bad; nor is it neutral.
29. Lamarck’s Principle (or “Use it or Lose it”): Use it or lose it (evolutionarily speaking, but also in the brain).
30. Lewis’s Law: The comments you’ll inevitably find on any article about feminism justify feminism.
31. Littlewood’s Law: Individuals can expect miracles to happen to them, at the rate of about one per month.
32. Maes–Garreau Law: Favorable predictions about future technology will fall at the latest possible date they can come true and still remain in the lifetime of the predictor.
33. Metcalfe’s Law: The value of a system grows as approximately the square of the number of users of the system.
34. Miller’s Law: To understand what another person is saying, you must assume that it is true and try to imagine what it could be true of.
35. Moore’s Law: Computation per dollar grows exponentially (or: number of transistors per circuit doubles roughly every 24 months).
36. Murphy’s Law: Anything that can go wrong will go wrong.
37. Alder’s Law: What cannot be settled by experiment is not worth debating.
38. O’Sullivan’s Law: All organizations that are not actually right-wing will over time become left-wing.
39. Pareto’s Principle (80/20 Rule): For many phenomena 80% of consequences stem from 20% of the causes.
40. Peter’s Principle: In a hierarchy, every employee tends to rise to his level of incompetence.
41. Poisson’s Law (or Law of Large Numbers): For independent random variables with a common distribution, the average tends to the mean as sample size increases.
42. Pournelle’s Iron Law of Bureaucracy: In bureaucracy, those devoted to the bureaucracy get control, those devoted to what it’s supposed to achieve lose influence.
43. Putt’s Law: Technology is dominated by two types of people: those who understand what they do not manage and those who manage what they do not understand.
44. Rosenthal Effect (Pygmalion Effect): High expectations lead to an increase in performance, low expectations to a decrease in performance.
45. Schneier’s Law: Any person can invent a security system so clever that she or he can’t think of how to break it.
46. Shermer’s Law: Any sufficiently advanced extraterrestrial intelligence is indistinguishable from God.
47. Zipf’s Law: The frequency of use of the nth-most-frequently-used word in any natural language is approximately inversely proportional to n (few words are used often, most are used rarely).
48. Wirth’s Law: Software gets slower more quickly than hardware gets faster.
49. Sturgeon’s Law: Ninety percent of everything is crud.
50. Stigler’s Law: No discovery is named after its original discoverer, including this one.

This piece was first written on July 6, 2020, and first appeared on my website on May 30, 2026.

In the attached image, there are two non-existent women (generated by a face generation AI set to generate “brown hair white adult female”). If these were real people, they would likely be treated differently throughout their lives due to very minor differences in facial structure and skin smoothness.

Based on their faces alone, there’s no way to know with non-negligible accuracy which of these people (if they existed) would be more hard-working, more moral, wiser, or otherwise in possession of personal traits that we actually might care about. So why are humans so obsessed with faces? It seems likely to be caused by a combination of two factors:

(1) Runaway Sexual Selection

If peacocks find large tail plumage sexually attractive, then even if those feathers are not useful for anything else, that still creates an evolutionary selection pressure where those with larger tail plumage are more likely to pass on their genes (due to improved chances of mating). Similarly, if certain humans are found to be more attractive based on their faces, that creates an evolutionary selection pressure in favor of mating with those people because then their children have a higher probability of finding mating success themselves (and hence passing on their genes). This phenomenon reinforces faces being attractive (because those attracted to “good-looking” faces mate with “good-looking” people more often, therefore their children are more good-looking and so have an easier time mating, plus have a preference for “good-looking” faces).

Today, this selection pressure is likely much weaker than it once was since most people now end up having children. For instance, now the vast majority of people in the US live to be at least 50, and only about 15% of women and 25% of men in the 40-50-year age bracket are childless. In contrast, tens of thousands of years ago, far fewer would make it to the point where they would have children.

(2) Health Correlations

In the environment we lived in tens of thousands of years ago, some aspects of a person’s face correlated with the likelihood of the survival of their genes, in particular ones related to disease (some diseases impact the face), genetic disorders (some of them cause facial changes), and development in the womb (where abnormal development can cause facial changes).

The correlation between health and facial features is likely to be lower now than it used to be back then. Today, a person’s facial features might still help to predict someone’s age, their most probable gender identity, and whether they have certain health conditions – but, of course, none of these give us any legitimate justification for treating some people better and others worse based just on their face.

It has been found that certain facial features do correlate with hormone levels (like testosterone). While testosterone levels may play a role in aggression (they may be part of the explanation for why men are violent so much more often than women), using these small correlations to make judgments about any one person is going to be both highly inaccurate and highly unjust. Some other personality traits may also be very weakly correlated with a person’s facial features, but talking to the person for 20 minutes will, of course, give you dramatically more information about what that person is like. Yet, we are prone to read so much into the way a person looks.

Note: there is an additional effect when it comes to faces, which is that we are sometimes taught by our culture to value certain facial attributes more than others. This can act above and beyond the previously mentioned two factors.

We humans act as though faces are incredibly important despite them being a substantially arbitrary mask our genes have programmed for us. And they often impact how we humans treat each other, despite this unequal treatment being both unjust and unjustified. If you ever notice yourself treating someone less well because of their face, take note and adjust your behavior.

I am not saying that people should, for example, date people they are not attracted to. Obviously, attraction is an important part of relationships for most people, and the face is one part of what determines attraction. (You may also care about your children one day having attractive faces, so they can more easily find life partners they like.) Rather, what I’m saying is that we should be very wary about making negative inferences about any individual person based on their face (which is something that, unfortunately, the human mind seems to do often). The face says too little about a person’s character to be useful for predicting at the level of any individual.

This essay was first written on July 2, 2020, and first appeared on this site on December 17, 2021.

We conducted a study of 999 people in the United States; recruited through our study platform at Positly.com. We looked for a correlation between 18 different personality traits (each trait being assessed with two questions) and life satisfaction. We examined the association each trait had with scores on the Satisfaction With Life Scale (a 5 question scale by Diener, Emmons, Larsen, Griffin). There are, of course, many personality traits beyond the 18 we measured!

We found that 8 of the 18 personality traits were strongly associated with life satisfaction, in the sense that the relationship held in the same direction whether measured:

(1) as simple correlations between each individual trait and life satisfaction

(2) controlling for all the other personality traits as well as income, age, gender and education (using linear regression), or

(3) controlling only for the other personality traits most strongly associated with life satisfaction (again using linear regression).

We also found a 9th personality trait with a slightly weaker association to life satisfaction; it was statistically significant, at p=0.05, using two of the three methods mentioned above.

The personality traits we found to be most closely associated with life satisfaction are listed below (ordered roughly by strength of association).

While reviewing the results, please keep a few things in mind:

A. The personality traits were reworded (to have the reverse meaning) in cases where there was a negative correlation to life satisfaction, so that all associations are positive.

B. Also, the exact cause of association is somewhat ambiguous so, for each trait, I cited three different speculative theories on why the relationship to life satisfaction may exist. For example, an association between X (e.g. a personality trait) and Y (e.g. life satisfaction) doesn’t necessarily mean that X causes Y (see this for all the meanings a correlation can have).

In the case of any particular association, it could be that:

• (a) the trait CAUSED greater life satisfaction (i.e. X->Y)
• (b) the trait is CAUSED BY greater life satisfaction (i.e. Y->X)
• (c) some other variable or variables increase the value of the trait and SEPARATELY cause greater life satisfaction. In other words, mutual exclusivity (i.e. Z->X and Z->Y).

The 9 personality traits together (in a linear regression without any other variables) explain about 23% of the variance in life satisfaction (i.e. adjusted R^2 = 0.23). Of course, we only studied 18 personality traits in this study, there are many others that exist that we did not investigate in this work.

Personality Traits Most Positively Associated with Life Satisfaction

1. At Ease – Seldom experiencing fear / Seldom worrying

Perhaps because…

• (a) High levels of anxiety cause people to feel unsatisfied with life.
• (b) People become anxious about the fact that they are unsatisfied with life (for instance, worrying because they feel like they “should” feel satisfied and that something must be wrong with them because they don’t)
• (c) Tough life situations (e.g. divorce or job loss) cause both lower life satisfaction and higher anxiety (separately).

Note: Many people learn to reduce their fear and worry. For instance, they may use Cognitive Behavioral Therapy (including Exposure Therapy for specific fears) or Mindfulness Based Stress Reduction. I think it’s likely this really does improve life satisfaction for people who don’t naturally feel “At Ease”. If you feel a lot of anxiety, fear, stress, or worry in your life, I recommend trying out our app Mind Ease.

2. Self-Valuing – Feeling of superiority / Feeling equal to others

Perhaps because…

• (a) Feeling inferior to others makes people feel bad about themselves which makes them feel unsatisfied with their lives.
• (b) Being unsatisfied with life causes people to feel inferior to others (e.g. due to blaming themselves for their unsatisfying situation).
• (c) Some commonly occurring bad life events (e.g. being harshly rejected or being put down a lot by family members) cause people both to become unsatisfied with life and to feel inferior to others.

Note: Some people learn not to view themselves as inferior to others by challenging their feelings of inferiority (e.g. using Cognitive Therapy techniques) or learning self-compassion (some have benefited from related literature like “Self-compassion” by Kristin Neff). I expect that for someone low on the trait of Self Valuing, learning to view yourself as equal to others may improve life satisfaction.

3. Warm – Laughing aloud / Expressing happy feelings

Perhaps because…

• (a) Being warm causes others to be drawn to us, this closeness improves life satisfaction.
• (b) Being satisfied with life causes people to feel more relaxed around others, which causes them to openly share positive emotions.
• (c) Having emotionally supportive caregivers growing up causes people to be both more comfortable showing their positive feelings, and more satisfied with life.

Note: People can get better at expressing their positive feelings (e.g. I have!), for instance by consciously channeling positivity when greeting people, consciously smiling when feeling happy (if it doesn’t happen automatically, which it doesn’t for some people), verbally expressing that something has made you feel good, etc. I suspect that for those low on the Warm trait these efforts might cause at least a minor boost to life satisfaction.

Personality Traits A Bit Positively Associated with Life Satisfaction

4. Unselfish – Putting others before oneself

Perhaps because…

• Being unselfish leads to better interpersonal relationships which makes people more satisfied with life.
• Being more satisfied with your life means you have less need to focus on yourself, hence, increased focus on others.
• When bad interpersonal experiences teach people that they shouldn’t trust others, they become less satisfied with life (due to worse social relationships) and become more selfish (because they don’t expect others to be unselfish toward them).

5. Forgiving – Not wanting or seeking revenge

Perhaps because…

• Harboring anger and dwelling on wrongs makes you feel bad about your life.
• Being more satisfied with your life makes it easier to minimize or forget about ways you were wronged by others.
• Growing up around forgiving people makes you both more satisfied with life as an adult and more forgiving of others.

6. Improvisational – Being adaptable or quick on your feet

Perhaps because…

• Being adaptable is often useful in the workplace, leading to increased effectiveness and work satisfaction; this leads to increased life satisfaction.
• When you are unsatisfied with life, that unhappiness causes a cognitive burden that makes it harder to be improvisational
• Negative thinking makes them less satisfied with life, and it also separately makes it harder for them to be think quickly on their feet because it’s distracting.

7. Self-Defending – Not being overly self-critical when things go wrong

Perhaps because…

• Blaming yourself for a lot of things may cause you to become unhappy about your life.
• Having a life that makes you feel satisfied may give you greater self-esteem (if you attribute that good life to your own character), which could prevent you from blaming yourself for problems.
• Being the sort of person that very rarely upsets other people could cause you to be both more satisfied with life (due to better social relationships) and less likely to attribute negative incidents to your own character.

8. Normal – Acting in a socially acceptable way

Perhaps because…

• Behaving normally may allow people to fit in better with their social groups; making them more satisfied with life.
• Being increasingly unsatisfied with life may cause people to engage in increasingly abnormal seeming behavior.
• Being a social outcast may cause people to be both less satisfied with life and less likely to care about their behaviors conforming to what others view as normal.

Personality Trait Possibly Associated with Life Satisfaction

(this final trait was not as consistently associated with life satisfaction as those above)

1. Unfeeling – being indifferent to the feelings of others

Perhaps because…

• Feeling too intensely when others suffer causes one to suffer; leading to lower life satisfaction.
• Being more satisfied with life makes people feel less need for maintaining existing social relationships causing indifference to the feelings of others
• People with a more positive disposition feel less negative emotions (causing more satisfaction with life) and less negative emotion in response to other people’s feelings (hence greater indifference).This fairly weak association could easily be a statistical fluke.

Note that there is some subjectivity in the rank order above because the strength order varied somewhat based on which method was used to measure the associations, and because there is a limit to the precision with which the associations can be measured using our data (which, naturally, depends on the sample size).

Also, since each personality trait was measured using only two questions (and, as a result, each trait was measured only superficially for each person), one might assume that these results fail to capture the true extent of the correlations. On the other hand, since we tested 18 traits and selected the strongest, a regression to the mean effect will be working in the opposite direction to some extent.

But all hope is not lost. We can effectively reason about which of our beliefs are more likely to be correct, and which are more likely to be in error. Even if we feel equally strong feelings of belief for two ideas, further considerations can make us realize that we are more likely to be correct in one of the cases than the other. In other words, there are traits that go beyond our strength of belief that can help us identify where we are likely to have made errors.

Consider the following properties that beliefs can have. Each of these is an indicator that a belief is less likely to be true.

1. Many smart, knowledgeable people disagree with you (e.g. you think that evolution didn’t happen). If many such people think you are wrong, it is not obvious why your belief is more likely to be correct than the beliefs of those who disagree.
2. You have a financial (or other) incentive to believe it (e.g. you think that the product you created really does regrow hair, and you value providing a product that helps people). When we have an incentive to think a certain way, we are less likely to seek out or listen to evidence that contradicts this way of thinking.
3. If the belief were not true you would find it psychologically disturbing (e.g. you believe that your wife does not fantasize about any other men). Our minds tend to veer away from thoughts that disturb us, making it less likely that we believe them, even when they are true.
4. You originally came to believe for reasons that don’t have much to do with logic, evidence or reason (e.g. growing up, your mom wouldn’t let you pet dogs on the street, so you believe that doing so is dangerous).
5. Your argument as to why your belief is true is long and complex (e.g. you believe that a convicted criminal is innocent, because when you evaluate the twelve pieces of evidence given against her, you find that they each don’t hold up). When our arguments are long and complex it is more likely that we have made an error at some point in our thinking.
6. There are lots of possible outcomes, and your belief is that just one of them will occur (e.g. you think Hillary will beat out the other seven candidates in this primary). Typically, the more possible outcomes there are, the less likely it will be that any particular one of them is correct.
7. A large number of factors influence whether your belief will end up being true (e.g. you’re convinced that GDP growth will decline over the next year). When many factors influence an occurrence, it is really hard to be sure that you have properly taken into account all of the important ones.
8. You don’t understand the arguments of those that disagree with you, or see how they could believe what they believe (e.g. you know that a fetus is obviously a person). When you don’t understand contrary opinions, it is an indicator that you have mainly researched one side of an issue, and so are less likely to have really weighed the strength of arguments on all sides.
9. You become emotional when people disagree with you about the belief (e.g. you think that insurance companies should not cap health expenditures for illnesses that are usually terminal, and you become upset when challenged on this issue). The problem here is that strong emotions can interfere with our ability to evaluate arguments objectively, and prevent us from engaging in open-minded discourse about a subject.
10. You can’t clearly explain what your belief means (e.g. you’re convinced that you have free will). When we find it hard to explain what we mean by one of our beliefs, it may be the case that we have merely become attached to an idea or intuition, rather than having considered the evidence and made a decision based on that.

To be good at identifying and stamping out our false beliefs, we need to go beyond just considering how strong our feeling of belief is. We need to consider the properties of our beliefs, and decide whether each is the sort of belief that we should have skepticism about.

Influences: Kathryn Schulz

• You know that “something” exists.
• If you have mental experiences, then you know that “you” exist (though coming up with a reasonable definition for what “you” means can be remarkably tricky).
• If your mental experiences are varied, then you know that whatever exists creates varied mental experiences.

With some cleverness, you may be able to add to this list a few more things that you know with total certainty, but not many. In fact, almost everything that we think we know we cannot be completely sure about. For instance:

• You take it for granted that five minutes from now you will still be alive, but there is a non-zero probability that your heart will give out before then.
• You feel as though the world you see around you is reality, and yet, that is also what it would feel like if you were in fact in an extremely detailed simulation of reality (for instance, your brain could be hooked up to some sort of computer). So you cannot know for certain that you are not now living in a simulation rather than having experiences based on physical reality itself.
• You cannot know, with total certainty, that you are not dreaming right now.
• You cannot know, absolutely, that you have not gone mad. Many things that you think are real could be delusions. Of course, you’d expect your delusions to feel totally correct to you, and not seem delusional at all.

Virtually everything involves some amount of uncertainty (even the best predictions will contain some error), and many things are highly uncertain (in the sense that our attempts to model them lead to uncertain predictions). Consider our uncertainty about processes on different physical scales.

On a large scale there are weather systems and economies, which depend on very large numbers of variables that are difficult to measure accurately. Weather depends on temperatures, pressures, geography, sun output, cloud cover, pollution, bodies of water, etc. Economies are impacted by spending behavior, regulation, war, technology, disease, culture, and so forth. What makes large systems especially hard to predict is that many of them are chaotic in nature, so small changes in the values of the variables by which they are influenced can lead to large changes down the line. For instance, a relatively small shift in the pressure and temperature in one area of the world may eventually lead to a hurricane hitting a far away area. Or, in an economy, a piece of legislation passed today that deregulates certain company behaviors could lead to increased use of leverage, which could result in a market collapse a decade later. Political events are another good example of large-scale uncertainty, because they tend to be highly complex and chaotic. That means they should be hard to predict. In fact, the empirical data says that most political forecasting “experts” are not significantly more accurate than random guessing or a non-expert predictor, though some fairly small fraction of such experts do seem to have meaningful predictive power.

On a medium scale, there are humans and roughly human-sized objects. Objects on a human scale, like balls and bookcases, tend to be quite each to predict, in part because they are accurately modeled by the relatively straightforward laws of Newtonian Mechanics. If you throw a ball you have a decent idea of where it is going to go, though the details of its trajectory will still be uncertain. If you drop a bookcase off of the Empire State Building, you know it will fall downward, and can roughly estimate when it will hit the ground, though exactly where it will land will depend on unpredictable factors involving air flow across its surface. However, predicting the location of the cracks along which it will shatter when it finally hits the ground is far beyond our ability to forecast.

Unlike balls, which are fairly straightforward, human minds are very unpredictable. The problem is three-fold: the complexity of the brain, our lack of knowledge about brains in general, and the inaccessibility of the present state of any particular brain. It can be hard enough to tell if someone you’re talking to is annoyed, engaged in a pleasant day-dream, or feeling awkward. But predicting what they will say next will be even more difficult. Brains can even act as uncertainty generators. You can give a man a physical object, like a book, which can lead him to believe some idea that he then feels compelled to spread. If he is successful enough in his proselytizing, he may spread this idea to hundreds of others, who themselves spread the idea to tens of thousands more. The large-scale fate of the world may be altered due to one particular person having seen one particular book, a result that was dependent on the great complexity of the human brain.

Knowledge of psychological studies does, of course, allow us to predict some things about human behavior. For instance, in Milgram’s famous experiment, it was found that if a person who seems sufficiently authoritative insists that a test subject administer high voltage shocks to another person, even though it seems obviously dangerous to do so, about two-thirds of the test subjects will comply. The trouble is that even robust psychological findings like this one involve much uncertainty. For one thing, it is difficult to predict in advance who the one-third of non-compliers are going to be. Furthermore, the fraction of participants that comply is influenced by a variety of factors, so one cannot assume that this two-thirds number will hold in similar situations. For instance, when the authority was not in the room with the test subject, communicating only by telephone, compliance fell from two-thirds to only about twenty percent. When other test subjects (actually actors who are confederates of the experimenter) refused to comply with the authority, only ten percent were willing to continue with the experiment.

While there are some human behaviors that can be predicted very accurately (if you touch a hot pan, I predict you will pull your hand away, if you are handed a deep-fried oreo, I predict you will eat it, if you attempt a certain extremely difficult logic puzzle, I predict you will not be able to solve it in 5 minutes). But, at this stage of our understanding, much of the human behavior that we care about predicting can only be modeled in a probabilistic fashion, with a lot of uncertainty in our estimates. We are forced to use statements like “most of the time when you put a person in situation X, they will have behavior Y.”

While one might hope that at the small-scale things become certain, this is unfortunately not the case. It is now known that atoms are governed by the laws of quantum mechanics, which involve inherent, fundamental unpredictability. In fact, in quantum mechanics it doesn’t make sense to even talk about the exact position and velocity of a particle. We can only know the probability of a particle being measured to be within a certain distance of a particular location, and having a velocity within a certain range of values. The Heisenberg Uncertainty Principle makes this idea more precise, encapsulating the idea that the more certain we are of the position of a particle, the less certain we must be of its velocity, and vice versa. As systems of particles become larger and larger (forming objects such as carrots and boulders) the uncertainty principle can be essentially ignored, because the size of the quantum uncertainty is very small compared to the size of the object itself.

Uncertainty is inescapable. It manifests in large-scale chaotic systems, like economies and weather systems, in the details of the trajectories of medium-scale objects, in human behavior, and in small-scale systems . For many of the most interesting predictions we would care to make, we cannot avoid making them in terms of probability. Rather than being able to say, “X will happen,” we end up only being able to speak in terms of “the probability that X will happen.”

To give one of many possible examples, when a group of students were asked to estimate how long their senior theses would take if everything went as poorly as it possibly could, the average estimate was about 49 days. In fact, the average time it took the students to complete these papers was about 56 days, 7 days worse than their worst case scenario estimates. Only about 30% of the students finished their projects in the amount of time they estimated. Other studies have demonstrated a similar optimistic bias on a variety of project types, from computer programming to tax form completion.

Why might we be bad at making estimates about our own projects? It is likely a combination of:

• Our tendency to plan as though the stages of a project will each go smoothly (when, in fact, one or more of these stages may have hitches).
• A self-serving bias, where we take credit for our past successes, but treat our past failures as being caused by unpredictable external events. This can lead us to have an inflated sense of our ability to complete projects.
• Our tendency to try to impress others by exaggerating how well we can perform (which becomes relevant when we are making our estimates in front of others).
• A wishful thinking bias, where our beliefs are influenced by how much we want something to be true. Since it is more pleasing to believe that a project will be completed quickly, in some cases we may be biased towards believing that.

So how can we correct this problem in our forecasting? Well, just knowing about it makes it possible for us to consciously make corrections for what are likely to be overly optimistic estimates. But even this approach often fails, as we may not adjust enough (being optimistic about the amount of bias that we have), or overact and adjust too much. Fortunately, there is a prediction method, known as Reference Class Forecasting, that has a tendency to be more reliable. At its core, this technique involves considering past cases that were similar to the project that you are now trying to make predictions about, and applying probabilistic thinking.

Rather than asking “Given what this project’s parts consist of, how long do I expect it to take?”, Reference Class Forecasting involves asking, “How long did similar projects I’ve done in the past take?” If you’ve never done a project similar to the one at hand, you can modify this question to, “Historically, how long have projects like this one taken for people with a level of skill that is similar to mine?” Once you have recalled or collected data on how long similar projects have taken, it is then easy to make estimates for how likely the project is to take different lengths of time. For instance, to estimate the probability that a project takes more than 30 days, we can just check what percentage of the time similar projects took us 30 days or more. The more data you have on similar projects, and the more similar those projects are to the project you are now doing, the less uncertain your estimates will be.

Reference Class Forecasting is useful, in part, because it gives us a way of predicting how long a project will take (or how costly it will be) that is unlikely to be influenced by our various biases. Wishful thinking, excessive optimism, and self-serving tendencies can be avoided simply by viewing our project as one among many, and thinking in terms of the probability of different outcomes. This process is certainly not perfect. For instance, it is not obvious which projects should count as “similar enough” to include in our analysis. And there will be a fundamental trade-off in this procedure between considering more past projects that are less similar, or considering fewer past projects that are more similar, and it isn’t clear what the optimal tradeoff is. But, nonetheless, this method often yields substantial improvements in prediction accuracy over other approaches.

Reference Class Forecasting can be used for many things besides planning projects. For instance:

• If you want to know how likely your friend is to cancel plans with you, consider the frequency with which they canceled in the past. For example, you can check your calendar to look at the last six times you had plans, and try to remember if they canceled on each of those occasions. If they canceled half of those times, it is quite likely they will cancel this time as well. If they didn’t cancel any of those times, then they probably won’t this time either. Of course, if you have extra information, such as that your friend has a cold and it is raining outside, you’ll want to try to make an adjustment to this probability, and conclude that your friend is less likely to show than normal.
• Suppose that you are stressing out about a test that is coming up a in a couple of weeks, and want to know how likely you are to do poorly. Well, consider your history of taking tests in the past that seemed to be of about this difficulty. If you never got a grade lower than a B on ten such exams, then it is quite unlikely that you will get a C on this next one.
• Perhaps you are feeling hopeless about finding a boyfriend, because you haven’t dated anyone for a while. Think back to your dating history, and note how long it took you to meet someone you liked in past cases. Unless something important has changed in your life that would affect your dating outcomes, this information can help you estimate how long it is likely to take you to find someone in the future. If this procedure tells you that you likely will have to wait a depressingly long time to find someone you like, consider the strategies you can use to increase your average romantic happiness.

Using the past to predict how well things will turn out in the future is certainly not an infallible method. Sometimes things change in a fundamental way such that past examples are just not relevant, or we lack knowledge about past examples. There are also fairly arbitrary decisions to be made during this process, like deciding which cases are similar enough to include. But, remarkably, using this simple procedure can often give us reasonable answers to questions that we care about, and produce predictions that may be less biased and more accurate than would typically be achieved by the methods we would have naturally relied on.

In cases where it is very important that our predictions are accurate, we can use both Reference Class Forecasting and other methods, and compare their results. When they agree, this should give us increased credence in our predictions. When they disagree, we can try to figure out why our prediction methods are diverging.

The next time you want to make a prediction, consider asking yourself, “Are there past examples similar to this case? What were the outcomes in those past cases, and how often did each of those outcomes occur?”

Consider the complete opposite of expectation adaptation: you have a single set of expectations that you hold all of your friends to. If these expectations are very high, and require that the person has a large number of positive traits, then what is likely to happen is that your friends fail at least one of these expectations from time to time. One friend will be kind and sympathetic when you have problems, but then keep you waiting for 45 minutes at a restaurant. Another, while very punctual, will often be a bit of a boring conversationalist. A third, while a lot of fun, wasn’t there for you during your recent breakup. Since your conception of friendship requires that people meet all of your different requirements, when they fail to do so you are likely to end up feeling that the terms of your friendship have been violated. This may lead to regular arguments with your friends, or feelings of disappointed, frustration and anger. As a result it is possible that you will end up with a small friend group (only people who satisfy all of your criteria quite reliably), or end up feeling regularly annoyed at the behavior of your friends.

If, on the other hand, you set your expectations for friends quite low, then the opposite sort of trouble can occur. You are willing to befriend people who don’t add much (or perhaps any) value to your life. You are likely to end up with a wide group of friends, but this wide circle may not be benefiting you very much. Whereas excessively high expectations may make you feel disappointment, excessively low expectations can lead to you feeling used or bored. And both have a tendency to produce frustration and arguments.

The ideal case for a fixed set of expectations would be if you could set them quite high, and still find a good number of people who easily exceed them. While this can work well, for many people it is unrealistically difficult to achieve. Most people have at least one significant flaw that impacts their friends.

The alternative to having fixed expectations is to have adaptive ones. You try to expect from each person what your understanding of them predicts it is realistic to expect. You ask yourself, “given what I know about this person, what do I expect them to do in this situation?” and then you expect that. You think about past situations you and your friend have been in, and anticipate that if he has canceled 40% of the plans you have made with you in the past, there is probably something like a 40% chance that he will cancel the next hangout that you schedule. If he has been unhelpful every time you have called him when you are feeling down, you start calling someone else when you are feeling down. If you have consistently found that hanging out with him is a bit boring, you assume you will be a bit bored if you choose to hang out with him.

Adapting expectations has a handful of advantages to using fixed ones. For instance, this mindset:

• Tends to reduce feelings of disappointment and frustration. If you know that Bill spends most of his time talking about his car, then when you make plans with him you are expecting this to happen, so are mentally prepared (or better yet, willing to talk about his car).
• Broadens your potential friend group compared to what you could have with high fixed expectations. Diana may be unreliable, but she really is fun to talk to. So you don’t rely on her for things that are important, but you do get together every so often for great conversation. Sally often cancels plans, so when you have plans with her, you figure out in advance what you are going to do if she cancels (e.g. pickup a movie to watch or a book you are going to read). And if you’re just not in the mood to have her cancel on you this week, you don’t make plans until next week.
• Tends to produce a mindset of problem solving. Timmy tends to be late whenever you hang out, so now you assume that this will happen and always bring a book with you, and show up a little late yourself. You assume that you could be waiting up to 40 minutes for him, and so just count on getting some reading in while you wait.
• Makes it easier to think about friendship in terms of the value, comfort, happiness and fulfillment that a person adds to your life, rather than in terms of whether they satisfy a set of abstract criteria. It might be that Don has a very serious flaw in his personality, but that overall he still adds a lot to your life. Carrie’s biggest personality flaw, on the other hand, might be such that interacting with her just makes your life worse. You can keep Don as a friend, doing everything you can to make his big flaw more tolerable so that he adds even more to your life, and choose not to be friends with Carrie. Even if Don would not satisfy the requirements a friend should theoretically meet if you had to make such a list, it is still quite beneficial having him in your life.
• Produces a realization that not every person needs to give you all things. Sue may be the best person you know for providing emotional support, Danny may be your favorite person to go out to bars with, and Annie may give excellent advice. It would be extremely hard to find a single person who is simultaneously as supportive as Sue, as good company as Danny, and as great an advice giver as Annie. But this is also unnecessary, as these three people together can play these roles in your life.

While adapting expectations can be quite a happiness generating and fulfilling view of friendship, it is worth noting that there are still some pitfalls to watch out for. For instance:

• If you become used to predicting from people what they are likely to give, it can be easy to forget that you do have some ability to modify the behavior of your friends when they are not acting in an ideal fashion. So it is important to not just predict the behavior a friend who normally have, but also, predict how likely the friend is to respond well to feedback and implement behavioral changes that you suggest. In practice, it is very hard to change people’s long-standing personality traits, though occasionally you can get them to change specific frustrating behaviors (especially if they are reflective types, who are willing to accept criticism and are motivated to become a better person). So maybe you can get Timmy to show up 15 minutes late instead of 30, if you give feedback in the right way and help him figure out how to correct his bad habit. But then again, maybe you can’t correct this behavior even if you try hard, so you might have to predict it and then adapt to it in order to avoid frustration and maintain your otherwise valuable friendship with him.
• It is also important to remember that just because you can, by managing expectations, be friends with people with many negative traits, and still find that your interactions add value to your life, that doesn’t mean that you always should be friends with such people. If you find amazing people, who meet most or all of your criteria for ideal friends, you should probably prioritize them. More generally, those people who have more of the traits that you desire and value are probably ones that you should invest more of your time and trust in.

If you are the sort of person that tends to hold friends to a strict set of criteria that you feel they must meet, consider changing the way you view friendship. Try to expect from people what they are likely to give you. Remember that people who have repeatedly behaved like X in situation Y, will likely behave like X in situation Y in the future. Ask yourself, “Given how I can expect this person to behave in the future, is he likely to add value to my life?” If so he may be worth keeping as a friend, even if he has some non-ideal traits. Then ask, “How can I adapt to the way he is likely to behave so that his negative traits become less problematic and bother me less?” Adapting expectations for friendship can allow you to be friends with more people, while experiencing less frustration and disappointment.

Our non-conscious systems primarily work using pattern recognition. Through a combination of genetic pre-programming and repeated exposure, your brain learns to label instances of things in the world as “dangerous” or “not dangerous”, “food” or “not food”, “person” or “not person”, “real smile” or “fake smile”. It learns to categorize and predict automatically.

When you see a chair you are immediately able to recognize it as a chair and associate it with your concept of chair, even if it is the largest chair you have ever seen, or the first one you’ve seen with a zebra stripe pattern. Your brain has effectively developed a way to rapidly recognize things as being “chairs” or “not chairs” based on whether they match a certain pattern. This pattern was learned automatically from exposure to many chairs in the past. You don’t have to consciously consider whether this particular object has the features necessary to make it a chair, your brain produces an answer before you are even aware of thinking about it.

To give another example, consider what happens when a chess Grandmasters looks at a chess board. In many cases, they are able to glance at a board positions and immediately (without conscious thought) identify them as being strong or weak, with good reliability. Of course, they might then choose to reason consciously about the positions in order to check or improve upon their gut responses.

Having considered these two ways of making predictions and decisions, the question then arises: when should we rely on thorough conscious analysis, and when should we use our automatic processing? To answer this, let’s consider the advantages and disadvantages of these two approaches.

Careful Conscious Analysis

Advantages

• Handles problems where multiple logical deductive steps are necessary to find a solution (e.g. I know that all A’s are B’s, and also that if something is a C it cannot be a B, therefore this particular A is not a C.)
• Allows the application of theories to problems, and in doing so lets you leverage the research and thinking of others (e.g. what does economic theory tell us will happen to soybean prices when the supply of soybeans dries up?)
• Allows you to develop theories of your own which can then be applied in future situations. (e.g. I’ve noticed that when I attack him on his left side he parries and counter attacks. So this time I will initiate a fast attack to the left, and as he begins to parry, I will drop the point of my blade to get underneath his sword, and carry the momentum into an attack to the right instead)
• Lets you compare the advantages and disadvantages of multiple possible methods or solutions. (e.g. The first theory predicts results will come out a certain way, but the second predicts a slightly different outcome. However, the first theory has a better track record on cases like this one.)
• Produces reasoning that can be communicated to others. (e.g. I came to my conclusion by evaluating the randomized controlled trials and noting that…)

Disadvantages

• Unless you have honed your truth discernment skills, you may fall prey to logical fallacies, become misled by cognitive biases, or rely on theories that have not been empirically validated. Performing truly excellent and reliable conscious analysis takes work, skill, self correction, good habits of mind, knowledge about potential pitfalls, and practice. (e.g. Due to the compatibility of your astrological charts, I think that he would make an excellent husband.)
• Thinking carefully is a slow process. (e.g. There is a deer in the road about thirty feet ahead of me and I’m going 60 miles per hour. If I hit the brake now will I have time to stop before I hit it? Will it be more or less dangerous to me if…)

Unconscious Automatic Processing

Advantages

• Operates fast (e.g. That ball he hit is going out-of-bounds so there is no need to run and try to catch it)
• Can be used in lots of situations where feedback is readily available, even when you don’t have a theory about how to solve the problem consciously (e.g. Now when I look at a cubist work I immediately know if is by Picasso, Braque or someone else, even if I’ve never seen it before.)
• Starts to get trained automatically without effort if you just do something enough times (e.g. Having listened to enough songs that I knew the titles of, I can now often guess what the title of a song is just from its lyrics)

Disadvantages

• Cannot effectively handle all types of problems. This method is especially bad for problems that require long deductive chains to solve, or where gathering various sorts of evidence together is necessary. (e.g. I know that the randomized controlled trials say otherwise, but my own experiences as a psychologist still tells me that psychodynamic therapy is more useful and cost-effective for treating depression than Cognitive Behavioral Therapy)
• May give you wrong answers when the problem changes compared to what you have been trained on (e.g. I know we’re playing squash now, but that move I did would have been perfect if we were playing tennis!)
• Requires repeated exposure, so won’t work well when you don’t get to practice a lot or see examples over and over again. (e.g. Well, I’ve only done heart surgery two times before, but from what I can tell this doesn’t quite seem to be in the right place, so I’m going to abandon the standard procedure and improvise)
• Doesn’t always provide us with conscious insight into why the prediction came out a certain way (e.g. I know that thing is designed to put food onto, but it still looks like a chair to me)

Careful analytical thought, and automatic processing relying on pattern matching are both extremely useful ways of approaching problems. When processing needs to be done quickly and you can train yourself through lots of examples with feedback, automatic processing may provide the best method for making predictions. But when chains of logical deduction are required, or evidence needs to be gathered and evaluated, or repeated exposure is not possible, or strong predictive theories exist, then as long as you have honed your thinking skills sufficiently well, conscious analysis will probably provide the best results.