Welcome. My name is David Siegel. I’ve written this essay to help young people in today’s society use the scientific method to understand the world and fix it. The world is complex, yet our minds tell us otherwise, making us overconfident and often wrong. In reality, cause and effect are difficult to determine. If you want to truly fix the world — and I assume that’s why you’re here — it would probably help if you understood what the problems are before trying to solve them. I think you’ll find things are a little more complicated than you thought before.
This page presents a curriculum that a person or a class could study to update their worldview. You could easily spend the next year on this page following all the links and learning new things. Eventually, I hope this will become a book, podcast, app, video series, documentary, meetups, conferences, and more. For now, it’s an essay © David Siegel 2019. Please do not copy this — link to inreality.show instead.
This is a curated list of links presenting up-to-date research on 150 topics. It has 20,000 words, 820 links, and 11 amazing videos. The 53 links with an * are particularly important. This page has had more than 30,000 visits.
NEWSLETTER AND BLOG NOW AT CUTTINGTHROUGHTHENOISE.NET
TIP: you can browser-search this page for any term, like “vaccines,” “bee colony collapse,” “robots,” “Bill Gates,” “grumpy,” “Mars,” “coral reefs,” or “gold,” and you’ll probably find something interesting. Skim it to get started, but don’t rush to judgment — follow the links before you make up your mind.
HOW DO WE KNOW WHAT WE THINK WE KNOW? Let’s start with the story of money, a mysterious topic to most of us. Niall Ferguson made a fantastic book and stunning documentary called The Ascent of Money. It was widely praised and viewed. The only problem is that he was mostly wrong. His narrative promoted many myths and beliefs about money that don’t fit the evidence available (I explore them below). In reality, most of our “common wisdom” fits the stories but not the facts.
Humans learn best through stories, so the best storytellers own the market in understanding the world, but they are usually the wrong people to understand a problem in the first place. Those who write books and give talks onstage are contest winners and storytellers, not statisticians and scientists.
The human mind is a miracle, and you will never see it spring more beautifully into action than when fighting against evidence that it needs to change. — David Wong, Six Harsh Truths that will Make You a Better Person*
We’ve been taking the blue pill all our lives, and now I’m going to hand you the red pill and show you how deep the rabbit hole goes. Let’s look at two quick topics and bring them into the light of reality.
The common idea that your weight equals calories in minus calories out may not be as true as we once thought,* and it hasn’t proven very useful as a regime for helping people lose weight. If it did, diets would mostly work, and they mostly don’t.
In reality, we simply don’t know. (I’d write a book called What we Don’t Know about Nutrition, but a) it would be huge, and b) no one would buy it.) Most people believe they have a nutritional mental model that fits reality, but it is based on stories not evidence. (Much more on this below.)
Understanding the Great Financial Crisis
A good way to see the storytelling effect is to look at the “common wisdom” of the Great Financial Crisis of 2008/9, an event that impacted every person on earth and destroyed a billion jobs. Almost everyone got the story wrong. Michael Lewis’s book and movie, The Big Short, was popular but completely missed the true cause and effect. So did Niall Ferguson and many experts.
In reality, two people — Kevin Erdmann, an investor and Scott Sumner, an economist — have shown that the “common wisdom” does not fit the facts. Using the scientific method and hard evidence, they show that the GFC was a result of bad reactions to scarce resources. It happened in four phases:
- A housing shortage that from 2000 to 2006 drove up home prices in both “closed access” cities and “contagion” cities and displaced workers in the US. (Not government incentives, lax lending standards, and predatory banks encouraging low-income home buyers to overleverage.)
- In early 2008, this caused US economic output to fall dramatically. The Fed — fearing inflation — set a contractionary monetary policy by not aggressively lowering interest rates and immediately injecting money into circulation. (Not a credit bubble, not an overheated economy.)
- This, along with several contractionary banking rules and policies, caused a general seizure of financial markets, resulting in foreclosures, market illiquidity, a huge drop in lending, and a general fear that “the banking system has failed.” (Not toxic debt derivatives sold by greedy investment banks — though there were some of those. Not perverse risk ratings — though there were some of those.)
- A similar reaction by most central banks imposed tight-money policies in most countries. This phase would not end until $3 trillion of quantitative easing had added enough money to the US money supply (and similar operations were conducted in Europe) to stimulate growth. Even so, the world lost about $1 trillion in human productivity forever. (Not “too big to fail” banks with inadequate balance sheets — though there was some fragility in the banking sector.)
The housing bust was not caused by too much money, too many mortgages, or too many homes. It was caused by not enough. — Kevin Erdmann
Though the Great Financial Crisis is the defining event for an entire generation (Millennials), this story hasn’t been told, and most people are confused about the true cause. The data does not support the common narrative, and that narrative still has a strong effect on policy and politics today. If we don’t understand the mistakes of the past, we are bound to repeat them (in fact, we may be about to repeat them).
Low unemployment is a very dangerous time for the US. We are a clumsy country, which fails to achieve soft landings. — Scott Sumner, The Money Illusion
If we were designing governments, institutions, and markets from scratch, we wouldn’t incorporate most of what we have today (though I would want to keep Trader Joe’s). By taking a fresh look, we can imagine building a much better world for the 21st century.
BELIEF SYSTEMS COME IN CONVENIENT, PRE-DEFINED PACKAGES, saving us the hassle of having to construct them. Here are several. You may disagree with my categories, you may have your own list, but the point is the same:
Fundamentalist: The church is the state. Literal interpretation of ancient texts, preservation of hierarchy, rules, prayer, and tradition. Jerry Falwell.
Conspiracy: Suspect the deep state and military industrial complex, paranoia, survivalism, gun ownership, tax evasion, etc. Oliver Stone.
Anarchy: Individualism. Marginalism. Let markets decide. Murray Rothbard.
Conservative: Small government, big defense industries, tradition, personal property rights, business-centric, religious values, closed borders, police enforcement, free trade, pro-life, gun ownership. Margaret Thatcher.
Fascism: The state is everything. No individual rights. Faith in “wise men” who run the state machinery. Dogma and propaganda, no free press. Donald Trump (slightly tongue-in-cheek).
Libertarian: Individual rights, non-aggression, property rights, autonomy, minimal state intervention, free press, free speech, free trade. Ron Paul.
Liberal: liberty, civil rights, secularism, free speech, social programs to support the poor, pro-choice, pro gun control, more equal distribution of wealth, unions, global warming, environmental issues. Noam Chomsky.
Neoliberal: Free markets, private enterprise, reduced government, less regulation, more competition. Ronald Reagan.
Progressive: Similar to liberal but with social and government reform. Less about unions and more about economic freedom. Bill Clinton.
Green: Environmentalism, global warming, social justice, secularism. At its most extreme, humans are seen as “the enemy of nature.” Jill Stein.
Christian democrat: social conservatism combined with market liberalism, fair competition, preservation of religious values. Angela Merkel.
Religious: power of god to control destiny, power of prayer, combination of church and state, pro-life, biblical beliefs over science. Rick Santorum.
Populist: majority rule, often nationalist, often anti-elite, often against free-trade. Donald Trump.
Socialist: collective ownership, equality, government-designed stability. Bernie Sanders.
Spiritual: magical thinking, supreme beings, hidden causes, peace, calm, kindness, “everything happens for a reason,” emotional. Mahatma Gandhi.
Minimalist: Less is more, fewer possessions, not voting is a choice, simplification, “do no harm.” [There are no famous minimalists.]
“Whole Earth:” Peace, global warming, whole grains, naturopaths, tea, herbs, organic, unshaven, balance of humans and nature. James Lovelock.
Skeptics: Skeptical empiricism, assuming that most claims are false, “extraordinary claims require extraordinary evidence.” Skeptics are negative on most future projections but, as a tribe, they have inexplicably embraced model predictions of catastrophic climate change. Michael Shermer.
Rationalist: Scientific method, Bayesian reasoning, evidence-based decisionmaking, randomly controlled trials, observe your own biases, change your view when you see new evidence, neutral. Eliezer Yudkowsky.
Separating people from platforms
These platforms are so well entrenched that most people choose a party and accept “the good with the bad.” Membership is part of their identity. A challenge to one of the beliefs of the platform is a challenge to the existence and well-being of the tribe. Tribe members must defend the entire platform — even if they suspect or know parts of it are wrong.
Someone who is a vegan atheist fiscal-conservative pro-abortion free trader who cares about the environment, supports government reform, supports gun control but not banking regulation, is skeptical about forecasts of catastrophic climate change, and supports the basic concept of income redistribution has no tribe, no allies, nowhere to call home. Most of us believe in a combination of our own “truths.” We go along with beliefs we really don’t support because changing tribes has a high social cost (unless you are Donald Trump).
Most groups are spearheaded by a small number of extremists fighting for the existence, identity, and leadership of their tribe. They are not allowed to question the platform. A huge amount of what you read comes from zealous authors who are not representative of the wider group. While this may be the only way to get a new group started, it may not be the best way to govern.
My goal is to help people decide for themselves. If you are ready to understand the world on a case-by-case basis; if you are willing to change your mind when you see new evidence and put in the work to challenge your own conclusions, you should keep reading. Please read this with your mind open to learning rather than defending your existing views. I changed my mind about several topics as I did the research for this project. Instructions for correcting me are at the end.
BELOW ARE 150 QUESTIONS AND SHORT DISCUSSIONS WITH LINKS TO RESEARCH. They are not meant as short-cuts or my “opinion,” but rather as launching points for your own research. The links are more important than the text. Feel free to skip to the section that interests you, the questions are in no particular order. Here are the categories again:
NOTE: I am a rationalist. I feel most at home in the rationalist tribe. Rationalists generally see things in a similar way, and we are happy to change our minds when we see new evidence that points in a new direction. When we disagree about something, we never “agree to disagree.”* Instead, we define a suitable test and make a friendly bet on the outcome. Whoever loses is happy to give up the money in exchange for a more accurate view of the world.
DISCLAIMER! This is not investing, health, legal, or policy advice. I give a lot of recommendations here. Some may be wrong (or wrong for you). You are responsible for your own actions and outcomes.
CHALLENGE: If you would like to try answering yourself before you see my research, I have created a question page — you can print it and work out your answers first, then come back here and compare with mine.
Should we use GDP as a measure of a society’s happiness and prosperity? It’s not perfect. But if we strive to maximize long-term, sustainable economic growth (measured by GDP), billions of people will be better off. In fact, it’s such a good indicator that I argue it’s better for most government agencies to simply focus on this one number rather than what they are doing now. If we can steadily grow GDP, we automatically improve a lot of humanitarian, well-being, societal, and environmental indicators, as we have seen in the past 40 years. Then, we would also want to factor in various externalities, so at some point we should look at using a better measure than GDP.
What is monetary policy? Central banks determine how much money is in circulation and often the inter-bank lending rate of interest. They are one hundred percent responsible for long-term inflation, and they are mostly responsible for recessions. I believe central bankers have more impact on our lives than presidents do. While most of the world has had several serious recessions, Australia hasn’t had one in almost 30 years. Most economies operate optimally at a rate of about 2 percent inflation, yet central bankers are still responsible for most recessions. Today’s policies are based on models that mostly haven’t worked. We can and should fix monetary policy.*
How can we fix monetary policy? If you don’t happen to like recessions, those of us who live in the US and UK can try to convince our central banks to implement NGDP level targeting, a policy proposal by Scott Sumner.* To improve lives of its citizens, the Fed should automate monetary policy informed by robust prediction markets and driven by continual adjustment of the money supply according to a formula based on the market predictions. The Fed should abandon control of interest rates and let the Treasury handle the balance sheet. Citizens should have electronic accounts with the central bank directly. This is called Central Bank Digital Currency (CBDC) — something many central banks are looking into (but not doing). The Bank of England should consider something similar. In this video, I explain NGDP level targeting:
What caused the Great Depression? It wasn’t consumer confidence, it wasn’t a banking failure, it was the gold standard. Gold hoarding caused deflation until Roosevelt revalued the dollar and made it illegal for US citizens to own gold in 1933. This is a fascinating story that mostly hasn’t been told but is only of interest to those who study economics.
What causes financial bubbles? I don’t believe they exist. They are a story told in reverse after a shock. Most “bubbles” — from the Dutch Tulipmania* to the Great Depression to the housing bubble — are very poorly understood by experts and the general public. Cause and effect is a story told in reverse but rarely applies again. In reasonably liquid markets, prices accurately reflect the market’s assessment of upside and downside . If traders could easily short (bet against the rise of) any asset, that would make markets even more efficient.
Does market regulation work? In general, regulation rarely achieves the stated result. The finance industry is a particularly good example. In reality, financial regulations are remarkably ineffective at catching “criminals,” yet since 2008 regulated banks have been fined more than $243 billion for committing fraud. Since 2000, Wells Fargo has been fined 93 times for committing bank fraud. Anti-Money Laundering laws don’t work, yet politicians can’t admit that. Who launders money? Banks do: 18 out of the top 20 banks in Europe have been fined for violating AML laws. The banking system isn’t failing — failure is the system. I believe regulations are mostly signaling. There may be good examples of effective regulation, but they are rare.
Does rent-control work? It works for the people who are renting those low-priced apartments! Many of them have nice second homes that they rent out or enjoy on weekends. This policy can interfere with market forces and reduce renovation and new building. Tenants argue in favor of rent control, and there may be other ways to achieve the same ends, but a thorough review of the literature concludes that it provides no net benefit.
Do minimum-wage laws work? Most of the research shows that increased wages are passed on to customers, employers hire fewer people, benefits are cut, and industry revenues are less overall. However, the debate on this worth understanding, with other studies showing different effects.*
Does technical trading work? It doesn’t. There are many chart-based and technical systems for trying to read markets. These systems are so widespread that they are already built into most prices. Going against them would have a better chance than following them. If you see candlesticks and shapes, run.
Does fundamental investing work? Not so much. Everyone has the same information, which means “the fundamentals” are already priced in, so you can’t get an edge by studying reports and analyses. Stock market investing is a game of wolves and sheep, where the wolves are those with a computing and algorithmic or data advantage, and the sheep are all the rest of us. Hedge funds are in an arms race to create profitable algorithms, and now most of those funds have reached the same plateau. Even Warren Buffett says he can’t beat the market (and hasn’t in over 20 years). Those who have an edge over the market usually make money until their edge disappears, often surprisingly quickly. The problem is skewness, which you can understand from reading this technical paper or by reading The Black Swan, by Nassim Taleb. My favorite book on investing is A Man for All Markets, by Ed Thorp.
So how should I invest? Passively. Most of us mortals can’t generate alpha, so it’s better to capture beta. That means take advantage of the average gains of various asset classes. Have exposure to global markets, real estate, and inefficient asset classes. Try to find “smart beta” opportunities. As Eliezer Yudkowsky says, “An efficient market is one where smart individuals should generally doubt that they can spot overpriced or underpriced assets.” Most markets are very efficient. If you can pick the winners ahead of time, by all means give it a try. You might get lucky.*
Should I buy or hold gold? Gold has traditionally been seen as a hedge against inflation. However, people don’t understand that the central banks cause 100 percent of long-term inflation. Central banks currently target 2 percent, and they are good at undershooting that target, making gold useless as a hedge against runaway inflation. A second fear is the total meltdown of a government or its fiat system of money. While it has happened in places like Argentina and Venezuela, only conspiracy theorists are worried about a major currency going under. Even F. A. Hayek explained that gold is a poor form of money. Real estate is generally a better, but less liquid, store of value.
Should I buy or hold cryptocurrencies? Cryptocurrencies are not money. We are decades away from any cryptocurrency becoming accepted as money. They are speculative assets. Bitcoin, for example, is much better than gold as a store of value, though it is taxed very inefficiently in most countries and regulators interfere with the market value. In general, cryptocurrencies are extremely volatile and risky. A little exposure isn’t a bad idea.
Should I invest in mutual funds? No, they add no value, they are “expensive sheep” getting eaten by the wolves. Some small number of mutual funds beat index funds each year, only to be dethroned by other lucky funds the next year. Passive investing produces the best long-term results on average; getting phenomenally lucky produces the best long-term results.
Do analysts produce fair analyses of investment products? Studies show that analysts game the system to produce the best results for themselves, not for their clients. This is true of journalists and analysts in every field. Journalists may mean well, but they care about their jobs, their careers, and hitting deadlines. They are rarely evidence-based. But some are.
How do we choose leaders in the US, and do they matter? The way we elect government officials is bizarre, people with deeper voices get elected more often, and presidents have less impact on the economy than we think. Studies show that CEOs are far less impactful on their firms than is generally believed. Presidents and CEOs tend to be much taller than average because we are biased toward thinking tall people are better leaders, but they aren’t.
What is the war on cyber crime? Is it working? Almost all cybercrime and security professionals admit: we are actively losing the war on cybercrime and it is guaranteed to get much worse in industry, finance, and government. Cyber crime has a power-law distribution, which means black swans can potentially be devastating. Regulation is not the answer. A better idea is decentralization of valuable assets. It’s very likely that the majority of funds spent to deter or punish cyber crime are wasted. Cyber-criminals have the advantage because there are so many targets to choose from, so many ways to attack, and so many non-criminals to hide among. I expect cyber crime will become a much bigger issue in the future, and blockchains can really help.
Does the income tax work? It manages to raise the money the government spends. But at what cost? In the US, the IRS spends $11 billion each year to collect $3 trillion, but this doesn’t count the cost of compliance to taxpayers and companies, which amounts to around $400 billion dollars, and another $400 billion is not collected. The US tax code is so complex that even the IRS doesn’t understand it, and it takes 6 billion hours each year to compute and pay (or not pay) it. The tax gap in Europe is on the order of $1 trillion per year. Taxing consumption and luxury aligns incentives better. One exciting proposal is the Fair Tax, which allows for a “prebate” that pays all citizens a certain amount, helping those who need it (also lets us experiment with universal basic income). Another is the Harberger scheme for valuing and taxing assets. Serious tax reform can reduce tax-collection costs by over ninety percent.
Does spending money on the military make us safe? Because military spending is highly political, and because war is a black-swan event, decisionmaking at the highest level is rarely based on cost-benefit analysis. The usual approach is to throw more money at the military and let the experts handle it. At the highest level, decisions to go to war are almost always seen later as wrong or based on faulty evidence. My guess is that most countries are wasting at least 2/3 of their military dollars. This matters to most people more than they realize, because it’s their money. I don’t have good data to support my hunches, so I need to learn more. Please send me any good studies/data.
What causes terrorists? Generally, terrorists are a reaction to bad policies. Most terrorists are not religious zealots, they are people disenfranchised by the system. One of the best books I’ve found on the topic is Root Causes of Terrorism: Myths, Reality and Ways Forward. We can generally look back 10–20 years before any terrorist event and see how a government policy or program created that terrorist. I need some help finding the data to understand this better.
Does voting work? Voting in many places is so poorly designed that the outcomes are often predictable using simple heuristics like asking kindergartners which candidate looks like “the captain of the ship.” Public choice is an active area of research with many well known principles and mechanism designs that almost no one uses. People are talking about quadratic voting to level the playing field and allocate resources using those who know the most about each topic. We could use a blockchain to run such a scheme. This is an active area of research that is far ahead of practice. It is time to overhaul voting (and reduce the need for it as much as possible).
Does representational government work? Elected officials always put themselves ahead of their jobs, creating the agency problem. As a result, most countries’ actual form of government is kleptocracy.* Running campaigns and winning elections requires a different set of skills than governing. We can now use technology to govern using new principles and mechanisms. One such principle is futarchy, where voting costs money and allocates resources efficiently. Liberal radicalism is a new idea for allocating resources. Several groups are experimenting with decentralized autonomous organizations. The ideas in the book Radical Markets* are food for thought.
Do immigrants improve our quality of life? In general, they do. Even in countries that are modern and have full employment, immigrants are usually worth bringing in because they add more than their fair share to the economy. In places like Europe and the US, with slowing and aging populations, immigration is an important driver of economic growth. We should be looking at open borders and new ideas in immigration policy. I haven’t seen any evidence-based arguments for closed borders, but I have seen emotional ones. Would it work to have a world of completely open borders?
Does management work? It’s not a straightforward answer. Certainly, some businesses are well managed. But the business press is full of fashionable gurus promoting the latest management fads. An excellent article shows that people quit jobs, not firms. The shift from top-down to flatter organizations probably hasn’t really done much to improve employee engagement or effectiveness. Self-management is an ongoing experiment. There are many management systems and anecdotes of success, yet those researchers who study failure find the same principles at work that are supposed to cause success. Some companies have found the secret recipe or developed systems that work, while others are still working on it and most others have not. If you think people work for money, watch Daniel Pink unpack the latest research:
Do boards of directors help a company? Probably not, though few researchers have studied this. CEOs choose board members who are aligned with their salary goals. Many boards are passive. They rarely prevent disaster, though some might. Studies point to long-term stock ownership rather than short-term compensation as part of the solution.
Is CEO pay justified? Here, the data is mixed. On one side, we have data showing there is no correlation between CEO pay and company performance.* A growing body of research also shows that pay for performance does not work, yet CEO salaries drift ever upward as a result of popular myths that these tall, mostly white male, lucky people are worthy leaders. On the other hand, one study showed that companies with well-compensated CEOs far outperform those with lower-paid CEOs and Tyler Cowen argues that CEOs capture less of the value they bring to corporations than other employees.* Correlation or causation? I guess it’s a little more complicated than I thought.
Should we have software patents? The patent system is under tremendous strain — another institution left behind by the 21st century. Many companies register patents defensively or as marketing requires. Some software patents are simply too short-sighted or too broad. Patent trolls are controversial. The current environment for patents does not encourage enough innovation. While there is no shortage of good ideas, patent reform is hard. Some people think patents are good for society and encourage innovation. I think patents are granted arbitrarily, there are different rules across borders, and software patents hold the world back. But — I might be wrong.
Does professional licensing benefit society? There is a rich body of literature showing that for everything from real-estate agents to insurance to wealth management to doctors to airplane pilots, professional licenses are made-up and serve no beneficial purpose. They are highly varied, inhibit innovation and automation, they make regional market for labor less responsive. The United States is over-licensed. Many studies have shown that people consistently game the system to improve their own outcomes, not the outcomes of patients, clients, and passengers. Behind every licensing body is a political lobbying organization that puts its own needs first. Licensing makes it hard for people to move to new areas when markets change. But mostly, licensing promotes a charade that licensed people are the only way to run an industry, when we have a) good examples of industries that run quite well without them and b) many examples of licensed professionals who don’t act in the best interests of their clients. Some favor reforming licensing, which is reasonable. I favor eliminating it, which is radical. I if we really need rules, we can code the correct behavior into software, not people.
How much should we trust experts? Not much. In engineering, they mostly agree. But in complex adaptive systems, there are many experts with opposing views. David Freedman has shown that experts are overconfident and often wrong. The more “important” they are, the less we look at their actual track records. Double-blind taste tests show that wine experts cannot consistently identify wines they have previously identified many times. Economists are worse.
What should we do about fake news? Recognize that it is not new. Is there such a thing as unbiased reporting? Has there ever been? “Facts” have a way of deteriorating and rarely remain factual for more than a few years. Many (probably the majority) of professional “journals” exist simply to deliver marketing messages. Can it ever be fixed? I don’t think it can. Messaging is an arms race — people will always try to game the system to tilt the playing field in their favor. Maybe the problem is truth itself.
Are diamonds really scarce? They are not scarce. To see how much they are worth, try to sell a used diamond. A diamond is simply a signal of spending money. The DeBeers company began a marketing campaign in 1938 to convince men to spend an entire paycheck on an engagement ring. It worked like a charm! Women now expect this symbol of forever and feel less worthy if they don’t get it. Diamond prices do not reflect supply and demand.
REMEMBER: THIS IS NOT MEDICAL ADVICE!
Is healthcare effective? There is no question that, around the world, basic health care is effective. I will pause to let the amazing Hans Rosling show us the good news:
Now for the bad news: most marginal (extra) health-care dollars do not bring more health. The Rand health care study extensively compared people spending different amounts on health care and found no difference in outcomes. Studies of states where health care costs differ widely for the same drugs/treatments show no difference in outcomes. The economics of health care are highly skewed, manipulated, and gamed. Doctors still don’t wash their hands! It should come as no surprise that doctors often treat according to what makes the most money, not what is best for patients. Canadians with Cystic Fibrosis live, on average, 10 years longer than Americans with the same condition. In the US, we are wasting between a quarter and half of our healthcare dollars each year, much of it on excessive administration, the most complex cases and end-of-life for elderly people, though those costs have started to come down in the last decade. Canada’s administrative fees are far lower than those in the US. Health care reform is necessary to save the US from spending ever more and getting ever less.
Is smoking really bad for you? Yes, it’s remarkably bad. Smoking causes one out of every four deaths from cardiovascular disease. Even one cigarette per day increases chances of heart disease. Taxing cigarettes reduces their use.
Should women get screened for breast cancer? If you get a positive mammogram, you are likely to be treated, and too many of those treatments are unnecessary and can lead to harm or death. This is a much-studied but very difficult topic that takes many years to see results. Mammography is the gold standard; thermography is quackery. Here are the current recommendations from Uptodate.com for women who don’t have any risk factors: Below age 40: no screening; 40–49: It’s up to you, but ask yourself and your doctor what you will do if you get a positive; 50+ they recommend screening every two years. Consult a qualified statistician before entering the medical sales funnel.
What about colonoscopy? Harriet Hall reports: “Colonoscopy cuts the incidence of colon cancer and of death from colon cancer by more than half. And it’s clearly superior to sigmoidoscopy in detecting cancers in the right side of the colon (proximal colon).” She continues: “There is still no evidence from randomized controlled trials directly comparing fecal occult blood tests (FOBT) with colonoscopy, but such trials are in progress.” Current screening recommendations: FOBT annually, sigmoidoscopy every 5 years, or colonoscopy every 10 years from age 50 to 75. I choose colonoscopy.
What are the recommendations for prostate cancer screening for men? No PSA tests, ever. If you have some risk factors for prostate cancer early in life, talk with your doctor, but if you have no risk factors, stay away from prostate tests. Don’t confuse enlarged prostate with cancer. I have written a short essay on what a scam the PSA test is.* Prostate cancer can metastasize and kill you, but it’s most likely to kill you after age 80, when any number of things can kill you (even then, it only kills one in 40). On the other hand, if you are treated, you may have to live many years with the results of the treatment, which can significantly impact your life. I assume I have prostate cancer, so I don’t need a test; I plan to live with it and die of something else.
What about hormone replacement therapy for women? The Million Women Study (2003) showed a strong increase in cancer development and recommended no HRT unless the alternative is worse. A large meta-study reported in JAMA (2002) concluded: “Benefits of HRT include prevention of osteoporotic fractures and colorectal cancer ... Harms include heart disease, stroke, thromboembolic events, breast cancer with 5 or more years of use, and cholecystitis.” The Women’s Health Initiative Study results (2002) showed that the health risks exceeded benefits. According to UpToDate: don’t use HRT to prevent disease, and treat post-menopausal symptoms with low-dose vaginal estrogen cream.
What about testosterone replacement therapy for men? This study (2000) shows benefits of using testosterone cream to improve ED and general virility. Because I’ve read The Trouble with Testosterone, I know it’s all relative, and taking testosterone can increase estrogen in an arms-race of increased dosage to get the same benefit. Using cream in strategic places seems to be the preferred treatment for both men and women who have symptoms.
Should we get regular check-ups? The studies are clear on this: They do more harm than good, both economically and medically. While there are a few outlying conditions that could be caught early in a check-up, it also puts many people on a slippery slope of medical intervention. A cost-benefit analysis says check-ups are a losing proposition. Same for dental check-ups. The rule is: don’t see a doctor unless you have symptoms you think won’t go away otherwise. The human body runs well on most fuel, without vitamins and other interventions. There are exceptions to this rule, but it’s generally a good one.
Are antibiotics a serious problem? Yes. Most antibiotics are given to livestock, get into our food supply, and cause antibiotic resistance. This is bad. New strains of bacteria have already rendered several classes of antibiotics ineffective. Antibiotics have been critical to the success of the human species, yet 80 percent now goes into animals we eat and 30 percent of prescriptions are unnecessary. The front line in this war is in hospitals, where the most potent antibiotics are creating the next generation of superbugs we won’t be able to fight later. Even antibiotics in soap could hasten the ineffectiveness of today’s antibiotics. Watch this shocking video to get a sense of how quickly germs can mutate …
In general, I recommend very light regulation. In this case, if we want our kids to have the benefits of antibiotics, we must review feeding antibiotics to livestock, prevent overprescription, ban over-the-counter antimicrobial soap, and look closely at limiting other anti-microbial products.
What can be done about arthritis? Arthritis affects a lot of people. This page at ScienceBasedMedicine explains the pathology and treatment options.* Exercise and weight loss are the first things to work on. Treatment comes in two forms: pain/inflammation management and cartilage repair. For the first category, NSAIDs (over-the-counter anti-inflammatories) can help if you tolerate them. Popular cures (soy, omega-3s, glucosamine, chondroitin, vitamins, willow, etc.) have no advantage over placebo. For the second, some cartilage-based therapies are promising but need more research. Ultimately, we will have stem-cell-based therapies, but they are decades away.
Do vitamins make us more healthy? There is a lot of good evidence that for all healthy people, vitamins are useless at best and harmful at worst. Taking a daily pill may have a placebo effect. Pregnant women should definitely get folate and may want to consider other options but read this first.* Those not planning to get pregnant should probably not take extra folate.
Don’t we need vitamin C? We do not. I don’t believe we should define scurvy as a “lack of vitamin C.” All animals need vitamin C, and most don’t get it in their diet. Most of them make vitamin C from precursors they get in their diet. Cows, cats, and cobras make their own vitamin C, and so do humans. If you eat nothing but fresh raw meat, I am quite sure you won’t get scurvy. I would define scurvy as “a lack of fresh food.” The way to get scurvy is to eat salted meats for months, and that killed more than 20 million sailors before James Lind finally figured it out.
Does having a child and breastfeeding prevent breast cancer? It seems so — earlier is better, and at least six months exclusive breastfeeding is better. To learn more about breastfeeding benefits for both mother and child, start with this summary of the evidence.
Why is testosterone going down? We know that testosterone in the US and a few other countries has been declining for decades (but not sperm counts). We don’t know why. Testosterone is relative — the absolute number doesn’t tell you very much. People suspect obesity, illness, medication, preservatives, even soy products, but no study has shown anything conclusive. Keep in mind: anything that raises estrogen has the effect of reducing testosterone, as they are relative. My personal candidate is porn. Since the early 1990s, porn has risen exponentially, just as erectile dysfunction in men under 40 has increased. Could pornography addiction be the real cause? We see more and more evidence that porn is having a very real effect on men.
Does using cell phones cause cancer? No they don’t. I’ll let Derek take you through the literature.*
Should you have your wisdom teeth removed if they don’t hurt? No.
What about metal tooth fillings? We don’t have any reason to believe that they are toxic or a danger.
What about recreational drugs? Despite decades of intervention, the demand for drugs continues to grow. Governments spend huge amounts trying to interdict agricultural products from places like Bolivia and Afghanistan, it’s a crime to sell these products, many people go to jail, and the market continues to get what it wants. The war on drugs has considerably increased the potency of most recreational drugs, and this is causing many deaths by overdose. Even drug-enforcement agents say it is a lost cause. By any measure, the war on drugs has failed and in fact is unwinnable. We should focus on education and demand, not supply.
What about addictive drugs like opioids? Opioid addiction and overdose kills hundreds of thousands worldwide each year. We now have opioids that are so concentrated, the amount in a small backpack can kill everyone in a large city. These drugs are now relatively cheap to make.* The days of coca leaves and Afghan poppy are over. We must focus on both demand and prescriptions as well as education. I believe the problem lies more with addiction than with the substances — two people can have very different addictive reactions to the same drug. Idea: make people get licenses for dangerous drugs, and if you abuse the drug (or your license) you lose your license. Use randomly controlled experiments to find new solutions.
Do osteoporosis drugs work? It’s complicated and discouraging. Diagnosis and treatment are highly variable. Several drugs have been shown to be effective at reducing fractures in frail, elderly people. But the side effects are extreme, and most patients should “take a break” from the drugs after five years. Even the recommendation to take calcium pills has now been withdrawn. Calcium probably doesn’t help anyway. The most recent “breakthrough” drug does seem to help (but also kills a small number of patients early). The cure is mostly worse than the disease! All studies are on postmenopausal women (the big market for drugs); we have almost no data on men. If breaking a bone could be fatal to you (generally, those over 85 years old), discuss your options with several doctors. For people over 40, I believe osteopenia is overdiagnosed — unless you have other risk factors, ignore it. Exposure to sunlight or UV radiation helps build bone (though taking vitamin D doesn’t) and not getting enough sunlight can be detrimental. Perhaps in ten years we’ll have better options. For now, Dr Susan Ott’s site is a good place to learn more. More on this topic in the Epilogue.
Do statins make you live longer? Only if you take them daily. Statins (cholesterol-lowering drugs) are some of the most studied drugs in history. We know from many many studies that they prolong life. Statins are cheap, have few side effects, and extend life. They aren’t for everyone — there are risk factors and side effects. Dr Rob Siegel (yes relation) says: “For patients who tolerate them, statins lower LDL and lead to health benefits that exceed what we would expect from LDL lowering alone.” If you tolerate them and you’re over 40, talk with your doctor about taking statins for the rest of your life. I do.
What about supplements and nutraceuticals? The worldwide market for supplements (vitamins, botanicals, minerals, amino acids, enzymes, etc) is already at $240 billion and growing fast. Believe it or not, a few of them actually work. Saw Palmetto has been shown to reduce symptoms of enlarged prostate. St. John’s wort is better than placebo (but not necessarily drugs) for treating mild depression, BUT St. John’s wort interacts negatively with many prescription medications. Green tea shows maybe some benefit, but even if the tea is strong, the evidence is weak. A few other supplements may turn out to be helpful and safe.* Most of the rest do nothing but help the companies that sell the products. Echinacea hasn’t been shown to prevent or reduce cold symptoms. Unfortunately, more than half of pharmacists recommend nutraceuticals. Most supplements do nothing, some cause real harm, a few work, and a surprising number are tainted with actual prescription drugs. Should this industry be regulated? What do you think?
What about herbal remedies? Three out of four Americans takes a supplement regularly, and they cause 23,000 emergency-room visits each year in the US. Herbal remedies are unregulated drugs. They can interact with each other or any medication you may be taking. Americans buy $8 billion of herbals (top of the list: horehound), but are they better off? Unfortunately, people take a natural pill, they feel better, they tell themselves a story of causation, then they tell their friends, and the news of a new cure spreads like wildfire. Entrepreneurs rush to fill the market need. Yet whenever scientists try to evaluate these drugs they find no effectiveness. Could it be that skepticism turns off the magic healing power of herbal remedies?
What about cannabinoids? Medical marijuana and cannabinoids do have some effects. However, there is much more enthusiasm for sales than for double-blind, placebo-controlled data. For epilepsy, researchers have found significant benefit, but for relief of anxiety, pain, nausea, and other symptoms, they haven’t seen any real signals yet. Steve Novella has a good summary, showing that we are just at the beginning to understand what the benefits and harms may be.
What about homeopathy? As placebos go, I highly recommend homeopathy, because it’s nothing but water. It has no side effects (unless the water is infected with bacteria — you have to watch quality control when you’re bottling water). On a cost/benefit basis, it’s about the best way to engage the power of placebos. The packaging, the seriousness with which the licensed naturopath chooses your remedy, the instructions, the price — all of this makes for the perfect placebo treatment. And placebos work amazingly well. It has to be at least as effective as (and safer than) much of the quack medicine practiced in hospitals and doctor’s offices. If it’s probably going to go away anyway — and most symptoms do — see your local homeopathist to make it go away faster!
What about chiropractic manipulation? There is nothing but anecdotal evidence for chiropractors and their charade of practicing “medicine.” There is no strong evidence in favor of spinal manipulation to cure various conditions. The cure is often worse than the disease, and chiropractic pediatrics should probably be outlawed. If you love your chiropractor, please click on and read those articles before you contact me to tell me your story.
What about eating grapefruit while on certain drugs? It turns out that grapefruit juice does interfere with the actions of many drugs. Your digestive tract has an enzyme called cytochrome P450 3A4 that’s responsible for the inactivation of about half of all oral drugs. Knowing this, drug companies give you more of those drugs than you can handle. Furanocoumarins — the active ingredients in grapefruit and a few other non-sweet citrus fruits — render the enzyme ineffective, which gives you an overdose of the drug. See this list of drugs affected. For some drugs, like statins, you would have to drink a lot of grapefruit juice to interfere, but other drugs are much more sensitive. If you drink a glass of grapefruit juice even once a week, always ask the doctor when getting a new prescription. A friend of mine says, “Wash your statins down with grapefruit juice — that’s something James Dean would have done.”
THIS IS NOT MEDICAL ADVICE, DIAGNOSIS, OR TREATMENT!
How much of a problem is obesity? Is obesity really bad for you? It’s a complex question, since other lifestyle and genetic factors could be more influential. Obesity probably contributes marginally to many diseases but is not the single driver of poor health. Being overweight is protective of certain diseases. Reducing obesity definitely improves health, but not by as much as you may think. One well designed meta-study showed that mildly overweight people outlive people with lower body-mass indexes.
Why are adults and children so much fatter than 40 years ago? People are definitely getting fatter, it causes more disease, and we don’t know why it’s happening. There are many interesting hypotheses and none of them has emerged as the clear cause. Could it be portion sizes? Urbanization? Gestational? Genetics? Carbohydrates? Breastfeeding? Ethnicity? Watching TV? Sugar may play a role but probably not as much as people think. It’s likely a combination and that genetic differences complicate the situation. We have more questions than answers.
What is a healthy diet? There’s no such thing as a “healthy diet” — research here is very poor, with very few claims verified. What works for one person does not work for all, there are many confounders and a lack of good studies. Bill and Melinda Gates, who want to help us understand what a good diet is, are wasting their money on rigorous nonsense. Michael Pollan’s advice to “Eat food, not too much, mostly plants” sounds good to me. Steve Novella says “Balanced macronutrients from a varied diet is likely best.” One recipe doesn’t work for everyone; we should treat different people differently.*
Are some people gluten intolerant? The market for gluten-free foods is over $4 billion, yet there is almost no real evidence for non-celiac gluten sensitivity.* It’s likely that most people who believe they are gluten intolerant aren’t, and a few are potentially pre-celiac or somehow have mild symptoms of celiac disease. If you believe you are gluten sensitive, then either you have some degree of celiac disease or you have something else you are not treating, and that could be dangerous. I have thought about how to make a home-test kit, but I’m not sure the market exists — people are willing to spend on gluten-free food but not on a test.
Do we need to get enough protein? We have no idea what “enough protein” is for humans. We only know a few things: 1) people who are undernourished need food, which includes protein, 2) most diets include 12–20 percent protein, it’s hard to get less or more than that, 3) Even weightlifters don’t need more protein than the general population, 4) people who eat low-protein diets may have more osteoporosis/fractures. This Harvard article gives a decent overview. This Stanford study (2019) shows that Americans get too much protein at the expense of the environment. One thing that may help: there aren’t any symptoms for not getting enough protein (nothing to do with being tired, headaches, weak nails or hair, etc). The symptom would be hunger. With only very rare exceptions, chronic symptoms are caused by other things. There is very little data on protein requirements for humans. We need more study in this area. See the epilogue.
Are trans-fats bad for us? Yes. Several studies confirm that margarine made with trans fats contributes to heart disease. One of the few solid findings in nutrition science!
What about whole grains, fresh fruits, vegetables, fiber, salad? There is more and more evidence that eating whole grains helps avoid some cancers and even other diseases. Some recent research points to more health benefits. But this research is difficult — you can’t hold everything else constant. Eating a lot of fiber probably reduces your risk of colorectal cancer. Consuming plenty of olive oil is almost certainly good for your heart (though we don’t really know why). Will you live longer if you eat brown, rather than white, rice all your life? My guess: statistically, those who eat whole grains all their lives may live “a bit” longer than others. I plan to die “a bit” early as a result.
Most studies are observational, relying on food diaries or the shaky memories of participants. There are many such studies, with over a hundred thousand people assessed for carbohydrate consumption, or fiber, salt, or artificial sweeteners, and the best we can say is that there might be an association, not anything about cause and effect. … the field has been undermined by the food industry, which tries to exert influence over the research it funds. — Eric Topol, The AI Diet
What about vegetarians and vegans? I’ve been vegan for 35 years. I am in good shape and people say I look ten years younger. Research says there’s no health difference between vegetarians and the rest of the population and that vegans maybe need to get some B12 and a few other supplements (we really don’t know). The most recent metastudy (not a huge amount of data, but probably the best we have to-date) shows a few general benefits to being vegan and no negative effects. There is some evidence that vegans have lower bone density and mixed evidence for higher fracture rates: no fractures for Asians (2012), yes fractures in elderly patients (2009. All this evidence is weak at best. This Stanford study shows that most people get too much protein and says: “Vegetarian and even vegan diets typically contain adequate amounts of protein, including adequate amounts of all 20 amino acids …” I have been “mostly vegan” for 35 years. I hope there is more study of these diets. More on my journey in the Epilogue.
Does drinking sugary drinks make people fat? Maybe not (2008). Or maybe yes (2010). Or maybe yes (2013). My guess: drinking a lot of sugary drinks definitely makes some people fat. For most people, the effect is probably less dramatic.
Do diets work? Almost all of them do. For a few weeks or months, sometimes as long as a year. And then, almost all of them fail. The weight comes back (though diet and exercise are better than either alone). The problem is the endocrine system, which is trying to go back to the “normal” weight from before, though not always. Restricting calories in the long term does not look like the key to losing weight.* Increasing exercise tends to be an arms race — you need to exercise more and more just to not gain the weight back. While some people get out of this trap, at least ninety percent do not. Here are the results of a two-year study showing overweight people losing 4kg on average (with no control group). What does this tell you?
Do low-fat diets work? It was once believed that saturated fats cause heart disease. We now know it’s more complicated than that. In general, the stigma against fats is gone, and eating fat in moderation is part of a healthy diet for most people. In 2005, we thought low-carb diets actually were better at losing and keeping weight off. A 2006 large study confirmed it, and a 2012 meta-analysis hinted in that direction. Today, we know less than we knew then. The trend now is away from low-fat (high carb) diets. Since the 1980s, we have known that unrefined carbohydrates are absorbed differently from highly refined grains and starches. The “glycemic index” probably helps steer people toward healthy options, but I expect it works better for some than others.
Do low-carb diets work? Low-carb diets may be slightly preferable for many people, but in studies up to one year, they still produce no better results than any other diet (see Chris Gardner’s video summarizing his study*). For the long term, we don’t really know. It’s very expensive to study people carefully for even three months. We definitely need to keep looking at this, because if low-carb diets really worked, we would see overwhelming evidence of that, and we don’t. (I hope John and Laura Arnold are funding more work in this area.) My guess is that we need to understand different genetic factors of glycemic response* before we can prescribe a particular weight-loss regimen.
What about a ketogenic diet? Nope. This diet may work for people with diabetes, but it doesn’t work for most.
What about the paleo diet? I think people just made up the idea of a paleo diet, since actual hunter gatherers had a wide variety of diets. We have almost zero evidence that the diet promoted in a popular book is beneficial.
What about fish oil and omega-3 fatty acids? We thought for a few decades that antioxidants would help people live longer, but it was really never more than a hunch. You can see the benefit in a petri dish but not in studies of human health. At this point, we don’t have any evidence that they are beneficial. If you’re taking them, read Steve Novella’s update.
Is a Mediterranean diet good for you? Probably, but it depends on several factors. It seems that olive oil is good for you, or a balance of carbs and fats is good, or working outside most of your life, or a glass of wine each day (or not (2018), or not! (2012), or some genetic component is really driving the results. It does seem to be better for your heart than other diets tested. A growing number of small signals points more and more to the Mediterranean diet as a pretty good way to eat, but it’s not a slam dunk and at least one carefully controlled trial showed no health benefit. Should all Chinese people should start eating lentils and ciabatta with olive oil? Maybe not, but Australians and Americans would likely be better off.
What about saturated fat? Prevailing wisdom, circa 1990: that steak is a widowmaker! In 2010: saturated fat does not cause heart disease. 2017: this excellent article concludes that saturated fat likely is a contributing factor to heart disease. A 2018 study drew the same conclusion. So the balance may be tipping again toward reducing saturated fat (but taking statins is probably more important). If you don’t have other risk factors and do eat a lot of saturated fat (remember, coconut oil is saturated), read the 2017 piece.
What about cholesterol? It’s clear that cholesterol is a driver of heart disease. Eating eggs raises cholesterol. The average American eats about 5–6 eggs a week. If your egg consumption is much higher, the current recommendation is to bring it down to normal. The advice on dietary cholesterol and outcomes (heart attacks, stroke, death) is not based on very solid evidence. Some researchers say take statins even if your cholesterol is low; others say use statins only if diet and exercise don’t lower your LDL numbers enough. Recent research suggests that some people may be “cholesterol sensitive.” This study shows the variability of individual response (see fig 4). I hope we’ll learn more about that in the future.
Does exercise help people lose weight? A bit. Not much. If it were a clear win, we would know. Some people do start an exercise program and manage to keep the weight off. I don’t think we can explain how it works for them but not for almost everyone else. Regardless, exercise helps make people more healthy, even if they don’t lose weight.
Does surgery help people lose weight? For obese and morbidly obese patients, surgery is far more effective than any other intervention. Here is the excellent story version.*
Is fasting beneficial? Can it make us live longer? Maybe. I expect that humans born in a hundred years will live much longer than we do today. Not because of fasting, but because we will learn how to cure ageing (possibly sooner than we had thought, watch this). We’ve known for decades that calorie restriction for lab animals lengthens their lives. But in humans, how will we know what works? A two-year study showed improved biomarkers for non-obese participants, but don’t think biomarkers are significant — what counts is longevity. It can’t hurt, but calorie restriction is unlikely to extend life very much. I started eating much less about 8 months ago and love it. I don’t fast, but I do go hungry. I don’t care if it lengthens my life; it increases the quality of my life tremendously.
Which foods cause cancer? We really have very few clues here, because cause and effect is so difficult to tease out of so many different variables. Processed meats very likely do contribute to colorectal cancer in some people, and well-done meat shows similar potential to cause cancer. Char-broiled meat probably has some carcinogenic effect. We might say the same for simple carbohydrates as well. Problem is that cancer takes a long time, and there are many potential confounders in the environment, genetics, and other factors. We can’t expect more information on this without long-term controlled trials, which are unlikely (Laura and John Arnold — can you help here?).
Does taking drugs or drinking while pregnant cause birth defects? We’ll probably never know, because the randomly controlled trial necessary to learn this is unethical. In most cases, it could easily be a case of anecdotal correlation, not causation. A woman who drinks often may not take folate and that increases the risk of birth defects. Yet in extreme cases, there is very likely a higher risk of damage to the child, whether in the form of birth defects or dependencies or various forms of mental health. I don’t think we’ll ever know if drinking a glass of wine a day has any effect — I suspect it doesn’t.
What about eating organic food? There is no measurable benefit to eating organic other than it costs more and makes money for organic producers. The Stanford study is the gold standard here, but it has been challenged. Meta-analyses (2012) show no health benefit to an organic diet. Organic food is almost certainly better for the environment, but whenever I say “almost certainly” I am usually wrong, so I’m looking for the data on this, too.
Which parenting paradigm is best? In my experience, bookstores are filled with two kinds of books: those that advocate parents to better understand the world of the child and those that give parents tools to bring children into alignment with the adult world. In practice, I don’t think it’s possible to stay completely on one side of that line. While I’m in favor of letting kids be kids and understanding their context, I get frustrated too easily and overcorrect. I know a lot of other parents do, too. Po Bronson and Ashley Merryman’s book, Nurture Shock, was the first evidence-based book on parenting. Emily Oster, a data-driven mother and author, has written a data-driven book* and a good article on parenting (and one on pregnancy). My suggestion: read Alfie Kohn’s books, subscribe to his newsletter, and try your best. I recommend making a deal with your kids that if you screw up and punish them, then you should get punished too. One more thing: if your kids see the back of your phone, you will soon enough see the back of theirs.
Should babies be breastfed? Babies benefit from breastfeeding, but for how long? This metastudy concludes that six months is a good minimum, but it’s likely that three months gives most of the benefit. And formula isn’t the worst thing. I was a formula baby, and look how grumpy I turned out!
Should we circumcise boys? There is no scientific evidence that we should. A small percentage of males are harmed irreparably in the name of religion and fashion. While it’s not practical to make it illegal, I do wish it would not be forced on boys but rather let adults make their own decision for themselves.
Is fluoride a public health concern? There is no basis for fluoride alarmism. Fluoridated water has been an effective public health intervention. The fluoride scare is a form of conspiracy theory, not based on scientific evidence. Fluoridated water = good policy.
Should newborn babies get vitamin-K shots? I’m surprised to say the answer is yes. Normally, I don’t believe in medical intervention with the birth process, but here the evidence is clear. It may not help your child, but it helps a small percentage of children, and we can’t tell who ahead of time — it’s easier to give the shot than to do the test.
Do vaccines actually work? Yes, vaccines work, they save lives. Arguments against vaccines are motivated by anecdotes, tribal belief systems, and misinformation. If you believe vaccines are harmful, you must address the outcome of the meta-study linked above and bring solid evidence to your argument. Measles outbreaks are becoming more common. We have to fight magical thinking* with facts and change our beliefs when the facts point in a different direction than we thought.
Does drinking milk build strong bones? Why should drinking cow’s milk be good for humans? We don’t have any evidence that drinking cow’s milk reduces hip fractures, which is how we diagnose osteoporosis. Also no signal against heart disease. If drinking milk helped build strong bones, then Asians and people in many other places that don’t drink milk would have weaker bones and more fractures. Do they? No (2005). No (2015). For children, it appears that drinking milk does improve bone mass, but that doesn’t mean they have fewer fractures when they are older, they don’t (2014). Milk gives you calories, but billions of people don’t drink it and they are fine.
Does playing violent video games make kid more violent, depressed, or antisocial? It’s a reasonable hypothesis, but you have to test it in the real world, not in your mind. In reality, suicide rates have increased in the US over the past few decades, but as Nir Eyal explains, it’s not safe to assume that screens are the cause …
How can we bring out children’s intrinsic motivation? Not by trading rewards for behavior. Maybe it isn’t about behavior at all. Punishing (usually called “consequences”) doesn’t work. Praise doesn’t work. It turns out that being there for your kids and loving them unconditionally actually works. It’s just hard to do.
Do horoscopes and astrology work? No. People who believe in astrology are confusing coincidence with cause and effect. Example: we know that children born in the first quarter of the year often outperform their peers, simply because they miss the age cut-off for school and end up as the oldest in their class, which gives them an advantage. This does not mean that Aquarius are smarter or better at sports than January Capricorns! It means that December Capricorns (youngest in class) are generally behind their January Capricorn classmates (oldest). Correlation is not causation.
What about afterlife and paranormal skills? There are many scientific studies showing that various people have paranormal skills. These studies are all — 100 percent — poorly done and cherrypicking the data. There are logical explanations for all the “amazing results” of astrology, of ghost sightings, for all UFOs, mediumship, and more. When we don’t understand something, our imaginations jump to fill in the gaps.
Do psychological drugs work? Some do, but most we have no idea. The vast majority of psych drugs are based on experiments that have not been confirmed, and the p-values cluster dangerously under 0.05.* This is the crisis of reproducibility.
Do people fear losses more than they celebrate gains? This is the principle behind loss aversion, a central pillar of prospect theory. Daniel Kahneman relied extensively on this theory for many conclusions in his work on behavioral economics. As we often learn, “it’s a little more complicated than that” — a recent paper titled The Loss of Loss Aversion challenges this theory and claims that most people give losses and gains roughly equal weight.*
What about choice overload? We’ve heard in books and TED talks that people don’t want too many choices, they have to be nudged, they seize up when presented with too many choices, and that they buy more when they choose from fewer options. That body of work was based on a tiny number of experiments. A large metastudy shows that overload is a function of choice-set complexity, decision task difficulty, preference uncertainty, and decision goal, and these factors make a simple model impossible. It depends a lot on context. In many cases, consumers have no problem with a lot of options.
Can people’s personality be characterized by a single test? Many tests claim to provide a profile of a person‘s personality: Myers Briggs, Jungian archetypes, Freudian analysis, Brainspotting, the marshmallow test, IQ, Activity Vector, DISC, etc. Most are rigorous nonsense. One test that seems to be holding up best is the “Big Five” personality trait test. There hasn’t been much research on it in the past twenty years, but there are some intriguing signs that there may be something there. I am interested to learn more.
How effective are police forces? Policing is subject to fads from various researchers, authors, and magazines. Policing suffers from “cost sickness” — police forces can’t keep up with the demand for their services, they are chronically decades behind. Research shows little evidence that punishment of low-level offenses reduces serious crime — and considerable evidence that it disproportionately harms communities of color. People who are disenfranchised, desperately poor, substance-addicted, and/or mentally ill take up much of a modern police force’s time. Arresting, charging, sentencing, and incarcerating these people is one of the least effective responses. Arnold Ventures is leading the way here.
How effective is the trial-by-jury process? Remarkably ineffective at determining who the true perpetrator of a crime is. Juries have no training in decision science, are not statistically literate, and are easily swayed by emotional arguments. It’s complicated, and the nuances are often beyond a jury’s capability to put into perspective. Some studies have shown that jury trials are largely effective. However, jury trials are very expensive, and a series of academic studies show that the court process is highly flawed. Sentencing is also extremely variable.
Does gun control work? Yes, gun control can save many lives. The US has one of the least effective gun policies in the world and the results to show for it. The gun lobby prevents evidence-based policy. It’s a complex problem, but gun laws, policies, and new technology can help.
Does prison work? The rate of incarceration in the US is unforgivable. The US leads the world in putting its citizens in jail. Prison is generally a) a poor deterrent and b) a cycle that, once started, tends to repeat. A randomly controlled trial in California showed that the amount of time people spend in prison does not correlate with how well they do once back outside. Parole boards don’t work. Finding what really works is very complex. The incentives are misaligned. Mental illness is the elephant in the room.
Do preschool programs work? Randomly controlled trials have shown that Head Start, a preschool development program, does not work better than no program. However, “no program” is probably worse for children living with bad parents. I believe other preschool programs have been found similarly lacking, but good research is showing the way forward.
Does testing students work? Absolutely not. Bryan Caplan explains: students who are good at taking tests are more obedient workers, but they don’t add more value than poor test takers. Teachers teach to tests and students learn to pass tests, nothing to do with real life after school. Grades only give a signal to employers who think degrees and grades matter. There are much better ways to design a curriculum.
Do small schools have better students than large schools? Bill Gates thought so. He gave more than $2 billion to the “Small Schools Movement,” until they realized that larger schools outperform small schools on almost all measures (we measure “quality” with test scores, which is silly). In reality, large schools have more resources and give kids more choices. I would ask Bill and Melinda Gates: What is school really for? How can we help society move away from tests and signals and toward more productive, fulfilling lives?
Does homework help kids learn? You would think so, but Alfie Kohn shows evidence that homework does not help. Perhaps if it prevents kids from doing things that would harm them, but there is very little academic benefit.
Does a 4-year college education help? For the vast majority of people going through it, the answer is: the education probably doesn’t, but the degree definitely does. We see an “education premium” among people with degrees in many areas (even bartenders with degrees make more money than those without), but I believe the effect is the degree (signal), not the education. The case against secondary education for most people* is remarkably strong, especially considering the debt incurred. Companies prefer to hire college graduates as a filtering mechanism. More and more, the best employees will not have college degrees. Solutions? Peter Diamandis has some good ideas. Xavier Niel’s 42, Udacity’s nanodegrees, Lambda School, Praxis, Sudbury schools, and Khan Academy are good places to start. I agree with Tyler Cowen that to counteract automation, we will need to learn continuously.
Do graduate degrees help? In hard sciences, graduate degrees have a strong track record. In business, psychology, economics, creative industries, and other soft sciences, the evidence is far less clear. Their major function is signaling to employers. The MBA in particular has no measurable effect on business. Jeff Pfeffer knows this better than anyone — he writes his brave reports from inside the machine. Station 1 may show the way forward.
Is research effective? Most of the findings published by the world’s most respected journals is wrong*. Research, funding, and publishing are all heavily biased and unlikely to help us learn much. Most research is statistically underpowered and unverified. Most research is gamed to produce the best career outcomes for researchers. Most research on animals is conducted on a single species of mouse that is very well understood but has little to do with human health. Derek Muller shows you in 12 minutes:
Is research funding effective? Many studies are well done and unbiased, yet many are directly or indirectly affected by their source of funding. Getting funding is as much a beauty/popularity contest as it is about the merits of the research. Funders either want a particular result ahead of time (especially in pharmaceuticals, where research costs are so high), or they want to be associated with the “big names” in research, they love being associated with spectacular newsworthy results. The majority of funded research projects should probably never take place, because they don’t have the statistical power to produce significant results, regardless of the outcome.
Is research publishing effective? Publishing scientific results has long been a game. Believe it or not, this is well documented in the published literature! The “big name” publishers continue to release a stream of sensational results by famous researchers, and the concept of peer review has not been relevant for decades. Peer review is not the gold standard — it is often gamed. Moreover, the writing is incredibly poor, papers are not machine readable (they could be), and the data is difficult for third parties to get and work with. This is a critical horizontal layer for science, and it is broken. We have the ideas to fix publishing— the problem is entrenched business models. I think one billionaire could make a difference here.
What does functional MRI imaging tell us about the brain? My guess is that 90 percent of fMRI neuroscience is rigorous nonsense. Signals in neurons travel at a velocity proportional to that of electricity in wires, while neuro-imaging measures blood flowing around the brain — a huge difference. Expensive machines around the world show regions of the brain “lighting up” under various stimuli, providing “clues” about how we think. But those clues are often statistically underpowered and most likely as accurate as astrology in determining cause and effect. On the other hand, researchers are always looking for ways to improve, and — even though some of us may be surprised — this field may eventually bring us closer to understanding ourselves.
Are fluorescent or LED light bulbs more efficient than incandescent? Yes. It was right to phase out incandescent bulbs. LEDs are better than fluorescent are better than incandescent. However, a new kind of incandescent bulb design may prove even better, and more breakthroughs are on the way.
Who is poor? Fewer are living on less than $2 a day than we think. A lot has changed in the past 20 years. Today, the middle class is rising around the world, even as the divide between rich and poor widens. As the rich are getting much more wealthy, standards of living continue to rise across all continents. Our common view is of a huge gap between rich and poor, but in reality, the rise of the middle class is the big story. Many things we expect to work don’t actually work. Smart, randomly controlled trials are showing us how to help poor people make their way toward the middle class.
Should we be concerned about growing human population? Paul Ehrlich famously predicted an apocalyptic world as a result of exponential human population growth. But his “population bomb” fizzled because the world can handle more people than we thought, affluence is the best birth control, and technology and innovation continue to raise the standard of living for everyone. The structure of our population growth this century is likely not only tolerable but desirable. I will let Hans amaze you:
Are robots and Artificial Intelligence stealing our jobs? Not in the next 30 years, no. History repeatedly shows that as technology replaces repetitive work, more jobs are created (technological re-employment*). So far, technological unemployment is largely a myth. We are in the middle of a long trend of wage polarization*, where middle-paying non-college jobs are increasingly being replaced, while wages for low-paying and high-paying jobs have increased. This is a complex topic with many surprising twists and turns.* I believe we are now (2019) at “peak office worker” for all of human history. I expect the next wave to be service jobs, which may peak around 2040, after which I think it will be all about mass replacement and niche specialization until early 2100s, when I expect most jobs will simply be done by machines.
Is our privacy really at risk? As long as the Internet is free, people will leave trails that companies like Google pick up, use, sell, and resell. Did you know that every prescription you’ve ever filled is available for marketers to buy? Privacy laws in Europe are hurting, not helping — they are too heavy to be sustainable. Privacy laws in the US can never keep up with giant technology companies that know every purchase, every web site, every search term, every mouse click. China is compiling a “social credit score” on its citizens that started out as a local incentive to behave well but is now digital, complex, and disturbing*. Algorithms are getting more and more sophisticated. “Free and easy” is expensive and limits your choices. My company is trying to change that, but we will need better protection for consumers.*
Does the art market price works fairly? The art market is almost entirely arbitrary. Fame and prices are controlled by a small number of people. There is really no such thing as “good” art, it’s just another business. Too many young artists think that they will make a living as an artist, when market insiders cultivate a small number of early career “darlings” to invest in, canonize, and profit from.
Does microcredit help the poor? It doesn’t look like it. Randomized controlled trials have shown very little benefit to borrowers, while lenders do fairly well. Trying to help people effectively is hard.
Does social capitalism work? Probably in some cases, but in reality we know at least one experiment that didn’t: After giving away over 60 million pairs of shoes to kids in developing countries, Tom’s Shoes conducted a randomized controlled trial and measured the results between kids who received shoes and kids who didn’t. Surprisingly, foot health was not improved, outdoor sports participation increased by a small amount, and children who received free shoes generally believed that foreigners should send them more of the things they need. The company now uses randomly controlled trials to measure the effect of all their giving (they found giving sports shoes as a reward for kids who do well in school has a remarkably positive impact on school performance, so that’s their new strategy — we’ll see how it turns out). Read Randomistas* for the whole story.
Does marriage work? Marriage has long been a political institution, subject to government by the majority. Why, exactly, can’t same-sex people or three people get married? Certainly, some people are very invested in the state-sponsored institution of marriage. I propose that we have the problem backward: why get married at all? Marriage is not necessary to require parents pay for and take care of their children. All countries already have laws for that. In the US, both marriages and divorces are down significantly, showing lower demand for state-approved marriage. Given that so many marriages end in divorce, I favor getting rid of state-recognized relationship licenses. Marriage should be a private contract between two parties.
Does God exist? Do you mean there’s a white man with a beard floating in the clouds or in space, even though white people evolved only about 20,000 years ago? Or do you believe in a higher power? I’ll let Richard Dawkins, Christopher Hitchens, and Eliezer take it from here. While plenty of people will disagree, I believe organized religion does more harm than good and works first and foremost to protect those at the top of the hierarchy (the agency problem again). Scientists who believe in god simply haven’t put enough work into deconstructing their own internal conflicts. Studies on happiness and life satisfaction are mixed but interesting. Whatever you believe, is there an amount of evidence that would change your mind? I walked away from organized religion at age 13 and never looked back. If I am struck by a bolt of lightning and killed, it will simply be because I was unlucky.
I don’t want to believe. I want to know. — Carl Sagan
Can a good player win at poker? Not against other players at the same level of skill. When someone wins a big tournament, he/she is hailed as the big new talent, but only one person has repeated as world champion since 1981, and the players themselves say that at the top tables it’s impossible to predict the winners. These days, top players bet on each other to spread their risk.
Is it a good idea to buy lottery tickets? No. Not because the odds are against you, but because the payoff isn’t big enough to give a positive expected value. Never in history has a fair US or European lottery had good enough odds to warrant buying a ticket. They disproportionately take money from the poor. Governments should eliminate public lotteries and heavily tax private lotteries.
Are lottery winners happier? Above a certain amount, yes, in general. Most lottery winners lead fairly similar lives to their previous lives and adjust well to their increased balance sheets. It is a myth that most big lottery winners end up penniless.
Is Social Security in the US at risk of failing? Yes. There isn’t enough money in the trust to pay future retirees. The longer we kick the problem down the road, the worse it gets. The biggest problem facing social benefits schemes around the world is the population structure* (you must click on the blue curve to see changes over time). Japan and the United States will almost certainly have to reduce benefits. People will have to work longer. In Europe, it helps to look at the incentives. In my view, Social Security wasn’t such a great idea the way they implemented it (today’s workers pay for today’s retirees). In other countries, I don’t know but am interested to learn.
Are GMOs dangerous? In general, they are not. However, there is black-swan potential here, and public perception is quite polarized. Because you can’t “put them back into the bag,” it makes sense to craft policies on genetically modified crops that help prevent worst-case scenarios. Keep in mind there is potential for harm without GMOs (the “Alar scare” was a misunderstanding followed by an aggressive media campaign, and there are “bad” molecules in the natural environment that haven’t managed to take over the earth). In my view, applying the “precautionary principle” is misguided. I think cost-benefit is a better approach. I believe we should look at each case on its own merits, and the Biofortified blog is a good source of that information.
IF YOU HAVE NOT BEEN SKIMMING TO THIS POINT, you must have seen many examples where your “common sense” notions or “the common narrative” has not held up very well to data-driven scrutiny. Now I will take that challenge up a level by presenting evidence and analysis showing that our environment is nowhere near as endangered as the press make it sound. If you have been sailing through until now, please fasten your seat belt.
Are polar bears threatened? Not at all, polar bear numbers are up dramatically over the past few decades. Much of the increase is due to reduced hunting of the animals, which is now down to about 600–800 bear-hunting permits per year. In the last ten years over 6,000 polar bears have been shot and killed by humans, and the populations are doing well. Polar bears are not threatened, they are thriving.
Are our forests being depleted? Believe it or not, we have more green acreage now than any time in the last several thousand years.* We do need to promote biodiversity and protect intact ancient forests. This are emergencies in a few places, but overall it’s not as bad as the press makes it sound.
How about DDT? DDT was widely vilified in the 1980s. Researchers claimed it was responsible for eggshell thinning and birth defects in raptors.* The alarm was sounded, and as a result DDT use was curtailed. That led to many more malaria deaths than we would have had if we had kept using DDT. Recent studies have shown only a tiny association with birth defects, the trade-offs are worth it to eliminate malaria, and the WHO includes DDT on the list of approved substances to kill mosquito eggs. Cost-benefit analysis is a better tactic than emotional alarms, scary imagery, and zealotry.
What about the Ozone hole? It’s likely that CFCs had nothing to do with the ozone hole and stopping them simply coincided with the natural reduction of the size of the “hole.” On the other hand, the basic chemical theory is solid, but we should always ask for direct evidence, not laboratory evidence. NASA recently reported that the size of the ozone hole naturally varies a lot over decades. The ozone hole is likely to continue expanding and contracting naturally. I don’t think there was ever any emergency. If you have interesting data on this, please send it to me.
Are we losing all our insect species? We are losing many species, but it’s not as bad as people think. It’s not an insect apocalypse, but we should definitely take steps to limit further breakup of intact ancient forests.
What about the bees? Bee colony collapse was very real in the first decade of this century. The causes were quite complex, involving many factors, but it only lasted a few years. Colonies routinely lose bees in winter; beekeepers have many options for restoring colony numbers in summer. In general, colony collapse disorder was overblown in the media. Bee numbers have been predictably steady for decades.
Should cities or states ban the use of plastic bags? If the replacement is paper bags: no; if the replacement is cloth bags people keep and reuse over and over: yes. How do you know which bags people will use? Soon, we will have algae and other organic-based plastic substitutes that will degrade gracefully.
Should we recycle plastic bottles? It’s complicated. There are a lot of variables. Many cities find it unprofitable. When the economics work, they don’t get turned into new bottles, they get turned into carpet fibers. We should reduce our need for one-time bottles, but in many cases, landfill may be the best place for them.
Should we recycle glass bottles? If it is put into concrete products, yes. If the colors are separated and your city is far from a coast, yes (Switzerland recycles three colors of glass profitably). If it is mixed, it isn’t going to become glass again in most cases, because sand is cheaper. That could change as sand prices rise. My guess is that a lot of cities run unprofitable recycling programs just so they can be seen as green. I need more data to understand this better.
Should we recycle paper? I don’t know. I can’t find any good studies on this. I’m guessing it has to do with how far away you live from a large forest. Please send me any data!
Should we recycle aluminum? It takes a lot more energy to mine and refine aluminum than it does to recycle it, so I thought there would be good empirical evidence on recycling economics. Even though about half of all aluminum is recycled, I can’t find an empirical analysis. Can you?
Is plastic and garbage polluting our oceans, killing all the marine life? No, not as much as you might think, there are no floating acres of plastic out at sea, no huge garbage patch, and the plastic found in the ocean is largely under 2mm in diameter (very small pieces) or microscopic. Half the plastic found in the oceans is floating fishing line and nets, but the amount overall is trending up. Despite the fact that our oceans are remarkably resilient, we still should try to reduce the flow of plastics into them.
Are we destroying fishing stocks by overfishing? Yes. We are removing and eating the predators. This is changing the balance of life in the oceans. A battle between fishermen and lawmakers is a battle the fishermen will first win, then lose. Humans will remove almost all predatory fish in the next several decades, reducing biodiversity and causing further knock-on effects. In reality, there will be plenty of fish in the future, but they will be small fish, not the fish we eat today, and that will change our oceans profoundly. I’m generally not concerned with most environmental alarms, but this one keeps me up at night.
Is wind power effective? In reality, there’s no such thing as wind power. There is only wind plus whatever you use to back it up when the wind doesn’t blow. Therefore, you have to compare wind plus existing power and factor in subsidies and upcoming rule changes (like California’s commitment to eliminate fossil fuels by 2030). Today, about six percent of world energy comes from wind, much of it heavily subsidized. Some places, like the UK, already have a lot of wind power and are probably starting to see diminishing returns. Other windy places with expensive energy will still benefit, and offshore wind farms are looking particularly unprofitable. Without continued subsidies, today’s wind technology, which is getting close to the Betz limit in efficiency, is unlikely to bring much value to the energy market (see table 4). Electricity costs are higher in places with more wind farms, because their stable energy sources run much less efficiently. There are problems with birds, wildlife, noise, human deaths, long-distance electric lines, and alternative uses for the land. The viability of wind also depends on storage costs coming down, because wind power doesn’t peak during the middle of the day, when demand is high, as solar does. Forcing wind power to be competitive with other sources will result in wind power where it truly works.
Is solar power the energy of the future? Solar power has much more promise than wind, because you can convert sunlight on your roof into energy to run your house. As prices come down and efficiency goes up, solar is definitely a smart source of power. We still need traditional power sources for cloudy days. In reality, electricity prices where there is a lot of solar are still higher than in the rest of the country because power plants are not designed to run as back-up generators. It makes sense to keep investing to make solar more effective, but it is not a full-time replacement for traditional power.
Does my electric car run on fossil fuel? In reality, it probably does. If your electricity comes from a coal-fired power plant (30 percent of US power) or from natural gas (34 percent), then yes, your eco-car runs on fossil fuel.
What about nuclear power? I believe nuclear power is the future. Fourth-generation reactors are safe, eat nuclear waste as fuel, and are coming down in size and price. Today’s nuclear power provides a reasonable trade-off against fossil fuels. Within 30 years it’s about 50 percent likely that we’ll have fusion power. If that happens, we’ll have unlimited clean power for almost no money. The technical term for that is “foom.”
Is the planet warming as a result of man? There is certainly a lot of consensus that extra CO2 is warming the earth, and I understand if you are reading this and believe it is true. But, as we have seen so many times previously, I think it’s worth challenging this common belief. How much of the data have you actually looked at? I spent a year on this topic looking at thousands of graphs and reading thousands of papers, and after about six months I changed my mind. What would it take to change yours? In the last 50 years, we had 30 years of warming followed by 20 years of no warming.* Model predictions of future temperatures are both flawed and much less precise than their creators would like us to believe. The earth hasn’t warmed significantly in the past 20 years,* defying model estimates. The debate is so political that it’s very difficult to discover the truth and the uncertainties. Research and data are highly skewed. IPCCC reports are politically driven. Did you know there are PhD’ed climate scientists who have quit the UN working groups and do not agree with the “consensus”? In reality, all the decarbonization we can do will have very little impact anyway. If you really want to understand this issue, you should first understand the uncertainties* (the science is not “settled”) and read Climatecurious.com.* If, after reading all the links in this paragraph, you still believe we have something to worry about, I will respect that view. And, I’ll be interested to bet on an outcome you can specify in the next ten years that proves your view correct.
Is sea level going to accelerate and threaten coastal cities? Sea level has been rising linearly roughly 7–10 inches (18–25 cm) per century for the past two centuries (long before CO2 started to increase). It continues in a linear fashion, even since the increase of CO2. This is heavily complicated by the fact that tide gauges are highly variable, land naturally sinks, and groundwater extraction causes land to sink further. In reality, impactful sea-level rise of more than 15" by the end of this century is so unlikely that it’s not worth preparing for on a cost/benefit basis. Dr Judith Curry has a careful evaluation of the evidence.*
Are reefs turning acidic and dying because of CO2 in the atmosphere? I don’t think they are. They die after El-Nino events, then they come back. They have been doing this for hundreds of millions of years, long before we got here. The thermal threshold for corals varies, and new corals easily replace old corals. There is good evidence that reefs are very resilient to temperature changes and routinely bounce back from shocks. The “scientific” research showing bleaching is heavily skewed by politics and the common myth of future climate change. (Sunscreen may play a role in destroying corals, but they are resilient creatures). More at climatecurious.com.*
Are duck and goose down products sustainable? Many manufacturers now adhere to “ethical down” practices. The National Sanitation Foundation’s Global Traceable Down standard and the Textile Exchange’s Responsible Down program are similar. Their approach is to gather feathers that fall naturally from geese or plucking after geese are killed for their meat (excluding force-fed). Many major brands now are certified under one of these programs. It makes sense to look for those labels.
NEWSLETTER AND BLOG NOW AT CUTTINGTHROUGHTHENOISE.NET
IN REALITY, THE WORLD IS FAR MORE DIFFICULT TO UNDERSTAND THAN MOST PEOPLE THINK. Here I offer a few principles to better align your own map with the territory.
Correlation is not causation. Now that even high-school kids are saying this, it seems trite. But it’s more true than ever. There are a lot of correlations, many of them spectacular. We understand very little about causation.
Facts are not evidence. Facts are everywhere, you can use them to justify anything. Evidence is everything — you must take the facts you don’t like along with the facts you like. My video series explains this.
Most scientific research is wrong. As Laura Arnold says, the four most dangerous words are “A new study says …”. If you didn’t watch Derek’s video above, please go back up and watch it now. The rest is in this book, which I highly recommend (it’s linked in the resources section):
Complex adaptive systems are not the same as complex systems. A complex system would be tectonic plates moving and causing earthquakes, or a famous experiment about grains of sand piling up and slipping. Those systems don’t continually evolve and learn. In a complex adaptive system, what worked last time won’t necessarily work next time.
Ecosystems are complex adaptive systems. Ecosystems are characterized by:
Hundreds of simultaneous arms races, many of which are invisible
Auctions, in which people make irrational, emotional choices
Beauty contests with winner-take-all economics
Moving, even elusive targets
Black-swan events, unprecedented surprises from unknown unknowns
Implicit games with unwritten rules
People gaming the system
Others pre-gaming and re-gaming the system in response
Lots of data, signals, statistics, information, and numbers
Solutions that are temporary at best
Adversaries and allies, many of whom are difficult to distinguish
The fast rise of new frontiers
Uncertainty, ambiguity, and volatility = the red-queen effect
Most signals are weak. This is an understatement. We continually learn that what we thought we knew was really just a misreading of the available information. We thought a low-salt diet was good. Now we think that a bit more salt is good for you.* Serious analysis shows that most antidepressants are better than nothing, but not a lot better.* Prevailing beliefs are often based on small signals with large margins of error.
Doctors really don’t know much. Studies show that doctors are overconfident, they are as susceptible to bias as anyone else, they often screw up patient diagnoses, many of them don’t keep up with the literature in their field, most can’t do basic Bayesian reasoning, they don’t wash their hands, and using simple checklists saves many lives. I certainly know some very smart doctors who understand all of this and are very good at statistics. Smart doctors say “I don’t know” or “it’s complicated” more often than other doctors do. If you have a serious medical decision to make, always consult a qualified (not certified) statistician.
Cause and effect are extremely difficult to determine. Complex adaptive systems are black boxes with inputs and outputs — don’t assume you understand the mechanism, because even thinking in terms of “mechanism” is wrong. Statisticians understand the world better, because they understand variance and uncertainty. Economists try to see the big picture.
Prediction markets are the best way to forecast the future. In complex adaptive systems, experts’ opinions aren’t very useful. Prediction markets have already proven themselves as excellent predictors, because people must put up their own money, and if they are wrong, someone else profits. Having skin in the game is the best way to get accurate forecasts. They can help inform policy, set monetary policy, colonize space, and much more. Unfortunately, most countries see these information markets as gambling and have outlawed them. This needs serious reform.
“The first principle is that you must not fool yourself, and you are the easiest person to fool.” ― Richard P. Feynman
All models are wrong; some models are useful. Nothing is perfect. We make decisions using very limited information. Yet limited information is better than no information. If your model isn’t improving over time, you’re probably not putting enough work into maintaining your tools.
Myths are stubborn. I’ve pointed out a few myths here, and there are many more. Once an idea gets momentum, it’s hard to stop. This is why I’m generally against learning from textbooks. I think if you want to become knowledgeable about a topic, it would be better to read ten different books than one textbook. Be wary of statements like “97 percent of scientists agree” — that’s a sign that someone is trying to sell something.
Don’t get a medical test unless you would be willing to make a change. People often get tests “to establish a baseline,” and this is generally useless. Before you take any test, ask yourself if you are willing to treat if positive and not treat if negative. If you are unwilling to treat, don’t test.
Most people are not statistically literate, including scientists, researchers, doctors, lawyers, judges, politicians, boards, and others who make rules and decisions. Anyone who has taken a few courses in statistics is dangerously overconfident. Start with the Marginal University course on econometrics, then dive into Khan Academy and read the books in the list below.
Most science is junk science. Most scientists are not strong statisticians. We have seen that even well trained statisticians can do a poor job of randomization. Given that most scientific research results are false, we should assume most science is junk science and be pleasantly surprised when an experiment is actually properly designed with enough statistical power to be worth conducting in the first place (and not designed a-priori to show a particular result that the funder would like to see).
The world isn’t black and white. There are rarely any concrete answers, context matters, details matter. There is always a continuum of possibilities and outcomes. Cost-benefit analysis is a better tool than rhetoric and cherry-picking the facts to suit the outcome you want.
Scientists try to disprove their theories. They know that almost all theories are later replaced by something better. A good scientist will actively try to disprove her theory. There may be heated debate about a topic for some time (evolution, germs, smoking), but then we all learn more and science moves forward. Should we think of it as “debate” or as “learning together?”
Fall in love with the problem, not the solution. — Uri Levine
It’s okay to be wrong. In my experience, at least 90 percent of people, when confronted with evidence that they are wrong, tend to not want to hear any more. They change the topic quickly or defend themselves. Only a small number of people want to learn what went wrong. Expect most people you encounter to be the first type but try to hang out with the second type.
It’s okay to be surprised. People generally don’t want to be surprised. If they think a particular explanation would be surprising, they tend to rule it out and look for less surprising narratives. Reading Randomistas will show you that surprise is the norm. If you are not surprised, you are probably not looking very hard.
Anecdotes do not generalize. How many books are there on management using examples of wolves, military, sports teams, and other nonsense? Storytellers easily convince their audiences that this particular story applies to you today. That’s what they think at Harvard Business school,* which produces graduates who are overconfident because they have read and discussed hundreds of stories. This leads us to differential diagnoses* — different people, economies, and countries need different solutions. One size does not fit all.
What would it take to change your mind? This is a big question among rationalists. You should be able to say ahead of time what it would take and then look for it. If you find it, don’t move the goal posts!
How much would you be willing to bet? This is a Bayesian measure of your belief in something. You may be 90 percent sure of something, but would you be willing to bet 90 percent of your assets on being right? How much would you be willing to bet? Bayesians never agree to disagree. In this introduction to Bayesian reasoning, take a quick test and see how well you do:
You should never be 100 percent confident of anything. That means you would bet your life on it, and that’s foolish. In fact, confidence should be nonlinear — it should take more and more evidence to move you one percent of the way toward “knowing something for sure” the closer you get to believing it. Never discount the possibility of unknown unknowns.
Context and details matter. Rationalists often say “I think you’ll find it’s a little more complicated than that.” ITYFIALMCTT for short. Every time we think we have discovered a “rule,” it turns out to be a pretty complex set of rules that apply in some situations but not in others. Alfred Sloane famously said in a board meeting:
Gentlemen, I take it we are all in complete agreement on the decision here. Then, I propose we postpone further discussion of this matter until the next meeting to give ourselves time to develop disagreement, and perhaps gain some understanding of what the decision is all about.
Variance is remarkably common and little appreciated. Complex adaptive systems have variance and uncertainty in several dimensions. Variance is often interpreted as signal, but variance is not signal, it’s noise. Statisticians ask: “How likely is it that this result is simply due to chance?” Danny Kahneman has recently been writing and talking about this.
Meta-studies are important. Not all metastudies are well done or statistically powerful. The rest probably summarize the best of our knowledge at any one time, though there are limits to their power. Meta-studies are done by statisticians. If you find a recent meta-study, you have to take it seriously. To find them, try the Cochrane Library and search Google Scholar for a search term with the words “meta analysis” added at the end.
Large, randomly controlled trials are important, but many are poorly designed. You need a proper control, which many studies don’t have, and you need double blinding whenever possible. The PREDIMED study, which made fantastic claims about the health effects of olive oil, was huge and highly regarded, until it was discovered that they had made significant mistakes in the design of the study. The benefits are likely significant but less than the original claims.* A weight-loss study that provides data at six months is meaningless, but one that shows results after two years would be a breakthrough. The Million Women study results are important. Yet randomization is harder than we thought.
Genetics are important. I expect human and microbial genetics will change many of today’s dominant paradigms in medicine, health, nutrition, and much more. Our treatments will get more and more nuanced, more individualized,* more targeted. What works for one person doesn’t work for another. This goes for cities, countries, cultures, and societies.
Anecdotes make for good hunches, not conclusions. If researchers find something they think is worth investigating, they formulate a hypothesis, try to design a statistically powerful study, determine feasibility (often they have to do a small trial study first), then pass the ethics board, put a budget together, find unbiased money to fund it, then report their unbiased data, findings, and confidence intervals. Then someone else needs to get the funding to verify those findings independently, using the same methods. At the end of all that, they usually find “no correlation.” Occasionally they find a few weak signals and “more research is needed.” Research is like mining for gold — few people hit anything big in a lifetime of digging. Resilience matters.
Destination is more important than instrumentation. How do we know coral reefs will all be destroyed by global warming? Scientists do tank experiments, where they raise the water a few degrees. They then release their findings to the news wire, claiming the oceans will all be death zones in a few decades (which the press eats like cake pops). In reality, that’s not how oceans work. Reef temperatures have been changing (variance) for hundreds of millions of years, and corals have been adapting, dying, and regenerating for just as long. Lab experiments can only give us hunches to test in the real world. Don’t look at instrumental variables like triglycerides and biomarkers — look at endpoints like heart attacks, strokes, and deaths. Don’t look at arrests or jail time, look at the measurable cost/benefit to society. Don’t measure carbon dioxide, measure long-term global mean temperature or sea-level trends. Similarly, there is no need to count calories, wear a device that measures your activity, or look at a bathroom scale. If your goal is to lose weight, look in the mirror to see your results.
It’s not always about data. You want to get whatever data you can, but sometimes can’t or it isn’t very good. I can’t present any data on abortion that will change anyone’s mind. Abortion is only about values. We have almost no data on what causes people to live longer, because it takes too long to do the experiment. This also appears in extreme value theory, where there aren’t enough events in the past to predict the future. Conflicting data is something else. When you see conflicting data, you are obliged to try to figure out why it conflicts and make sense of it — you can’t write it off as “self canceling.” As Andrew Leigh says, “If you want to know about the impact of denuclearization on the Korean peninsula, a randomized trial is probably not your best way of working it out.”
Systems beat individual judgment. Forget about goals, forget about talent. Stop trying to keep everything in your head. This video explains how Scott Adams used a system to beat the system.* At Menlo Innovations, they use a system to do all their project management.* In fact, systems are replacing project managers around the world.
Skeptics tend to be right most of the time, but optimists move the world forward. Alternate between the two. The Apollo program was a remarkable see-saw between huge audacious goals and pragmatic prototyping, testing, and failure. Same with the Wright Brothers. This ability to make grand leaps but let the experimental evidence lead you forward is a hallmark of many big successful projects. To understand the world, be about 70 percent skeptic, 30 percent open to new ideas.
Trial and error is a remarkably good strategy. We call it “explore and exploit.” Imagine if I ran for president and said, in answer to any question, “I’m really not sure, I would want to do randomly controlled trials and see where they lead.” Do you think I would get elected? Maybe not, but I’m an atheist, so there’s no chance anyway. It may not get you elected, but it’s good advice for you in your business and in your life.* To optimize your engagement with the world, make trial and error your normal procedure.
Hold strong opinions loosely. I learned this from Doug Hubbard after reading his books. I’ve gotten better at saying “I don’t know, what can we learn?” When talking with experts, I’m never surprised to hear: “I think you’ll find it’s a little more complicated than that.”
NEWSLETTER AND BLOG NOW AT CUTTINGTHROUGHTHENOISE.NET
Thinking Fast and Slow (beginner), by Danny Kahneman
Factfulness (beginner), by Hans Rosling
Mindware (beginner), by Richard Nisbett
Wrong (beginner), by David Freedman
Pull (beginner), by David Siegel
The Elephant in the Brain (beginner), by Kevin Simler and Robin Hanson
The Black Swan (intermediate), Skin in the Game (intermediate), and Antifragile (advanced) by Nicholas Nassim Taleb
Uncontrolled: The Surprising Payoff of Trial-and-Error for Business, Politics, and Society (intermediate)— Jim Manzi
Radical Markets (intermediate), by Weyl and Posner
The Case Against Education (intermediate), by Bryan Caplan
Randomistas (intermediate), by Andrew Leigh
Statistics Done Wrong (advanced), by Alex Reinhart
Stubborn Attachments (intermediate), by Tyler Cowen
How to Measure Anything (intermediate), by Douglas Hubbard
Expert Political Judgment (intermediate), by Philip Tetlock
Principles (beginner), by Ray Dalio
The Rational Optimist: How Prosperity Evolves (intermediate), by Matt Ridley
Shut Out: How a Housing Shortage Caused the Great Recession and Crippled Our Economy (advanced), by Kevin Erdmann
The Flaw of Averages (advanced), by Sam Savage
Inadequate Equilibria (advanced), by Eliezer Yudkowsky
Rationality, from AI to Zombies (advanced), by Eliezer Yudkowsky
The Checklist Manifesto(intermediate), by Atul Gawande
The Money Illusion — Scott Sumner
Macro Musings — David Beckworth
Essays by David Siegel
The Rational Optimist — Matt Ridley
Marginal Revolution — Tyler Cowen
Overcoming Bias — Robin Hanson
Idiosyncratic Whisk — Kevin Erdmann
Slate Star Codex — Scott Alexander
Crooked Timber — Various
Sciencebasedmedicine.org — Steve Novella et al. (please support them!)
Straight Talk on Evidence
Evidence-Based Programs (that work)
IN THE PAST SIX YEARS, I have broken both my hips at the femoral neck. I got a bone scan and learned that I have osteoporosis, which is extreme for someone in his 50s who is as sporty as I am. My doctor prescribed Fosamax, which I learned is not for active people (I’m not convinced it benefits anyone, but it may help frail elderly people).
There is no correlation between being vegan and fractures, but there is a small amount of data on older people with low-protein diets having thinner bones and breaking them. According to that study, increasing the protein helped that population prevent further fractures. This isn’t much to go on, but it’s more than I had before. I’ve decided to stay vegan and increase my protein intake. In a few years, I may then go for another bone scan to see if it has made any difference (or not — you only take a test if you would do something different depending on the outcome of a test, remember?). Since we know so little about men in their 50s and osteoporosis, I’m doing an experiment where the number of subjects (n) equals 1. Very few people eat a low-protein diet. Adding plant protein to my diet isn’t difficult. I will be interested to see the results (if any).
For the record, I also take B-12 tablets, not because I really think I need them but because they are delicious and it doesn’t hurt. I know there are recommendations that vegans require extra B-12, but I believe the evidence for that is very weak. In general, I don’t think vegans need any special supplements, but I do hope we learn more in the coming years.
What do I worry about? Mostly, I worry about overconfidence, statistical ignorance, and rigorous nonsense, which divert resources to made-up problems like global warming, regulation, and going to Mars. I worry about people focusing on near-term problems like tax rates and minimum wage, rather than big important problems, like maximizing the rate of sustainable economic growth. I worry that most of our institutions are optimized for the world we had back in 1950.* I worry about the poverty trap, the health care system, our political systems, central banks causing recessions, not investing in nuclear power, the signaling economy, social safety nets, the slow pace of innovation, and the future of humanity in an AI/robot/brain-emulation world (after 2050). I worry that children born today will go through a school system designed for the industrial age. I worry about the fact that banking, insurance, and other industries are held back by rules and structures from the last century, when we now have technology that can replace those structures. I worry that $13 trillion of impact investments don’t have much impact. I worry that the playing field is designed for huge companies to get bigger and that makes us all vulnerable.* I worry about cybercrime and invasion of privacy — not just from criminals but from governments like the United States, Russia, and China. The more connected we become, the easier it is for them to discover and exploit vulnerabilities. When I have a spare moment, I worry about overfishing, antibiotic resistance, and dietary ignorance.
Mostly, though, I don’t worry. I’m glad things are generally improving, they aren’t as bad as people think, and we are naturally finding solutions to many big problems. Even though the gap between rich and poor is widening, the poor are much better off than they were 20 years ago. The more connected we become, the faster ideas spread.
In my research, I have been most impressed by the Laura and John Arnold Foundation. They are fighting the right battles. I hope they can scale up their model to meet the needs of the entire world. Aside from the Arnolds, I’m shocked at how much money wealthy people waste on frivolous “save the world” projects without understanding the real problems first. It seems like philanthropy is mostly virtuous signaling. I’m not surprised at all how bad we are at trying to help people who are less fortunate. I just wish it weren’t that way, because it doesn’t have to be.
I will continue to do research on these and many other topics. My main hope is that you’ve enjoyed reading this as much as I’ve enjoyed researching and writing it. My second hope is that someone will want to help turn this into something much bigger.
If you learn to look at the world this way, I think you’ll find it’s a little more complicated than you thought before. And that might help the next generation of problem solvers make the world a better place.
My thanks to: Kevin Dick, Rob Siegel, Scott Sumner, Kevin Erdmann, Satish Luintel, and Mike.
If you want to contact me because you have corrections, can fill gaps in my knowledge, want to show me evidence or analysis, or you would like to discuss taking this work to the wider world, or join our small team of research volunteers, please email me (david at dsiegel dot com). If you have suggestions for topics, I won’t be able to answer them. If you disagree with me, that’s really not the point. The point is to give you links to data and analysis, so you can look at new evidence and learn. Did you?
NEWSLETTER AND BLOG NOW AT CUTTINGTHROUGHTHENOISE.NET
David Siegel is a serial entrepreneur in London. He is the founder of the Pillar Project and 2030. He is the author of The Token Handbook, Open Stanford, The Culture Deck, Climate Curious, and The Nine Act Structure. He gives speeches to audiences around the world — see his speaker page if you would like him to speak at your next event. His full body of work is at dsiegel.com.