One Helpful Idea: Small effects, large sample

Studies with lots of study participants are better than smaller ones, but also have a unique way of being misleading.

Scientists will run a large study and then conclude "X causes Y and the result is statistically significant (p<0.05)." For instance, they may determine that "this specific diet causes increased blood pressure."

But, for large studies, this can happen for really small effects. In those cases, it can be MORE accurate to say "X has no effect on Y" than "X has an effect on Y" even though, technically, there is an effect.

This highlights the difference between importantly distinct ideas which easily get conflated: how likely it is that there is any effect, and how big that effect likely is.

Scroll down for the video of the week, question of the week, etc.

Video of the week

6 Paradoxes That Will Break Your Brain
(12 minutes)

Also, if you are interested, I'd really appreciate it if you'd subscribe to me on YouTube.

Question of the week
 

What common human experience can you not relate to at all? Bonus question: why can't you relate to it?

You can read other people's answers or join the ongoing discussion about this question on X or Facebook.

C.R. Says:
"‘Showing off expensive materialistic goods’ for example showing off a car or designer items. I can’t relate to it because it seems to be a liability and drawing the wrong type of attention. I don’t particularly like drawing attention to myself because I have social anxiety."

Anders Sandberg says:
"Sport. I feel mildly impressed that athletes can do extreme things, but not excited by the fact or drawn to see them struggle. Team sports are hard to even parse for me, and I don't feel drawn to tribal identification."

N.T. says:
"The ability to enjoy a rave. The sensory overload feels violent for me, and the environment feels claustrophobic. Yet I have many friends who tell me it is enjoyable."

Poll of the week

Podcast episode of the week

Seeing through cognitive traps
(with Alex Edmans)

Key Takeaways:

🔑 Treat correlations as prompts for causal hypotheses, not proof, and first rule out statistical flukes and alternative causal structures before acting on them.

🔑 Confirmation bias is powerful partly because beliefs serve comfort and identity as well as truth, so even small value‑judgments about what we prefer to be true can tip us toward accepting weak causal stories.

🔑 Scientific publishing incentives often reward quantity, safety, and exhaustive editing over clear, faster insight, as editors default to asking for all reviewer demands instead of a focused core (what the reviewers agree on).

Essay of the week

Categorizing The Causes Of Bad Things In The World

What causes bad things? It sounds like a huge question, but maybe it’s not as big as it seems. Here’s my updated/improved list of high-level causes of bad things in the world. Note that these are not mutually exclusive categories. I’ve also added some potential solutions for each cause.

I’d be interested to know: what is missing from my new list of causes of bad things and potential types of solutions? Thanks to those of you who commented on my prior version!

Causes of bad things in the world:

—

1) EXTERNAL CAUSES

1i) Nature or evolution (e.g., malaria, cancer) -> Potential solutions: technological development, such as medical cures

1ii) Bad luck (e.g., landslides, earthquakes, droughts) -> charity, government programs providing social safety nets

1iii) Scarcity (e.g., insufficient food or water in an area) -> migration away from high scarcity areas, technological development to increase food production

—

2) FAILINGS OF HUMAN NATURE

2i) Highly selfish actions by non-evil people (e.g., some of the crimes that are committed, some of the manipulation that occurs) -> cultural norms discouraging selfishness, cultural norms to punish those taking highly selfish actions

2ii) Harmful actions taken in highly emotional, confused, or desperate mental states (e.g., crimes of passion, harmful, desperate reactions out of fear, harm caused during extreme mental illness) -> widely available and effective mental health treatment, widespread education/training related to mental health and emotional regulation

2iii) Well-intentioned ideologues who are convinced that their simple but wrong model of the world is the absolute truth (e.g., some of the genocides and wars, many harmful yet well-intentioned policies) -> rationality education/training, a robust culture of respectful disagreement and debate

2iv) Cognitive biases leading to actions with severe negative consequences (e.g., greatly misjudging whether a project will bring enough benefit to be worth the cost, excessive fear towards or devaluing of ‘othered’ outsiders leading to mistreatment, or harm to outsiders, lack of preparation for likely occurrences that are not salient, under-recognition of some beings' or some groups' moral status) -> rationality education/training, careful design of systems to counteract biases, strong moral norms of respect towards all, moral circle expansion

2v) Retaliation or revenge (e.g., cycles of retribution) -> a culture of forgiveness, effective dispute resolution methods and institutions, reliable enforcement of laws

2vi) Evil people acting alone (e.g., serial murder, child abuse) -> effective police forces, high crime clearance rates, enforcement of laws, scientific investigation into the root causes of evil

2vii) Evil people who rally supporters (e.g., some genocides and wars, some extractive government policies) -> strong norms around truth telling and social punishment for lying, a robust culture of respectful disagreement and debate, a culture of empathy toward and acceptance of those who are different than you, a well-educated and informed citizenry, scientific investigation into the root causes of evil, a strong constitution, a strong independent judiciary, strong norms around maintaining freedom and independence of thought

2viii) Harmful cultural norms and destructive ideas (memes) that spread through populations -> (e.g., honor cultures that create cycles of harm, stigmas around healthy behavior, positive viewpoints toward healthy behavior) -> active rigorous debate of ideas with, a norm of openness toward challenging norms

2ix) Mismatch between evolved minds and modern environments (e.g., anxiety that gets triggered frequently without helping people achieve their goals) -> effective psychological treatments, redesigning our environments to better match our nature

—

3) CHALLENGES OF COORDINATION AND INFORMATION

3i) Negative-sum competition (e.g., fighting over food when there isn’t enough to go around) -> technological innovation to increase abundance, thoroughly enforced laws forbidding negative-sum behaviors

3ii) Unintended side effects of actions that are not innately unethical (e.g., addiction caused by the invention of social media, new promising-seeming medical treatments that turn out to have horrendous side effects, dangerous technology that's created by people who do not see it as dangerous) -> a robust and low-transaction cost systems for those who were harmed to be compensated by those who caused the harm, hard to undermine enforced regulation requiring organizations to ameliorate harms once they have been identified

3iii) Collective action problems and negative externalities caused by individually reasonable behavior (e.g., pollution, climate change, overuse of resources) -> methods for assigning prices to negative externalities so that someone bears the cost, regulation to limit negative externalities

3iv) Prisoner’s dilemmas and difficulties of pre-commitment and coordination (e.g., arms races, such as with nuclear weapons) -> technology to facilitate coordination and simultaneous action, public projects by governments and private donors

3v) Misaligned incentives and principal–agent problems (e.g., politicians being incentivized to do what gets them re-elected even when it's not good for the people they represent) -> redesign incentive structures, increase transparency of how people act and the impacts of those actions

3vi) Harmful path dependency and lock-in (e.g., structural injustice, lock-in of harmful technologies) -> allowing for change and innovation, redesign of existing systems

What other broad causes of bad things or potential types of solutions am I missing?

Spencer used generative AI to produce the image at the top of this newsletter.