Welcome to The Nonlinear Library, where we use Text-to-Speech software to convert the best writing from the Rationalist and EA communities into audio. This is: CEA’s events team: capacity building and mistakes, published by Amy Labenz on November 3, 2021 on The Effective Altruism Forum.

For CEA's Q3 update, we're sharing multiple posts on different aspects of our work.

This post was written before EA Global: London 2021, and refers to that event in future tense.

Events enable attendees to make new connections, learn about core concepts, share and discuss new research, and coordinate on projects.

In summary:

We put a lot of effort into building up our team.

We ran two well-received medium-sized events.

We made some significant mistakes with one of these events, which means that the counterfactual impact might have been negative. We plan to reflect on this after EA Global is complete, and we expect to make significant changes to address this.

In the past few weeks (as of 10/26/21), we’ve managed to roughly double the capacity of EA Global: London, in response to an unexpectedly high number of very strong applications. We’re currently focused on this event.

Depending on how EA Global: London goes, we expect to increase the total number of new connections (our key metric) relative to previous years. Optimistically, we might double the number of connections relative to 2019 (though the increase will likely be somewhat lower).

Metric: connections

Past surveys (e.g. Open Phil’s survey) suggest that connections between individuals are the key source of impact from our events. So we focus on the number of new connections we make at our events.

Currently, we calculate this as follows: number of attendees x average number of self-reported new connections. We define a new connection as “someone you feel comfortable reaching out to to ask for a favor”, so this is a relatively high bar. We expect that this is an overestimate, since people who engage a lot with the event are more likely to fill in our post-event survey, and also more likely to make lots of connections. However, we think it allows for some comparison between events and years.

We’re currently working on fine-tuning this metric, finding better ways of measuring it, and thinking through other important outcomes to consider for measuring event outcomes.

Broadly, we think that we’re on track to nearly double new connections compared to 2019. This would be a slight increase from 2020. However, we think this methodology somewhat overestimates growth relative to 2019, and underestimates growth relative to 2020:

We think this methodology overestimates the number of connections at virtual events relative to in-person events[1]; all events in 2020 were virtual, and some events in 2021 were virtual, while all 2019 events were in-person.

2020 figures include EAGx, whereas 2019 and 2021 figures don’t.

See footnote #2 for details on how we came up with our "projected" figure.[2]

Staffing

Overall, we’re currently in a transition period for the team: we’ve made some very strong hires who we are now onboarding, or who are joining later this year. During this onboarding process, we’re still understaffed. But we think the team will be stronger in about 6 months.

Additionally, Amy is transitioning from a role focused on direct work on events to a role more purely focused on management, with support from Max (Executive Director).

We hired.

Lizka Vaintrob for the Events Generalist role: we expect her to focus on communications and help out with impact analysis.

Ollie Base as our Community Events Manager. He will help us to develop our overall events portfolio, with a particular focus on expanding our community-led events.

Pat Magcalas as a Personal Assistant for Amy Labenz at the start of October.

Barry Grimes is transitioning to the Happier Lives Institute starting in November. We’re grateful for all of his hard work on events over the past 2.5 years.

We’re working with Anine Andresen (previous...

Welcome to The Nonlinear Library, where we use Text-to-Speech software to convert the best writing from the Rationalist and EA communities into audio. This is: Liberty in North Korea, quick cost-effectiveness estimate, published by MichaelStJules on November 3, 2021 on The Effective Altruism Forum.

In 2019, Liberty in North Korea (LINK) claims to have rescued 222 North Koreans, spending $3,660,223 across all programs, including refugee resettlement. In 2018, it was 326 rescues, spending $3,604,423. Combining, that's 548 rescued for $7,264,646, or $13,257 per rescue. Rescue expenses alone (so excluding employee pay, other program expenses, and everything else) over both years totaled $932,456, which would come out to $1,702 per rescue. For comparison, GiveWell's recommended life-saving charities are estimated to save a life for $3000-5000 on average.

EDIT: What they do: They rescue already escaped North Koreans from China using secret routes, since if they are caught, Chinese authorities would send them back, or the women could be sold into sex trafficking or as brides. LINK doesn't help North Koreans escape in the first place. LINK claims thousands of North Koreans attempt to escape each year. They also write:

It costs $3,000 to help a North Korean refugee travel from Northern China to safety in South East Asia

The Changing Costs of A Rescue

The cost of rescues varies on where a North Korean refugee begins their journey. The closer they are to the North Korean border when our partners find them, the greater the cost because of heightened security and increased travel time. A generous foundation funds these high-risk extractions from the border region.

There are likely several ways this cost-effectiveness estimate may be off (including possibly having the wrong sign) due to indirect effects and not accounting properly for the counterfactuals, and a cost-effectiveness model should take such considerations into account. Here are some:

Less cost-effective: How likely is it that these people would have succeeded anyway, on their own or with the help of another group? Their website had someone who made 4 escape attempts (Jo Eun, on this page). How much earlier does LINK move their success if they would have been caught and made further attempts?

Unclear, lean more cost-effective: What are the risks or benefits to the families of rescues, and how likely are they? If someone is caught, their family may be punished, e.g. with labour camps or execution, and by rescuing them, we may prevent this. On the other hand, we may incentivize further escapes, which risk punishment for them and their families. To what extent are the families also aware of and accept these risks? (Credits to Bruce Tsai and Denis Drescher.)

Less cost-effective: They only rescued 15 North Koreans in 2020 due to increased security due to COVID, and maybe we should expect it to be similar going forward.

Unclear: Does this undermine reform in North Korea, by taking those disproportionately likely to push for reform? (Credits to edwardhaigh) On the other hand, LINK says that refugees send money and information back to their families in North Korea, and it's possible this could undermine the regime. (Credits to Khorton)

Less cost-effective: How would North Korea respond to increasing rescues? Harsher punishments and security? Further encouraging births?

Unclear, but lean less cost-effective: Where does marginal funding actually go? Would they actually rescue more, or just spend more on resettlement and other services? In what proportions? If we funded them, could we get them to spend disproportionately more on rescues? How much could this scale before further rescues became very difficult?

Unclear: How are descendants affected? Escapees who are caught and sent back may be prevented from having children, or have children anyway in North Korea (this is something to check). This may extent to their families, as well. Rescued escapees seem likely to have children outside o...

Welcome to The Nonlinear Library, where we use Text-to-Speech software to convert the best writing from the Rationalist and EA communities into audio. This is: Liberty in North Korea, quick cost-effectiveness estimate, published by MichaelStJules on November 3, 2021 on The Effective Altruism Forum.

In 2019, Liberty in North Korea (LINK) claims to have rescued 222 North Koreans, spending $3,660,223 across all programs, including refugee resettlement. In 2018, it was 326 rescues, spending $3,604,423. Combining, that's 548 rescued for $7,264,646, or $13,257 per rescue. Rescue expenses alone (so excluding employee pay, other program expenses, and everything else) over both years totaled $932,456, which would come out to $1,702 per rescue. For comparison, GiveWell's recommended life-saving charities are estimated to save a life for $3000-5000 on average.

EDIT: What they do: They rescue already escaped North Koreans from China using secret routes, since if they are caught, Chinese authorities would send them back, or the women could be sold into sex trafficking or as brides. LINK doesn't help North Koreans escape in the first place. LINK claims thousands of North Koreans attempt to escape each year. They also write:

It costs $3,000 to help a North Korean refugee travel from Northern China to safety in South East Asia

The Changing Costs of A Rescue

The cost of rescues varies on where a North Korean refugee begins their journey. The closer they are to the North Korean border when our partners find them, the greater the cost because of heightened security and increased travel time. A generous foundation funds these high-risk extractions from the border region.

There are likely several ways this cost-effectiveness estimate may be off (including possibly having the wrong sign) due to indirect effects and not accounting properly for the counterfactuals, and a cost-effectiveness model should take such considerations into account. Here are some:

Less cost-effective: How likely is it that these people would have succeeded anyway, on their own or with the help of another group? Their website had someone who made 4 escape attempts (Jo Eun, on this page). How much earlier does LINK move their success if they would have been caught and made further attempts?

Unclear, lean more cost-effective: What are the risks or benefits to the families of rescues, and how likely are they? If someone is caught, their family may be punished, e.g. with labour camps or execution, and by rescuing them, we may prevent this. On the other hand, we may incentivize further escapes, which risk punishment for them and their families. To what extent are the families also aware of and accept these risks? (Credits to Bruce Tsai and Denis Drescher.)

Less cost-effective: They only rescued 15 North Koreans in 2020 due to increased security due to COVID, and maybe we should expect it to be similar going forward.

Unclear: Does this undermine reform in North Korea, by taking those disproportionately likely to push for reform? (Credits to edwardhaigh) On the other hand, LINK says that refugees send money and information back to their families in North Korea, and it's possible this could undermine the regime. (Credits to Khorton)

Less cost-effective: How would North Korea respond to increasing rescues? Harsher punishments and security? Further encouraging births?

Unclear, but lean less cost-effective: Where does marginal funding actually go? Would they actually rescue more, or just spend more on resettlement and other services? In what proportions? If we funded them, could we get them to spend disproportionately more on rescues? How much could this scale before further rescues became very difficult?

Unclear: How are descendants affected? Escapees who are caught and sent back may be prevented from having children, or have children anyway in North Korea (this is something to check). This may extent to their families, as well. Rescued escapees seem likely to have children outside o...

Welcome to The Nonlinear Library, where we use Text-to-Speech software to convert the best writing from the Rationalist and EA communities into audio. This is: Paths Forward: Scaling the Sharing of Information and Solutions, published by Zvi on November 2, 2021 on LessWrong.

Follow-up to: An Unexpected Victory: Container Stacking in the Port of Long Beach

My chronicle of how Ryan managed to change the container stacking rule in the Port of Long Beach quickly became the most viewed post I’ve ever had. A lot of people were excited to spread the word about this story. This makes one think about the stretch goal, where we get to do things like this more often, potentially even creating this virtuous cycle:

Identify a concrete problem.

Explain the gears behind the problem.

Present a solution to the problem, and explain how it would work.

People notice this, and amplify the signal.

Signal causes people with ability to implement solutions to notice.

Solution is implemented.

Victory.

Except, you know, with planning and strategy involved.

Every time this happens, it becomes more normal to do it again, and it becomes easier to get good solutions amplified and implemented.

We get to thinking we can solve problems, and change things for the better.

Including when the solutions are less and less trivial

Including they require non-zero downsides or non-zero amounts of time

It becomes standard to solve problems and blameworthy to not solve them.

Setting aside for now how much the rule change mattered. For this to be exciting, it must be an asymmetric weapon. Does this work for positive-sum solutions to problems without working, or working less well, for zero-sum resource grabs?

My instinct was clearly yes, this only worked because it was obviously correct. If it had not been obviously correct, the effort fails, for overdetermined reasons. This isn’t obvious to others, and a lot of this post explores the gears involved.

Asymmetric Attention Weapons

I was highly frustrated to see this reaction, which I instinctively (if unfairly) translated as ‘actually, attempting to communicate information that might cause people to do things is bad, because the reference class is zero-sum games that compete for attention’:

It’s important to intuition pump why this is the opposite of most of the attention economy, and why such strategies are asymmetric. The information and solution both being true and important was central to the Tweetstorm’s success.

From my previous post‘s ‘I see what you did there’ list.

Starts with a relatable physical story of a boat ride, and a friendly tone.

Tells a (mostly manufactured) story that implies (without saying anything false) how the ride led him to figure these things out, which gives rhetorical cover to everyone else for not knowing about or talking about the problem. We can all decide to pretend this was discovered today.

Then he invokes social consensus by saying that ‘everyone agrees‘ that the bottleneck is yard space. Which is true, as far as I can tell, everyone did agree on that. Which of course implies that everyone also knows there is a bottleneck, and that the port is backed up, and why this is happening. The hidden question of why no one is doing much about this is deflected by starting off pretending (to pretend?) that the boat ride uncovered the problem.

Describes a clear physical problem that everyone can understand, in simple terms that everyone can understand but that doesn’t talk down to anyone. He makes this look easy. It is not easy, it is hard.

Makes clear that the problem will only get worse on its own, not better, for reasons that are easy to understand.

Makes clear the scope of the problem. Port of Long Beach effectively shuts down, we can’t ship stuff, potential global economic collapse. Not clear that it would be anything like that bad, but it could be.

Gives a decision principle that’s simple, a good slogan and again can be understood by everyone, and that doesn’t have any obvious objections: ...

Welcome to The Nonlinear Library, where we use Text-to-Speech software to convert the best writing from the Rationalist and EA communities into audio. This is: Paths Forward: Scaling the Sharing of Information and Solutions, published by Zvi on November 2, 2021 on LessWrong.

Follow-up to: An Unexpected Victory: Container Stacking in the Port of Long Beach

My chronicle of how Ryan managed to change the container stacking rule in the Port of Long Beach quickly became the most viewed post I’ve ever had. A lot of people were excited to spread the word about this story. This makes one think about the stretch goal, where we get to do things like this more often, potentially even creating this virtuous cycle:

Identify a concrete problem.

Explain the gears behind the problem.

Present a solution to the problem, and explain how it would work.

People notice this, and amplify the signal.

Signal causes people with ability to implement solutions to notice.

Solution is implemented.

Victory.

Except, you know, with planning and strategy involved.

Every time this happens, it becomes more normal to do it again, and it becomes easier to get good solutions amplified and implemented.

We get to thinking we can solve problems, and change things for the better.

Including when the solutions are less and less trivial

Including they require non-zero downsides or non-zero amounts of time

It becomes standard to solve problems and blameworthy to not solve them.

Setting aside for now how much the rule change mattered. For this to be exciting, it must be an asymmetric weapon. Does this work for positive-sum solutions to problems without working, or working less well, for zero-sum resource grabs?

My instinct was clearly yes, this only worked because it was obviously correct. If it had not been obviously correct, the effort fails, for overdetermined reasons. This isn’t obvious to others, and a lot of this post explores the gears involved.

Asymmetric Attention Weapons

I was highly frustrated to see this reaction, which I instinctively (if unfairly) translated as ‘actually, attempting to communicate information that might cause people to do things is bad, because the reference class is zero-sum games that compete for attention’:

It’s important to intuition pump why this is the opposite of most of the attention economy, and why such strategies are asymmetric. The information and solution both being true and important was central to the Tweetstorm’s success.

From my previous post‘s ‘I see what you did there’ list.

Starts with a relatable physical story of a boat ride, and a friendly tone.

Tells a (mostly manufactured) story that implies (without saying anything false) how the ride led him to figure these things out, which gives rhetorical cover to everyone else for not knowing about or talking about the problem. We can all decide to pretend this was discovered today.

Then he invokes social consensus by saying that ‘everyone agrees‘ that the bottleneck is yard space. Which is true, as far as I can tell, everyone did agree on that. Which of course implies that everyone also knows there is a bottleneck, and that the port is backed up, and why this is happening. The hidden question of why no one is doing much about this is deflected by starting off pretending (to pretend?) that the boat ride uncovered the problem.

Describes a clear physical problem that everyone can understand, in simple terms that everyone can understand but that doesn’t talk down to anyone. He makes this look easy. It is not easy, it is hard.

Makes clear that the problem will only get worse on its own, not better, for reasons that are easy to understand.

Makes clear the scope of the problem. Port of Long Beach effectively shuts down, we can’t ship stuff, potential global economic collapse. Not clear that it would be anything like that bad, but it could be.

Gives a decision principle that’s simple, a good slogan and again can be understood by everyone, and that doesn’t have any obvious objections: ...

Welcome to The Nonlinear Library, where we use Text-to-Speech software to convert the best writing from the Rationalist and EA communities into audio. This is: I’ll pay you a $1,000 bounty for coming up with a good bounty (x-risk related), published by EmersonSpartz on November 2, 2021 on The Effective Altruism Forum.

I think bounties and requests for proposals (RFPs) are powerful and underutilized in EA. This is just a quick experiment to see if there are any interesting ideas floating around that might be bounty-able or RFP-able.

If your idea is interesting to me, I’ll shop it around to smarter people with relevant expertise, and if it gets funded, I’ll pay you $1,000.

Ideas can be either be objectively verifiable or they can be subjectively evaluated by a judge. For example, to minimize Goodharting the bounty, you might propose that a panel of AI safety experts judge the bounty submissions.

The deadline is November 15th. If it goes well, I'll open it up to more submissions.

Email me your ideas - rough ones are fine! - at [email protected], DM me on Twitter @EmersonSpartz, or submit them here.

Thanks for listening. To help us out with The Nonlinear Library or to learn more, please visit nonlinear.org.

Welcome to The Nonlinear Library, where we use Text-to-Speech software to convert the best writing from the Rationalist and EA communities into audio. This is: I’ll pay you a $1,000 bounty for coming up with a good bounty (x-risk related), published by EmersonSpartz on November 2, 2021 on The Effective Altruism Forum.

I think bounties and requests for proposals (RFPs) are powerful and underutilized in EA. This is just a quick experiment to see if there are any interesting ideas floating around that might be bounty-able or RFP-able.

If your idea is interesting to me, I’ll shop it around to smarter people with relevant expertise, and if it gets funded, I’ll pay you $1,000.

Ideas can be either be objectively verifiable or they can be subjectively evaluated by a judge. For example, to minimize Goodharting the bounty, you might propose that a panel of AI safety experts judge the bounty submissions.

The deadline is November 15th. If it goes well, I'll open it up to more submissions.

Email me your ideas - rough ones are fine! - at [email protected], DM me on Twitter @EmersonSpartz, or submit them here.

Thanks for listening. To help us out with The Nonlinear Library or to learn more, please visit nonlinear.org.

Welcome to The Nonlinear Library, where we use Text-to-Speech software to convert the best writing from the Rationalist and EA communities into audio. This is: 2020 PhilPapers Survey Results, published by RobBensinger on November 2, 2021 on The Effective Altruism Forum.

In 2009, David Bourget and David Chalmers ran the PhilPapers Survey (results, paper), sending questions to "all regular faculty members" at top "Ph.D.-granting [philosophy] departments in English-speaking countries" plus ten other philosophy departments deemed to have "strength in analytic philosophy comparable to the other 89 departments".

Bourget and Chalmers now have a new PhilPapers Survey out, run in 2020 (results, paper). I'll use this post to pick out some findings I found interesting. Keep in mind that I'm an opinionated LessWrong contributor focusing on topics and results that dovetail with things I'm curious about (e.g., 'why do academic decision theorists and LessWrong decision theorists disagree so much?'), not giving a neutral overview of the whole 100-question survey.

The new survey's target population consists of:

(1) in Australia, Canada, Ireland, New Zealand, the UK, and the US: all regular faculty members (tenuretrack or permanent) in BA-granting philosophy departments with four or more members (according to the PhilPeople database); and (2) in all other countries: English-publishing philosophers in BA-granting philosophy departments with four or more English-publishing faculty members.

In order to make comparisons to the 2009 results, the 2020 survey also looked at a "2009-comparable departments" list selected using similar criteria to the 2009 survey:

It should be noted that the “2009-comparable department” group differs systematically from the broader target population in a number of respects. Demographically, it includes a higher proportion of UK-based philosophers and analytic-tradition philosophers than the target population. Philosophically, it includes a lower proportion of theists, along with many other differences evident in comparing 2020 results in table 1 (all departments) to table 9 (2009-comparable departments).

Based on this description, I expect the "2009-comparable departments" in the 2020 survey to be more elite, influential, and reasonable than the 2020 "target group", so I mostly focus on 2009-comparable departments below. In the tables below, if the row doesn't say "Target" (i.e., target group), the population is "2009-comparable departments".

Note that in the 2020 survey (unlike 2009), respondents could endorse multiple answers.

1. Decision theory

Newcomb's problem: The following groups (with n noting their size, and skipping people who skipped the question or said they weren't sufficiently familiar with it) endorsed the following options in the 2020 survey:

5% of decision theorists said they "accept a combination of views", and 9% said they were "agnostic/undecided".

I think decision theorists are astonishingly wrong here, so I was curious to see if other philosophy fields did better.

I looked at every field where enough surveyed people gave their views on Newcomb's problem. Here they are in order of 'how much more likely are they to two-box than to one-box':

(Note that many of these groups are small-n. Since philosophers of computing and information were an especially small and weird group, and I expect LWers to be extra interested in this group, I also looked at the target-group version for this field.)

Every field did much better than decision theory (by the "getting more utility in Newcomb's problem" metric). However, the only fields that favored one-boxing over two-boxing was ancient Greek and Roman philosophy, and aesthetics.

After those two fields, the best fields were philosophy of cognitive science, applied ethics, metaphilosophy, philosophy of mathematics, and 17th/18th century philosophy (only 4-5% more likely to two-box than one-box), followed by philosophy of religion, normative ethics, and metaphysics.

...

Welcome to The Nonlinear Library, where we use Text-to-Speech software to convert the best writing from the Rationalist and EA communities into audio. This is: 2020 PhilPapers Survey Results, published by RobBensinger on November 2, 2021 on The Effective Altruism Forum.

In 2009, David Bourget and David Chalmers ran the PhilPapers Survey (results, paper), sending questions to "all regular faculty members" at top "Ph.D.-granting [philosophy] departments in English-speaking countries" plus ten other philosophy departments deemed to have "strength in analytic philosophy comparable to the other 89 departments".

Bourget and Chalmers now have a new PhilPapers Survey out, run in 2020 (results, paper). I'll use this post to pick out some findings I found interesting. Keep in mind that I'm an opinionated LessWrong contributor focusing on topics and results that dovetail with things I'm curious about (e.g., 'why do academic decision theorists and LessWrong decision theorists disagree so much?'), not giving a neutral overview of the whole 100-question survey.

The new survey's target population consists of:

(1) in Australia, Canada, Ireland, New Zealand, the UK, and the US: all regular faculty members (tenuretrack or permanent) in BA-granting philosophy departments with four or more members (according to the PhilPeople database); and (2) in all other countries: English-publishing philosophers in BA-granting philosophy departments with four or more English-publishing faculty members.

In order to make comparisons to the 2009 results, the 2020 survey also looked at a "2009-comparable departments" list selected using similar criteria to the 2009 survey:

It should be noted that the “2009-comparable department” group differs systematically from the broader target population in a number of respects. Demographically, it includes a higher proportion of UK-based philosophers and analytic-tradition philosophers than the target population. Philosophically, it includes a lower proportion of theists, along with many other differences evident in comparing 2020 results in table 1 (all departments) to table 9 (2009-comparable departments).

Based on this description, I expect the "2009-comparable departments" in the 2020 survey to be more elite, influential, and reasonable than the 2020 "target group", so I mostly focus on 2009-comparable departments below. In the tables below, if the row doesn't say "Target" (i.e., target group), the population is "2009-comparable departments".

Note that in the 2020 survey (unlike 2009), respondents could endorse multiple answers.

1. Decision theory

Newcomb's problem: The following groups (with n noting their size, and skipping people who skipped the question or said they weren't sufficiently familiar with it) endorsed the following options in the 2020 survey:

5% of decision theorists said they "accept a combination of views", and 9% said they were "agnostic/undecided".

I think decision theorists are astonishingly wrong here, so I was curious to see if other philosophy fields did better.

I looked at every field where enough surveyed people gave their views on Newcomb's problem. Here they are in order of 'how much more likely are they to two-box than to one-box':

(Note that many of these groups are small-n. Since philosophers of computing and information were an especially small and weird group, and I expect LWers to be extra interested in this group, I also looked at the target-group version for this field.)

Every field did much better than decision theory (by the "getting more utility in Newcomb's problem" metric). However, the only fields that favored one-boxing over two-boxing was ancient Greek and Roman philosophy, and aesthetics.

After those two fields, the best fields were philosophy of cognitive science, applied ethics, metaphilosophy, philosophy of mathematics, and 17th/18th century philosophy (only 4-5% more likely to two-box than one-box), followed by philosophy of religion, normative ethics, and metaphysics.

...

Welcome to The Nonlinear Library, where we use Text-to-Speech software to convert the best writing from the Rationalist and EA communities into audio. This is: Transcript: "You Should Read HPMOR", published by TurnTrout on November 2, 2021 on LessWrong.

The following is the script of a talk I gave for some current computer science students at my alma mater, Grinnell College. This talk answers "What do I wish I had known while at Grinnell?".

Hi, I'm Alex Turner. I’m honored to be here under Sam’s invitation. I'm in the class of 2016. I miss Grinnell, but I miss my friends more—enjoy the time you have left together.

I’m going to give you the advice I would have given Alex2012. For some of you, this advice won’t resonate, and I think that’s OK. People are complicated, and I don’t even know most of you. I don’t pretend to have a magic tip that will benefit everyone here. But if I can make a big difference for one or two of you, I’ll be happy.

I’m going to state my advice now. It’s going to sound silly.

You should read a Harry Potter fanfiction called Harry Potter and the Methods of Rationality (HPMOR).

I’m serious. The intended benefits can be gained in other ways, but HPMOR is the best way I know of. Let me explain.

When I was younger, I was operating under some kind of haze, a veil, distancing me from what I really would care about.

I responded to social customs and pressure, instead of figuring out what is good and right by my own lights, how to make that happen, and then executing. Usually it’s fine to just follow social expectations. But there are key moments in life where it’s important to reason on your own.

At Grinnell, I exemplified a lot of values I now look down on. I was extremely motivated to do foolish or irrelevant things. I fought bravely for worthless side pursuits. I don’t even like driving, but I thought I wanted a fancy car. I was trapped in my own delusions because I wasn’t thinking properly.

Why did this happen, and what do I think has changed?

On Caring

First, I was disconnected from what I would have really cared about upon honest, unflinching reflection. I thought I wanted Impressive Material Things. I thought I wanted a Respectable Life. I didn’t care about the bible, but I brought it with me to my dorm anyways so that I’d be more “wholesome” according to my cultural background. Chasing something someone convinced me to believe I wanted, but which I don’t care about.

I became motivated to unironically reflect on what is good, how I want the universe to look by the time I’m done with it—to reason about what matters without asking for permission. Not so that you can show how caring you are on social media. But because some things are fucking important. Peace, learning, freedom, health, justice. Human flourishing. Happiness.

When I inhabit my old ways of thinking about altruism, they evoke guilt and concern: “The world will burn. I have to do my part.” If, however, I’ve discharged my duties by donating and recycling and such, then I no longer feel guilty. But the cruel fact is that no matter what I do, millions of people will die of starvation this year. Due to a coincidence of space and time, none of these people happen to be my brother or sister, my mother or father. None are starving two feet away from me. But who cares if someone starves two feet away, or 42 million feet away—they’re still starving!

What I’m saying here is subtler than “care a lot.” I’m gesturing at a particular kind of caring. The kind from the assigned essay. Since you all read it, I probably don’t need to explain further, but I will anyways. Some extreme altruists give almost everything they have to charity. It’s natural to assume they have stronger “caring” feelings than you do, but that may not be true.

The truth is that I am biologically incapable of caring as much as 9 million x (how much I would care if my brother starved). My internal “caring system” doesn’t go up that many decibels, it just silently throws an emotion overflo...