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: 300+ Flashcards to Tackle Pressing World Problems, published by AndreFerretti on July 11, 2022 on The Effective Altruism Forum.

How often did you read an article to then forget it? With these 300+ flashcards, you can memorize key facts and test your knowledge of pressing world problems such as animal welfare, global health, longtermism, and meta-effective altruism. Get these flashcards at your fingertips from Thought Saver in the four embedded quizzes below, or download the Anki decks at the bottom.

[Note: Sources are in the Thought Saver/Anki flashcards, as I didn't want to fill this post with links. Also, applying and understanding is more valuable than memorizing facts—see Bloom's Taxonomy. My favorite sources to dig deeper are Our World in Data, 80,000 Hours' problem profiles, and Doing Good Better by William MacAskill.]

Animal Welfare

1% of US animal donations went to farmed animal organisations in 2015

3,000 US farm animals were killed for every shelter animal death in 2015

3% of US donations were aimed at helping the environment or animals in 2020

75% of agricultural land is used for livestock (including grazing and land to grow animal feed)

30B chickens were alive in 2020

A centralized nervous system is what enables animals to have experiences

Animal shelters received 65% of US animal donations in 2015

Broiler chickens in factory farms are usually killed when 6 weeks old

China is the country that produces the most meat in tons

China is the country that produces the most pigmeat

China is the country that produces the most seafood

Clean meat Meat grown in cell culture rather than in an animal's body

Cows eat 6 calories for each calorie of beef produced

Does the majority of seafood production come from wild fish catch or fish farming?Fish farming

Eating 300 eggs indirectly kills one chicken

Eating 3,000 calories of chicken meat kills one chicken

Effective animal campaigns affect 10-100 chicken-life years per $ spent

Farmed hens make 20 times as many eggs as they were born to do

Fish farming has increased 50-fold globally from 1960 until 2015

Flexitarian A person who has a primarily vegetarian diet but occasionally eats meat or fish

foodimpacts.org An online tool ranking animal foods based on suffering and emissions

Global meat production grew 200% between 1970-2020

Humanity farmed 1 trillion insects in 2020

Humanity killed 100B farmed fish in 2017

Humanity killed 2,000 land animals every second in 2016

Humanity killed 300B farmed shrimp in 2017

Humanity killed 69B farmed chickens for meat in 2018

Humanity killed 70B land animals in 2016

Less than 0.1% of global donations are aimed at helping farmed animals

Mark Post developed the first cultured meat hamburger in 2013

Open Wing Alliance A coalition aiming to eliminate battery cages for chickens

Over 90% of global farm animals lived in intensive farms in 2018

Peter Singer wrote Animal Liberation

Speciesism Treating members of one species as morally more important than members of other species

Switching to a plant-based diet spares 100 vertebrates every year (mostly fish and chicken)

The average meat consumption per capita in China has grown 15-fold since 1961

The average meat consumption per capita in China was 60kg in 2017

The average meat consumption per capita in India was 4kg in 2017

The average meat consumption per capita in the United States was 120kg in 2017

The average global meat consumption per capita has grown from 20kg to 40kg between 1961-2014

The EU will ban cages for farmed chickens by 2027

The Humane League launched the Open Wing Alliance

The United States is the country that produces the most cattle and poultry

Three foods to avoid that remove the most animal suffering from your diet-Chicken -Eggs -Fish

Top Charities recommended by Animal Charity Evaluators in 2021 -Faunalytics -The Humane League ...

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: How do AI timelines affect how you live your life?, published by Quadratic Reciprocity on July 11, 2022 on LessWrong.

This question is more about personal decision-making rather than for example deciding to work on AI safety for altruistic reasons. If I were thinking about this from a purely selfish perspective, it seems pretty likely that if I expect transformative AI to arrive in 20 years, I should live my life a bit differently than people who don't expect to get TAI in their lifetimes. I'm curious about if your own beliefs on AI timelines have affected anything that you do in your personal life - perhaps decisions related to saving money, personal relationships, health etc.

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: Does biodiversity loss warrant being it’s own cause area?, published by emwalz on July 10, 2022 on The Effective Altruism Forum.

The impacts of biodiversity loss is not a well-researched topic in EA. A small group of us have begun doing some research into the impacts of biodiversity loss (including mass extinction and its connection to ecosystem collapse) as a potential cause area that deserves more attention and resources.

After preliminary research we found very little active research within the EA community about this topic and only minor explorations from core EA organizations. If you know of substantial, recent research on this topic we would love to read it.

As a quick way to help our community by contributing to an under-explored topic, we invite you to fill out this 3 minute survey or respond in the comments to the question:

What is the strongest argument you know of for & against biodiversity loss as a cause area?

We also want to acknowledge that we have seen substantial research into climate change. We believe biodiversity loss warrants its own targeted research as a separate but related trigger for compounding devastating risk for life on Earth.

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: The Alignment Problem, published by lsusr on July 11, 2022 on LessWrong.

Last month Eliezer Yudkowsky wrote "a poorly organized list of individual rants" about how AI is going to kill us all. In this post, I attempt to summarize the rants in my own words.

These are not my personal opinions. This post is not the place for my personal opinions. Instead, this post is my attempt to write my understanding of Yudkowsky's opinions.

I am much more optimistic about our future than Yudkowsky is. But that is not the topic of this post.

Humanity is going to build an AGI as fast as we can.

Humanity is probably going to build an AGI, and soon.

But if an AGI is going to kill us all can't we choose to just not build an AGI?

Nope! If humanity had the coordination ability to "just not build an AGI because an AGI is an existential threat" then we wouldn't have built doomsday weapons whose intended purpose is to be an existential threat.

The first nuclear weapon was a proof of concept. The second and third nuclear bombs were detonated on civilian targets. "Choosing not to build an AGI" is much, much harder than choosing not to build nuclear weapons because:

Nukes are physical. Software is digital. It is very hard to regulate information.

Nukes are expensive. Only nation-states can afford them. This limits the number of actors who are required to coordinate.

Nukes require either plutonium or enriched uranium, both of which are rare and which have a small number of legitimate uses. Datacenters meet none of those criteria.

Uranium centrifuges are difficult to hide.

Nuclear bombs and nuclear reactors are very different technologies. It is easy to build a nuclear reactor that cannot be easily converted into weapon use.

A nuclear reactor will never turn into a nuclear weapon by accident. A nuclear weapon is not just a nuclear reactor that accidentally melted down.

Nuclear weapons are the easiest thing in the world to coordinate on and (TPNW aside) humanity has mostly failed to do so.

Maybe people will build narrow AIs instead of AGIs.

We [humanity] will build the most powerful AIs we can. An AGI that combines two narrow AIs will be more powerful than both of the narrow AIs. This is because the hardest part of building an AGI is figuring out what representation to use. An AGI can do everything a narrow AI can do plus it gets transfer learning on top of that.

Maybe building an AGI is really hard—so hard we won't build it this century.

Maybe. But recent developments, especially at OpenAI, show that you can get really far just brute forcing the problem with a mountain of data and warehouse full of GPUs.

The first AGI will, by default, kill everyone

Despite AGIs being more useful than narrow AIs, the things we actually use machine learning for are narrow domains. But if you tell a superintelligent AGI to solve a narrow problem then it will sacrifice all of humanity and all of our future lightcone to solving that narrow problem. Because that's what you told it to do.

Thus, every tech power in the world—from startups to nation-states—is racing as fast as we can to build a machine that will kill us all.

But what if whoever wins the AGI race builds an aligned AGI instead of an unaligned AGI?

Almost nobody (as a fraction of the people in the AI space) is trying to solve the alignment problem (which is a prerequisite to building an aligned AI). But let's suppose that the first team of people who build a superintelligence first decide not to turn the machine on and immediately surrender our future to it. Suppose they recognize the danger and decide not to press "run" until they have solved alignment.

AI Alignment is really hard

AI Alignment is stupidly, incredibly, absurdly hard. I cannot refute every method of containing an AI because there are an infinite number of systems that won't work.

AI Alignment...

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: EA for dumb people?, published by Olivia Addy on July 11, 2022 on The Effective Altruism Forum.

I've been involved in EA for nearly a year now. At first, it was super exciting. I resonated so much with the core ideas of EA, and I couldn't wait to get started with doing the most good I possibly could. I had no idea there was so much opportunity.

As I got further into it, my hopes started to fade, and I started to feel like I didn't really fit in. EA is pitched to the super intelligent in our society, those who did super hard degrees at Oxford or Harvard and learned to code at age 8. For me, I'm just average. I never stood out at school, I went to mid-ranking university and studied sociology (which has a reputation for being an easy degree). I graduated, got an average job and am living an average life. I don't have some high earning side hustle and I don't spend my spare time researching how we can make sure AI is aligned with human values.

I do however, care a lot about doing the most good. So I really want to fit in here because that matters a lot to me. I want to leave the world a better place. But I feel like I don't fit, because frankly, I'm not smart enough. (I'm not trying to be self deprecating here, I feel like I'm probably pretty average among the general population - and I didn't really ever feel 'not smart enough' before getting involved in EA)

I totally understand why EA aims at the Oxford and Harvard graduates, of course, we want the most intelligent people working on the world's most pressing problems.

But most people aren't Oxford or Harvard graduates. Most people aren't even university graduates. So do we have a place in EA?

I want to be a part of this community, so I'm trying to make it work. But this leads me to be worried about a lot of other people like me who feel the same. They come across EA, get excited, only to find out that there's not really a place for them - and then they lose interest in the community. Even the idea of giving 10% of your salary can be hard to achieve if you're balancing the needs/wants of others in your family (who maybe aren't so EA minded) and considering the rises in the cost of living currently.

I'm guessing here, because I have absolutely no stats to back this up and it's based on mostly my anecdotal experience - but we could potentially be losing a lot of people who want to be a part of this but struggle to be because EA is so narrowly targeted.

Whenever I come on the EA forum I literally feel like my brain is going to explode with some of the stuff that is posted on here, I just don't understand it. And I'm not saying that this stuff shouldn't be posted because not everyone can comprehend it. These are really important topics and of course we need smart people talking about it. But maybe we need to be aware that it can also be quite alienating to the average person who just wants to do good.

I don't have a solution to all this, but it's been on my mind for a while now. I re-watched this Intro to EA by Ajeya Cotra this morning, and it really re-invigorated my excitement about EA, so I thought I'd put this out there.

I'd be really keen to hear if anyone has any thoughts/feelings/ideas on this - I'm honestly not sure if I'm the only one who feels like this.

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: My EA Failure Story, published by semicycle on July 11, 2022 on The Effective Altruism Forum.

Content warning: depression, failure.

...If you want to do as much good in expectation as possible, ideally you should be aiming for those best-case, [high risk, high return] outcomes. And probably, you’ll fail — more likely than not.¹

An Engineer's Attempt at a Sales Pitch

Do you worry that selection effects are blinding you to the reality that most ambitious EA's have mediocre careers? Worry no more as you hear the depressing story of how I did not become like Sam Bankman-Fried or the Moskovitz's.

"If you shoot for the moon..."

At some point in early childhood development, well-adjusted children realize that the phrase, "you can do anything you set your mind to," isn't meant to be taken literally. Yes, you should take "personally responsibility" for cleaning your room, but you could not have stopped -- nor are you "personally responsible" for -- the mess caused by the collapse of the Twin Towers.²

Poorly-adjusted over-achiever that I was, I decided to bear the entire world's suffering on my shoulders. Somewhere in my teens, I concluded that my path to ameliorate suffering was earning-to-give: I would become a billionaire entrepreneur and give my money away. The only way to fail was to not try hard enough -- reasonable expectations, amirite?

"...you land on your head and break your neck"

Shockingly, I didn't become a billionaire college dropout like I planned. Nor did I do well enough in undergrad to pursue my MBA, so the backup plan was shot too. However, by working harder than everyone else, I managed to make slightly more money than my engineering peers.

In 2013, a few years into my career, I discovered GiveWell. Fortunately, I could now save people's lives even before becoming a billionaire. Unfortunately, every frivolous purchase sentenced someone to death-by-malaria. The upshot was that I found it easy to live below my means, donate 10-20% of my earnings, and save up for the day when I would launch a successful start-up.

(If anyone is still reading this, you're probably thinking: this guy needs professional help. More on that later.)

I learned the hard way that most people don't want to risk their life's savings on a start-up, and I failed to find a co-founder.³ Undeterred, I struck out on my own. With 13 months of hard work, I built a thriving business had absolutely nothing to show for it. Apparently, I inherited my family's "entrepreneurial spirit," but my phenotype was 100% engineer and 0% salesman (which you guessed from my sales pitch above... or not because you're not reading this...). It took me a surprisingly long time, but eventually I learned that,

At this point in the story, I'm on the backup plan of the backup plan of my earning-to-give strategy, but I thought maybe I could climb the ladder as a manager. I loved leading teams and always did right by them. Two years and two "Manager" titles later, I was forced to conclude that either (a) I was incapable of politics or (b) I hated politics too much to try. Regardless, my career had peaked in my 30's, and the path to billionairedom had officially dried up.

Failed Attempts at Direct Work

Meanwhile EA funds continued to grow, and 80k kept talking about talent constraints and how some donors had more money than opportunities, so I began considering direct work more seriously. However, I don't live in an EA hub like SF or London, and pre-COVID very few EA jobs were remote-friendly. I interviewed with one organization anyway, but they confirmed that I would need to move, which wasn't possible for family reasons. (Did I mention I got married and had kids? They're great!)

I tried contributing some bio-security research on my own, but I couldn't get anyone's attention, which in retrospect wasn't surprising.

Post-pan...

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: Meat Externalities, published by Richard Y Chappell on July 11, 2022 on The Effective Altruism Forum.

[Thanks to Pablo S. for suggesting that I cross-post this here.]

A year ago, Scott Alexander argued that it’s better to eat beef than chicken, on the (interesting but questionably relevant) grounds that it’s vastly cheaper to offset the climate costs of beef than the animal welfare costs of chicken:

Eating beef causes more climate change than eating chicken, but eating chicken causes more animal suffering than eating beef. Offsetting the climate change effects of beef would only cost $22 per year, which seems really good. Offsetting the animal suffering effects of chicken might only cost $360 per year, but this is a very tentative estimate and maybe shouldn't be taken seriously. Also, these only work if you're actually doing the offsetting. If not, you should probably default to eating beef over chicken, but I can't prove it.

So I was excited to hear Kevin Kuruc’s talk on ‘Monetizing the Externalities of Animal Agriculture’ (you can find the full paper on his website), extending standard economic models of the “social cost of carbon” to additionally estimate the “social cost” of factory-farmed animal suffering. The headline result:

We find that the welfare costs of global animal agriculture are very large in the case that animals do not have net-pleasurable existences: the monetized costs of producing the meat consumed for the Standard American Diet (SAD) for one person is on the order of $100,000 per year under our baseline parameters. In other words, eliminating the production of meat required for one individual’s diet for one year confers social welfare benefits equal to the benefits of increasing annual global output by more than $100,000.

While this particular number is highly sensitive to the baseline parameters of the model, the broader conclusion that animal welfare costs completely swamp the climate costs of eating meat turns out to be almost unavoidable once you grant that factory-farmed animal lives are net-negative.

Note that, while the animal welfare costs of the Standard American Diet are on the order of $100,000 per year, the climate costs are a mere $47. Combining these with Scott Alexander’s estimates of offsetting efficacy yields some interesting results. Suppose for simplicity that the SAD involves equal parts beef and chicken. (This is presumably incorrect, so take the following numbers with a large grain of salt.) $47 worth of climate costs could then be prevented with a mere $11 to effective climate charities. (Good value!) Whereas >$100,000 worth of animal suffering costs could be prevented with ~$180 donated to effective animal suffering charities. (Insanely good value!)

I should re-emphasize that these numbers are unreliable. But they indicate at least some back-of-the-envelope reason to expect effective animal charities to do (vastly!) more good per $ than climate charities. And even when replacing all chicken with beef, the welfare costs still dwarf the climate costs by many orders of magnitude. So for reasons of both efficacy and reparative justice, it seems that meat eaters should eat (only) beef and then prioritize further offsetting the animal welfare costs of being non-vegetarian, rather than merely offsetting the (comparatively trivial) climate costs.

So that’s an interesting result. One thing I’d be curious to hear more about is how the ~$100,000 in negative externalities compares to the (i) intrinsic value, and (ii) positive externalities of a Standard American year of life. IIRC, the value of a statistical life-year in the US is itself on the order of $100,000. So the harm of a Standard American Diet roughly balances out the intrinsic value of a standard American life (on standard economic approaches). If you’re neither vegetarian ...

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: Information Design of the Library of Effective Altruism (1/2), published by annaleptikon on July 10, 2022 on The Effective Altruism Forum.

This post describes the reasons for, process, and result of creating an information design poster of the essential books on effective altruism. Fundamentally this is an attempt at a visualization of an effective altruism library.

However, this is only half the process. The current design is meant as an approximation, as a surface to collectively disagree with and to add perspectives and knowledge. It was designed with Cunningham's Law in mind, which states that the best way to get the right — if not any — answer on the internet is not to ask a question; it's to post the wrong answer. Once the comment and survey activity subsides I will synthesize the information into the final design (and second post).

The final design will be a digital document with hyperlinks or a physical scratch-off poster.

Background

Goals

The goal of this endeavor is an information visualization that represents the most important books on and related to effective altruism. This product then affords various functions; it can, e.g., get printed as a scratch-off poster that can be handed out at introductory events and efficiently give a first overview of the knowledge space of effective altruism. Part of the value of the artifact is, however, the collective process of creating it, namely the creation of a first draft and external representation of the knowledge space, offering a surface to quickly show gaps or disagreements, which can then be integrated into the next version.

Funding

With these goals in mind, the project was funded by a one-month salary from CEA to conduct the scratch-off reading list. More complex information design projects will potentially be conducted. This will depend on how well this project is received and what was learned from the process of creating it. The reception of this project will be judged by how widely it will get used by different organizations and by the engagement with this EA Forum article. Ideas for what to visualize next are welcome.

Intentions

Why is creating an information visualization poster of books on effective altruism a meaningful endeavor? It helps to resolve many different issues, offers various affordances, and, accordingly, is a robustly valuable case:

An easy way for new aspiring effective altruists to get an insight into the knowledge space of effective altruism. This process of getting accustomed to the ideas, as you might know, can quickly be very overwhelming, as it is simply impossible to know and read it all, and one might fall down one particular rabbit hole and miss out on the big picture. This design can offer precisely this grasp of the big picture.

By Matthew Barnett and Jackson Wagner in the EA Dank Memes Facebook Group

It can be a guiding light in a self-administered curriculum for studying effective altruism-related resources, either by reading the most introductory books from the list first or by taking the book that is currently the most far away from one’s current thinking. The self-administered study is still relevant for many people who don’t have the privilege to live in a place with a developed EA community.

The poster can function as a form of a quality control measure in effective altruism education, assuring that the breadth of its thinking is represented in an easily accessible way so that everyone shares some basic knowledge. Deeming this critical stems from my personal experience. I have repeatedly met people who considered themselves effective altruists but didn’t know ideas I and many others would consider foundational. I’m talking about “Who is Esther Duflo? Who is Peter Singer?”. It is OK to disagree with all foundational ideas. However, the lack of awareness was shocking and m...

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: What I mean by "funding overhang" (and why it doesn't mean money isn't helpful), published by Owen Cotton-Barratt on July 10, 2022 on The Effective Altruism Forum.

I was a bit surprised by this post arguing against using the term "funding overhang". In part it made me realise that others are maybe thinking of the term differently than I am. I'm not sure whether we should stop using the term, but I thought I'd quickly write-up my understanding of the term, so we could have that as background for the conversation.

My basic picture is that there's some flow of people turning up willing to do projects, and some flow of money turning up willing to fund them. Then there's a matching process, and some work happens.

If they're roughly balanced, increasing the flow of either has some effect. Increasing the flow of labour, as well as displacing less talented labour, may also drive wages down, and result in more total work. Increasing the flow of money can help to fund reasonable marginal people, or attract more people to the field.

In the case of a labour overhang, there are many more qualified people turning up than there is money to support them. This means that:

Most of the people turning up end up not getting funded, and going away to do something else instead

More people turning up can improve the quality of the work done, but they don't significantly increase the total quantity of people employed

More money turning up can easily increase the total quantity of people employed (it usually has a smaller proportional effect on the total work output, because the top people were likely to get employement in any case)

I think that many academic fields have something of a labour overhang (e.g. lots of competition for PhD slots, and lots of competition for professorships among PhDs).

In the case of a funding overhang, there's significantly more money turning up willing to pay for good projects than there are people to execute on them. This means that:

Most of the money turning up ends up not funding anything

It's not (usually) wasted! But redeployed to other areas or saved for better opportunities

More money turning up doesn't significantly affect the number of people employed (it can have some affect, by boosting salaries)

More people turning up can easily increase the total quantity of people employed (since sufficiently good people with good projects can usually find funding without replacing other people getting funded)

I think that direct work in longtermist EA is in something of a funding overhang regime at the moment, and has been for a while (but wasn't 10 years ago), and will be for a while (but likely not indefinitely).

Does this mean money isn't helpful for longtermist EA? No!

The most robust reason is that extra money will typically not be deployed this year, but held for deployment in some future year. We're likely at some point to stop being in a funding-overhang situation, and then marginal money will be important. (It's possible to stop being in a funding-overhang situation even while there are large amounts of money left in the war-chest that aren't being spent, since people spending the chest have to make judgement calls about the margins of allocation today vs in future years.)

Another reason that funding overhangs don't necessarily mean that money isn't helpful is that there are multiple things we might spend money on. It could well be that one domain (like hiring alignment researchers into well-run orgs) has a funding overhang, while another still-useful domain can absorb a lot of money (even if marginal hires are less useful) and doesn't have a funding overhang.

But if there is a funding overhang for everything you might donate to, at least that means that you shouldn't donate to object-level charities today? Still no!

As a floor on how good it is to give, i...

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: Avoid the abbreviation "FLOPs" – use "FLOP" or "FLOP/s" instead, published by Daniel Eth on July 10, 2022 on LessWrong.

Especially in discussions about AI, the abbreviation "FLOPs" is being used for both "floating point operations per second" (a measure of computational power) and "floating point operations" (a measure of total computations, and equivalent to the previous term times seconds). This is ambiguous and confusing. For clarity, I propose people avoid this specific abbreviation and instead use the alternatives of "FLOP" (for floating point operations) and "FLOP/s" (for floating point operations per second).

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