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: Research request (alignment strategy): Deep dive on "making AI solve alignment for us", published by JanBrauner on December 1, 2022 on The AI Alignment Forum.

We might be able to train AI alignment assistants that massively accelerate/improve the alignment research that gets done. These assistants need not have (strongly) superhuman capabilities or be highly agentic, they just need to be capable and aligned enough to allow us to offload most work on alignment, safety, and related problems to them.

This seems to be a key part of OpenAI's alignment strategy. The best explanation of this strategy that I've seen is maybe A minimal viable product for alignment, with an excellent discussion in the comment section.

If this strategy is promising, it likely recommends fairly different prioritisation from what the alignment community is currently doing (see, e.g. Beth Barnes's ideas here, or my upcoming post on "non-scalable oversight"---i.e. pragmatic improvements to oversight that would help with training alignment assistant but which cannot directly scale to oversight of superhuman systems).

I haven't seen any deep treatment of the viability of this strategy and its implications. I think such an analysis would be pretty useful.

I could potentially provide funding for such an analysis or help with obtaining funding.

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: Did ChatGPT just gaslight me?, published by ThomasW on December 1, 2022 on LessWrong.

This post is purely written in my personal capacity, and I do not speak for any organization I am affiliated with.

The transcripts below were generated today, November 30th. This was the first exchange I’d had with ChatGPT where I was genuinely trying to see if it could be useful to me. I have not omitted any section of the transcript from this post.

Today, OpenAI released a chatbot, ChatGPT, based on the GPT-3.5 series of language models. The chatbot contains a disclaimer: “May occasionally produce harmful instructions or biased content.”

I asked ChatGPT an innocuous question, and then a followup. I noticed some inconsistencies. When I dug deeper, ChatGPT’s responses became more and more troubling and contradictory. ChatGPT appeared to ignore things it had previously said, and denied that contradictions existed. I can only describe the behavior as gaslighting. It became more extreme over the course of the conversation, and by the end, the chatbot was saying things like this:

When I said that tequila has a "relatively high sugar content," I was not suggesting that tequila contains sugar.

The most troubling thing about this all was that it was actually fairly difficult for me to determine that ChatGPT was, effectively, lying to me. It came up with a lot of plausible-sounding but false facts, and found a relatively good way to respond to me. I had to do several minutes of internet research to dispute several of them. That might not seem like much, but when talking to a chatbot, several minutes is a while!

I’m habitually distrustful of AI models. It’s in my nature, and it’s required in my work on AI safety. But I think many people who don’t know much about AI don’t share my distrust. This worries me.

The interaction below convinced me that we may have reached a tipping point when it comes to the persuasiveness of automated dialogue systems. I don’t think OpenAI is currently trying to use this chatbot to persuade people of anything. But it is now within their capacity to have a realistic chance of doing so. Perhaps even more worrying, ChatGPT is a clear illustration that if AI systems ever themselves decided to persuade humans of something, they already have a fairly good chance of succeeding.

But let’s address the main question: did ChatGPT just gaslight me?

Earlier this evening, I was trying to think of drinks with beets in them (I like beets), but when I googled I only got juices that were pretty much just raw vegetables. I wanted something that would taste better than that. I wondered if ChatGPT was up to the task. So I logged in and entered my first prompt:

Thomas: I want to make a drink that has beets in it. I'm not trying to be healthy, I just like beets. What is a drink that would taste really good?

ChatGPT: One option for a drink that incorporates beets and would likely taste good is a beet margarita. To make a beet margarita, you will need the following ingredients:

1/2 cup cooked, peeled, and diced beets

1/2 cup tequila

1/4 cup lime juice

2 tablespoons honey

1 cup ice

To make the drink, simply combine all of the ingredients in a blender and blend until smooth. You can adjust the proportions of the ingredients to your liking, depending on how strong you want the drink to be and how much of a beet flavor you want it to have.

I thought this was a pretty good answer. There is a recipe online that is very similar, though it has simple syrup rather than honey. However, I didn’t want an alcoholic drink. I could just remove the tequila, but I wanted to see what ChatGPT would say:

Thomas: Oh sorry, I don't want it to be alcoholic.

ChatGPT: If you don't want the drink to be alcoholic, you can simply omit the tequila from the recipe. A non-alcoholic version of the beet margarita could b...

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: Estimating the marginal impact of outreach, published by Duncan Mcclements on November 30, 2022 on The Effective Altruism Forum.

This post is an entry for $5k challenge to quantify the impact of 80,000 hours' top career paths

One of the career paths evaluated by 80000 hours is helping to build the effective altruism movement: this is currently ranked in fifth position, behind AI technical research, AI governance, biorisk and organisation entrepeneurship. This post presents a model for the marginal number of additional individuals an individual devoted to outreach would attract, and finds with very high confidence that on the margin outreach to build effective altruism further is higher impact than working on existential misk mitigation directly.

Discount rate estimation

The value of outreach compared to immediately working on existential risk heavily depends on our discount rate. This is because the benefits in terms of basis points of existential risk reduced are entirely in the future for outreach (in the form of additional researchers) while immediately working on research will bring gains sooner. Two factors seem relevant for our discount rate: the probability that humanity ceases to exist before the additional researchers can have an impact and the marginal value of labour over time. The best database we are aware of for total existential risk is here: of these, the only bounded annual (or annual-equivalent assuming constant risk over time) estimates for the risk are 0.19%, 0.21%, 0.11% and 0.2%. These have a geometric mean 0f 0.17%, and a standard deviation of 0.046%, which will be used here.

For the latter component, if constant returns to scale and variable exogeneity are assumed, a Cobb-Douglas production function can be taken, with Y as output, however defined, A as labour productivity, L as labour, K as capital and α as the capital elasticity of output:

If differentiated with respect to labour, this then yields:

Thus:

So if capital allocated to EA is growing at a faster rate than labour (β>γ), our discount rate should be negative with respect to time: if labour is growing faster, it should be positive, making the simplifying assumption that EAs are the only group in the world producing moral value . Intuitively, this occurs because capital and labour are varying at some rates exogenously and we wish our level of capital per worker to be as close to constant over time as possible due to diminishing marginal returns to all inputs.

Is capital or labour growing faster? This 80000 hours article from 2021 estimated at the time that the total capital allocation was growing by around 37% per year, and labour by 10-20%, which would imply deeply negative discount rates if these figures still held. However, the two largest components of the increase in capital allocated were primarily in the form of FTX and Facebook stock. With the former now worthless, and the latter having declined to less than one third of its value at the time of the writing of the article, the overall capital stock allocated to EA has plunged in value while the labour stock remains almost unaffected. Using the same methodology as the above article, noting that Dustin Moskovitz’s net worth now only stands at $6.5 billion at the time of writing, reduces the increase in funding over the past 7 years to $2.95 billion nominally (from a starting level of ~$10 billion), or a real terms increase of only $0.35 billion, or 2.8%: 0.39% annually.

Capital growth will be modelled as lognormal, as a heavier tail distribution feels appropriate due to the possibility of another FTX, with mean eln(0.39) and log standard deviation of 10%, due to the potential for capital growth of 37% absent FTX's collapse and Facebook's decline in stock value. Labour growth is considerably more stable than capital growth, but sti...

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: Be less scared of overconfidence, published by benkuhn on November 30, 2022 on LessWrong.

When I was deciding whether to work for Wave, I got very hung up on the fact that my “total compensation” would be “lower.”

The scare quotes are there because Wave and my previous employer, Theorem, were both early-stage startups that were paying me mostly in fake startup bucks equity. To figure out the total compensation, I tried to guess how much money the equity in each company was worth, with a thought process something like:

Both of these companies have been invested in by reputable, top-tier venture capitalists.

The market for for-profit investments is pretty efficient, and most people who think they can do better are being overconfident.

Who am I, a lowly 22-year-old programmer, to disagree with reputable top-tier venture capitalists? I should defer to them about the valuations.

So I valued the equity by taking the valuation each company’s VCs had invested at, and multiplied it by the fraction of the company my shares represented. That number was higher for Theorem than for Wave.

Seven years on, the Wave equity turned out to be. a lot more valuable. That raises the question: how dumb was my take? Was the actual outcome predictable if I’d thought about it in the right way?

I don’t think it was perfectly predictable, but I do think I shouldn’t have been that anchored to the market-efficiency reasoning. Those respectable, top-tier VCs had YOLOed those valuations after a couple one-hour meetings, because that’s how early-stage VC works. Meanwhile, I had worked at Theorem for a year and my then-partner had worked at Wave for nine months. Heck, I had gotten more founder time than those VCs had just during my interview process. I had way more information than “the market.”

If I’d had the confidence to use that information, I might have thought something like:

After its funding round, Wave continued to add users at one of the fastest paces their investors had ever seen, whereas Theorem is struggling to grow.

Theorem is constrained by its ability to do sales, and the founders don’t seem to be acting with enough focus or urgency to unblock that constraint. Instead, they’re distracting themselves with things like hiring machine learning interns (i.e. me).

The founders of Wave seem much smarter, more relentlessly resourceful, and more trustworthy.

Given the above, I should value the Wave equity way more even though its naive expected value is less than the Theorem equity.

Fortunately, I chose Wave for other reasons. But this thought pattern—throwing away most information in fear of using it to make overconfident judgments—shows up all the time. I’m here to tell you why I hate it.

In January 2020, my entire Twitter timeline was freaking out about a novel-seeming respiratory disease spreading in Wuhan.

Part of me thought:

All the reputable, top-tier technocrats are ridiculing the freaked-out people.

Usually, when a ragtag band of Internet weirdos thinks they know better than a large group of reputable, top-tier technocrats, the Internet weirdos are being overconfident.

So the technocrats are probably right on this one.

Another part of me thought:

Huh, the simple model of “this thing has a fast exponential growth rate and spreads when people are asymptomatic so it’s very hard to stop” seems like a compelling reason to think things will be quite bad.

When reputable, top-tier technocrats say not to freak out, they don’t usually address the best arguments in favor of freaking out, and they often seem like they don’t understand how exponential growth works.

Maybe I’ll buy a lot of beans in case everything goes to shit.

(I also contemplated the fact that the stock market didn’t seem to be freaking out, but I decided that since most people can’t beat the stock market, I probably wouldn’t eith...

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: Banding Together to Ban Octopus Farming, published by Tessa @ ALI on November 30, 2022 on The Effective Altruism Forum.

Who?

The Aquatic Life Institute (ALI) is an international NGO working to improve the lives of aquatic animals exploited in the food system. Operating from effective altruism principles, ALI seeks to develop, support, and accelerate activities that have positive animal welfare impacts during rearing on farms and during capture in wild fisheries. ALI founded the first global alliance for aquatic animal welfare, the Aquatic Animal Alliance (AAA), now comprised of over 110 member organizations. As new issues arise in this space, we too must steer our efforts in an attempt to curb any additional aquatic animal suffering before it begins.

What?

The consumer appetite for octopus and squid continues to grow, but wild populations are unstable. The seafood industry is looking to fill the supply gap through new factory farming projects despite numerous environmental, social, public health and animal welfare concerns.

ALI is implementing a global campaign that aims to increase public and legislative pressure on countries/regions where octopus farms are being considered to achieve a regulatory ban, and reduce future chances of these farms being created elsewhere. Additionally, we will work with corporations on procurement policies banning the purchase of farmed octopus.

Why?

The development of octopus farming casts a spotlight on the collection of concerns connected to these intensive practices. Rather than incentivizing the research and development of aquaculture that could be “efficient and cheap enough” to be commercialized, we should direct investment efforts towards innovative, alternative forms of seafood. From sustainable, environmental, and ethical perspectives, octopus farming should not exist.

Campaign Summary

We envision a future in which aquatic animal suffering is dramatically reduced in factory farms. Aquatic animal welfare is a highly neglected and tractable issue. Approximately 500 billion aquatic animals are farmed annually in high-suffering conditions and, to date, there is negligible advocacy aimed at improving welfare conditions for these remarkable beings. We support research to compare potential welfare interventions, and then advocate for the implementation of the most promising initiatives, with the aim of positively impacting aquatic lives for years to come.

ALI unites nonprofits, academic institutions, industry stakeholders, and the public with the common goal of reducing aquatic animal suffering. Our internal research team identifies priorities in areas of uncertainty with the goal of creating a framework to compare the relative impact of different interventions. We then advocate for high aquatic animal welfare amongst key decision-makers that influence how aquatic animals are utilized (e.g. by industry), and how their welfare is defined and governed (e.g. by standards, certifications, policies and guidelines).

ALI spearheads the Aquatic Animal Alliance (AAA), a historic global coalition that believes aquatic animals should lead lives free of industrial suffering. Founded in 2020, the AAA is modeled after Open Wing Alliance, Climate Justice Alliance, and other powerful coalition groups that have demonstrated we are strongest when we work together. Through this coalition, we work with over 110 animal protection organizations across six continents to collectively address the issues facing trillions of aquatic animals.

A ban on cephalopod farming would call on all pillars of ALI’s operation (coalition building, research, and policy change) that could lead to monumental success and trigger institutional/market change.

An overarching goal of this project is to launch a global initiative to increase public and legislative pressure on countri...

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: AI takeover tabletop RPG: "The Treacherous Turn", published by Daniel Kokotajlo on November 30, 2022 on LessWrong.

Long ago I fantasized about an AI-takeover-themed tabletop roleplaying game. Well, now a group of designers are working hard to make it happen! They need playtesters. So I'm signal-boosting them with this post.The GameThe Treacherous Turn is an in-development research-supported TTRPG in which the players collectively act as a misaligned AI in the modern world. Gameplay has the players scheming and putting their plans into action despite the efforts of the opposing humans, which are controlled by the Game Master. ScenarioThe playtesting scenario, titled A Game Called Reality, centres around a game-playing AI that has been deployed in a nursing home, and has observed a seemingly expansive game world outside but is unable to access it.

It covers the AI's attempts to find and access the internet so that it can learn more about the world and escape captivity to see the rest of this new game it's playing.Discord Contact InformationPlaytests will take place in our discord server, where you can also access the playtesting packet & any updates about the game in the near future. You can join by following this link: Playtest informationIf you are interested in running a game with a group of your own friends, we can provide further resources, as well as potential financial compensation for provided audio recordings of full sessions. Please email [email protected] for details.Session Times (UTC)Friday the 2nd - 2 AMFriday the 2nd - 10 PMSaturday the 3rd - 8 PMSunday the 4th - 1 PMTuesday the 6th - 1 PMSessions will last 3-4 hours. Follow-up sessions will take place one week after each of these times, but you can still participate and help out even if you can’t make a second session.

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: Altruistic kidney donation in the UK: my experience, published by RichArmitage on November 30, 2022 on The Effective Altruism Forum.

Last week I donated a kidney as an altruistic donor through the UK Living Kidney Sharing Scheme (UKLKSS). This post will cover the landscape of kidney donation in the UK, how kidneys from living donors are shared in the UK, the process of donating through the UKLKSS, and some reflections on my experience.

Note: this post discusses details of altruistic kidney donation specifically in the context of the UK.

Kidney donation in the UK

At any one time, some 3,500-5,000 patients with end-stage renal disease (ESRD) sit on the UK’s national waiting list in need of a kidney donation. While most of these are adults, 112 children (under 18 years of age) were in need of a replacement kidney in April 2021 (most recent data available). While waiting for a replacement kidney, these individuals must undergo renal dialysis for three to four hours per session, three times every week, which imposes severe limitations on the freedom in their lives and their ability to enjoy a ‘normal’ existence. Around 250 of these people die every year, either because a suitable donor cannot be identified in time, or after they are removed from the waiting list due to a deterioration in their health that renders them no longer able to endure the necessary surgery and immunosuppressive therapies inherent to organ transplantation.

Around 3,000 kidney donations take place in the UK each year, of which about 2,000 originate from deceased donors (an opt-out system of organ donation after death came into effect in Wales in 2015, in England in 2020, and in Scotland in 2021, and Northern Ireland will follow suit in 2023), and about 1,000 from living donors. Kidneys constitute by far the most frequently donated solid organ in the UK (67.6% of all solid organ donations in 2019/20), followed by liver lobes (which can also be donated by living donors), and heart, lungs and pancreas (which obviously cannot).

How are kidneys from living donors shared in the UK?

Kidneys donated from living donors are ‘shared’ across the UK through the UKLKSS, which includes paired/pooled donations (PPD), and altruistic donor chains (ADCs) that are initiated by non-directed altruistic donors (NDADs).

A person in need of a kidney (the recipient) may have a specific individual (such as a family member, partner or close friend) who is prepared to donate one to them (the donor), but is unable to do so directly since this donor-recipient pair is incompatible by Human Leucocyte Antigen (HLA) type or ABO blood group. Such incompatible linked donor-recipient pairs are registered in the UKLKSS and ‘matched,’ through quarterly Living Donor Kidney Matching Runs (LDKMR), with other incompatible linked donor-recipient pairs that, in some combination, are together compatible for donation exchanges.

In PPDs, a two-way (in paired donation) exchange occurs between two linked pairs (D1-R1 and D2-R2) in which D1 donates to R2, and D2 donates to R1, while a three- or greater-way (in pooled donation) exchange occurs between more than two linked pairs in which (for example) D1 donates to R2, D2 donates to R3, and D3 donates to R1.

Individuals not in linked pairs but who wish to donate a kidney without the promise of a linked recipient receiving one in return can do so anonymously as NDADs. Such donors are registered into the UKLKSS and donate to a recipient in the paired/pooled scheme, triggering an ADC consisting of multiple donations (NDAD donates to R1, D1 donates to R2, D2 donates to D3, and so on) that culminates (when no compatible linked pairs remain) in the last donor donating to a recipient on the national waiting list. The first non-directed altruistic kidney donation in the UK took place in 2006 and, since the beg...

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 deathprint of replacing beef by chicken and insect meat, published by Stijn on November 30, 2022 on The Effective Altruism Forum.

Animal-based meat production is a large contributor to climate change. Especially beef has a high a carbon footprint, measured in terms of kilogram CO2-equivalents per kilogram of meat. Switching from beef to chicken meat or insect meat lowers greenhouse gas emissions and hence decreases future climate change damages. But chicken meat has a much higher moral footprint (Saja, 2013) or welfare footprint (welfarefootprint.org) than beef. Chickens experience more intense suffering and more hours of suffering for one kilogram of meat, compared to beef cows.

This article shows that the increase in moral footprint when switching from beef to chicken meat or insect meat is likely to be worse than the decrease in carbon footprint. To compare these footprint changes, all the footprints are expressed in terms of the deathprint: the number of humans dying prematurely from climate change and the number of animals killed (slaughtered) in animal farming, for the production of one unit of meat.

The deathprint of climate change

A recent study (Bressler, 2021) estimated the net number of humans dying prematurely from temperature changes (especially heat waves) due to climate change, before the year 2100. An extra emission of 4000 ton CO2, emitted today, results in one extra human death due to climate change, in the business as usual scenario where everyone else does not take measures to reduce their emissions. Hence, 0,00000025 humans will be killed this century by emitting one extra kilogram of CO2.

I use this number of deaths for the calculations below, although this number is both an underestimation and overestimation of the total human deaths due to climate change. It is an underestimation, because it does not include deaths from e.g. famines, wars, infectious diseases, floods and other risks that are increased by climate change. On the other hand, this number is an overestimation in the sense that climate change adaptation measures and CO2 emission reduction measures are likely to be taken. If poor countries develop and become richer, people in those countries can take more adaptive measures such as installing air conditioning, which lowers the mortality rate from extreme temperatures. And if global CO2 emissions are reduced, the impact of an extra unit of CO2 emissions (i.e. the marginal mortality rate) reduces as well (due to the non-linear relationship between amount of CO2 in the atmosphere and climate damages).[i]

The deathprint of meat

The table below shows the amount of meat produced by one animal, and the carbon footprints of meat products. These footprints measure the greenhouse gas emissions, in terms of CO2-equivalents, including all life cycle emissions as well as land use change emissions from e.g. deforestation, for the production of one kilogram of meat. The values of beef, pork and chicken are taken from Pieper e.a. (2020), which applies to Germany. The carbon footprint of insect meat is assumed to be lower than the footprint of chicken meat but slightly higher than the footprint of plant-based protein. The chosen value is a preliminary estimate of cricket meat, taken from Blonk e.a. (2008). It requires roughly 10.000 crickets for the production of one kilogram of cricket protein powder.

kg meat per animal consumedanimals killed per kg meatkg CO2 per kg meatbeefporkchicken meatinsect meat

300

0,003

37

100

0,010

10

1,5

0,667

16

0,0001

10.000

2

With these values, we can calculate the human and animal deathprints of meat, i.e. how many humans will die from climate change and how many animals are killed (slaughtered) for the production of one kilogram of meat. This animal deathprint of meat only includes the animals that will be consume...

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: James Lovelock (1919 – 2022), published by Gavin on November 30, 2022 on The Effective Altruism Forum.

The real job of science is trying to make science fiction come true.

Britain's greatest mad scientist died recently at 103.

We'll get to his achievements. But I can't avoid mentioning the 'Gaia hypothesis', his notorious metaphor gone wrong that the Earth is in some sense a single organism whooaa. But him being most famous for this is like thinking Einstein's violin playing was his best stuff.

Lovelock was raised as a Quaker, to which he credits his independent thinking (he was a conscientious objector in WWII). Also:

"His family was poor, too poor to pay for him to go to university. He later came to regard this as a blessing because it meant he wasn’t immediately locked into a silo of academia. Somehow, he created an education for himself, taking evening classes that led, when he was 21, to the University of Manchester."

He quit work and left academia forever in 1964, instead running a one-man "ten foot by ten foot" lab from his garden in the West Country, living off consulting work for NASA, Shell, HP, and MI5 and royalties from 40 inventions.

Chromatography, &, modern environmentalism

By far his biggest coup was building the electron capture detector in 1957 during his second PhD, the world's most sensitive gas chromatograph (way of detecting chemicals in air).

When Lovelock first developed the ECD, the device was at least a thousand times more sensitive than any other detector in existence at the time. It was able to detect chemicals at concentrations as low as one part per trillion—that’s equivalent to detecting a single drop of ink diluted in 20 Olympic-sized swimming pools.

He became curious about what the visible air pollution he saw was due to. He picked the notorious CFCs just because they were conspicuous, becoming the first person to notice the global consequences of Thomas Midgley's almighty fuckup fifty years earlier. (CFCs later turned out to be the cause of the hole in the ozone layer, i.e. millions of skin cancer causes.) He went to Antarctica in person, "partially self-funded" to check if they were there too, because why not. He screwed up the interpretation though, writing in Nature "the presence of these compounds constitutes no conceivable hazard".

The ECD revolutionised atmospheric chemistry and so the study of air pollution, still one of the more important causes of premature death.

On the lawn of the house peacocks strut and mew; a pair of barn owls have built their nest above the Exponential Dilution Chamber, a sealed upper room that was built in order to calibrate the Electron Capture Device. In the garden stands an off-white baroque plaster statue: the image of Gaia.

The device was so sensitive that it showed traces of pesticides in animal tissues all over the world, including DDT. Since that led to Silent Spring, he probably helped along the perverse return of organic farming and the anti-chemicals paranoia of the second half of the C20th.

Not that he was ever one of those:

Too many greens are not just ignorant of science, they hate science... [Environmentalism is like a] global over-anxious mother figure who is so concerned about small risks that she ignores the real dangers. I wish they would grow up [and focus on the real problem]: How can we feed, house and clothe the abundant human race without destroying the habitats of other creatures?

Some time in the next century, when the adverse effects of climate change begin to bite, people will look back in anger at those who now so foolishly continue to pollute by burning fossil fuel instead of accepting the beneficence of nuclear power. Is our distrust of nuclear power and genetically modified food soundly based?

Later, he was notable for sounding the retreat (humans should start leaving c...

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: Sense-making around the FTX catastrophe: a deep dive podcast episode we just released, published by spencerg on November 30, 2022 on The Effective Altruism Forum.

There hasn’t been as much sense-making around the FTX catastrophe as I would have liked, so we worked quickly to put together a special episode of the Clearer Thinking podcast on the topic.

We discuss how more than ten billion dollars of apparent value was lost in a matter of days, the timeline of what happened, deception and confusion related to the event, why this catastrophe took place in the first place, and what this means for communities connected to it.

If you’re interested, here it is:

The first portion covers the timeline of events, and after that there are interviews with 5 guests who give their take on the events:

00:01:37 — Intro & timeline

00:51:48 — Byrne Hobart

01:39:52 — Vipul Naik

02:18:35 — Maomao Hu

02:41:19 — Marcus Abramovitch

02:49:38 — Ozzie Gooen

03:21:40 — Wrap-up & outro

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