The Nonlinear Library

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: Introducing Hive: A rebrand for Impactful Animal Advocacy, published by Hive (formerly Impactful Animal Advocacy) on May 7, 2024 on The Effective Altruism Forum. Last month, we made an April Fools Day post about rebranding to Shrimpactful Animal Advocacy. It was a two-sided joke, because we are rebranding. As of today, Impactful Animal Advocacy is now Hive. Our new name won't have the word "impactful" in it - but it's still an important part of our mission: to end factory farming through creating impact-focused communities. We are excited to announce that Impactful Animal Advocacy is now Hive, the same organization with a fresh look and feel. After much thought and consideration, we envision Hive to better reflect who we are - capturing our collaborative spirit and commitment to generating impact within the farmed animal advocacy movement. Why rebrand? This change is, in part, a response to common feedback that the name Impactful Animal Advocacy is too long (10 syllabus!), hard to remember, and difficult to recognize as the acronym IAA. We want to empower animal advocates to do work that's effective and impactful. And if they are, we want them to feel comfortable calling their work impactful animal advocacy without worrying about overstepping into an org's name. Why we chose Hive We want to convey a sense of friendliness and approachability, combined with a rational and results-driven approach. Additionally, we wanted to avoid using an acronym if possible. Name: Hive. The word hive is a place full of activity, and in the context of bees, it's a place where many gather. Hive symbolizes collaboration and coordination, while still bearing a subtle animal element. Logo: Hexagon with 3 typing indicator dots. The hexagon is a reference to bee hives. The three typing indicator dots are common on digital platforms, and help explain that we are a largely online community that thrives on communication. Colors: Deep and light orange gradient. We think these warmer colors will help establish the brand as a place for fresh ideas and energy. The fluid nature of the gradient represents the diversity of viewpoints and collaborative nature of our community. How this change affects our community or programs You will see that our name, logo, and website will be changed. Besides that, our programs (Slack, newsletter, etc) will run as usual. Other changes that you may see, such as Community Calls, are not related to the rebrand and are instead part of our continuous improvements. We thank many people and organizations: Vegan Hacktivists for their initial logo design and brand kit - it helped us get started quickly and we are thankful this service helps other orgs get off the ground also. Our trusted advisors and mentors have been critical in supporting us through all the organizational growth we experienced in 2023. Thank you to Cameron King, David Meyer and Tracy Spencer, Rockwell Schwarz, Aaron Boddy, Nicoll Peracha, Jay Barrett, Rachel Atcheson, Jacob Eliosoff, Monica Chen, David Nash, Steven Rouk, Wanyi Zeng, David Coman-Hidy, Tania Luna, Alyssa Greene-Crow, Nicole Rawling and Ana Bradley. You have all been so helpful and we appreciate your time. Our funders and donors who have helped not just with funding but with valuable advice that helped us create and follow a focused and solid strategy, including Open Philanthropy, Lush, Veg Trust, Craigslist Charitable Fund, Humane America Animal Foundation, and all the individual donors who believed in us and supported us so early in our journey. James and Amy Odene from User-Friendly who helped us come up with a new name and design the Hive logo and website. You have been incredibly helpful and patient in this process. We love the end product! All of our loyal Slack members - for being the impactful and helpful advocates that you are! We are...

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 moral ambiguity of fishing on wild aquatic animal populations, published by MichaelStJules on May 7, 2024 on The Effective Altruism Forum. Summary The net welfare effects of fishing, changes to fishing pressure and demand for wild-caught aquatic animals on wild aquatic animals seem highly morally ambiguous, in large part because there are 1. tradeoffs between species due to predation, e.g. larger (respectively smaller) populations and life expectancies for one species results in smaller (respectively larger) populations and life expectancies for their prey and competitors, and this cascades down the food chain, 2. uncertainty about moral weight tradeoffs between affected species, and 3. depending on the moral view, uncertainty about whether the directly and indirectly affected animals have good or bad lives on average. Acknowledgements Thanks to Brian Tomasik, Ren Ryba and Tori for their feedback on an earlier draft. All errors are my own. For prior related writing that is more comprehensive and suffering-focused, see Tomasik, 2015-2017. This piece focuses on population effects, overlapping largely with Tomasik, 2015. Population effects and welfare uncertainty Increasing the fishing of a species will tend to decrease their population, and decreasing their fishing will tend to increase their population, all else equal. For a given species, the marginal and average effects of fishing on the number of them alive at any time are typically at least several times greater than the effects on their annual catch under standard single-species fishery models.[1] With fishing deaths making up such a small share of a life on average, even if intense (although stunning during capture may be more widely used in the future), the effects on the size of the population and resulting welfare effects could be more ethically important than the effects on the number of fishing deaths. And then the effects on the populations could be morally ambiguous. First, it may be ambiguous whether the average individual has a good or bad life, so that reducing their population through fishing and increasing their population by reducing fishing could be morally ambiguous. Second, there are also population (and life expectancy) tradeoffs between species due to predation, with increasing the population of one species reducing the populations of its prey species.[2] Fishing reduces the populations of the directly fished species and (I'd normally guess) species up the food chain that depend on them,[3] while increasing the populations of the (unfished) prey (and competitors) of the directly fished species. But then the increased populations of the (unfished) prey (and competitors) may result in decreased populations for their prey. And so on. For example, Peruvian anchoveta, the species most wild-caught for feed and the most wild-caught fish species by numbers of individuals and tonnage (Mood & Brooke, 2024, Supplementary material 3, Supplementary material 6, Borthwick et al., 2021), primarily eat krill and copepods (Espinoza & Bertrand, 2008, Espinoza & Bertrand, 2014), so fishing Peruvian anchoveta presumably decreases Peruvian anchoveta populations but increases krill and copepod populations.[4] I illustrate with the effects of fishing on the fished pieces and their (unfished) prey: Fished animals have good lives Fished animals have bad lives Unfished prey have good lives Bad for fished animals Good for unfished prey Good for both Unfished prey have bad lives Bad for both Good for fished animals Bad for unfished prey Effects of fishing on fished animals and their unfished prey, in a simple model ignoring the rest of the food web. Fishing decreases the populations of the fished animal populations and increases the populations of their unfished prey, and these lives could be good or bad. Similarly, fi...

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: An impactful opportunity to take action for animals, by today (May 7th): Reply to UK government's open consultation on animal welfare labelling, published by David M on May 7, 2024 on The Effective Altruism Forum. I saw this post from Doing Westminster Better and wanted to blast it widely for people to take action today as it seems like a great opportunity, and a great ratio of effort to expected impact. Defra - the UK's department for the environment and, in some sense, the ministry that affects more lives than any other, as the body which regulates the farming of over a billion farmed animals in England - is currently considering the most impactful animal welfare regulation in years. This is your last day to tell the government that you care about animals and support this regulation. Here's what you need to know: the government is considering introduce animal welfare labeling. This would support consumers who still eat meat to make better informed choices about the animals they eat, whether that means helping you find the higher welfare product, or nudging you to recognise the suffering you're enabling by buying low welfare meat. Details continue in the linked post. The call to action says it takes 30 minutes. Plausibly more MVP, copy-paste actions are also possible, but I don't know whether copy-pasting answers into a government consultation helps. I'm going to try and find time to do this tonight. 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: Observations on Teaching for Four Weeks, published by ClareChiaraVincent on May 7, 2024 on LessWrong. I just finished a program where I taught two classes of high school seniors, two classes a day for four weeks, as part of my grad program. This experience was a lot of fun and it was rewarding, but it was really surprising, and even if only in small ways prompted me to update my beliefs about the experience of being a professor. Here are the three biggest surprises I encountered. 1: The Absent-Minded Professor Thing is Real I used to be confused and even a little bit offended when at my meetings with my advisor every week, he wouldn't be able to remember anything about my projects, our recent steps, or what we talked about last week. Now I get it. Even after just one week of classes, my short-term and long-term memory were both entirely shot. I would tell students things like, "send that to me in an email, otherwise I'll forget" because I would. Now that the program is over, things are slowly getting better, but I'm still recovering. I can't really tell why this happened, but there are two obvious theories. The first is just that two classes at the same time is too many names and faces (plus other personal details) at once and so much information just overwhelmed me. The other is that there's something unusual about teaching in particular. I noticed that I was doing a lot more task-switching than normal. Most jobs and most of my research experience involves working on projects for long blocks of time, multiple hours or sometimes multiple days with few distractions aside basics like eating and sleeping and commuting. But teaching involves changing your focus over and over. I've led recitation sections as a teaching assistant, but for some reason this was so much worse. That makes me think that it's more likely to be the task-switching. As a recitation leader, you have to remember a lot of names and faces too. But once you're outside of class you can mostly go back to work as normal, there's not so much task-switching. This project was in a high school but my students were all seniors, so I think this is what it would be like to teach college too. Most of them were already 18 so you can barely tell the difference. I was helping them with projects so I think it's a bit like being a PhD advisor too. So it could also be the load of keeping track of lots of research projects, more than just keeping track of lots of people. 2: Teaching Makes You Dehydrated For this program I taught only two days a week, just two classes, on Monday and Wednesday afternoon. But even with only two classes per day and two days per week, I became seriously and uncomfortably dehydrated. This had all kinds of weird knock-on effects with my digestion and my ability to concentrate. It was really very unpleasant. Part of this is that you have to be talking and meeting all the time. But mostly I got dehydrated because of the logistics. If you drink enough water, then halfway through the class you have to go to the bathroom and you're either super uncomfortable and distracted all session or you have to awkwardly walk out in the middle of class. Even if it doesn't hit right away, a 10-minute break between classes isn't enough time to go to the bathroom, especially since some students show up early from the next class and others stay late. So you're trapped. I had some success on days when I showed videos and could sneak out the back while they were watching. But overall this was bad for my teaching and my quality of life. 3: Teaching is a Grueling Job Even Under the Best Circumstances I didn't really like high school. Classes were too easy and too boring, and even though no one was asking very much of me, I felt like I was being taken advantage of. Implicitly I assumed that the teachers were the ones ta...

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 open AI models affect incentive to race?, published by jessicata on May 7, 2024 on LessWrong. I see it said sometimes that open models contribute to AI race dynamics. My guess is that they don't, and if anything, reduce AI race dynamics. I will consider a simplified model that only takes into account the cost of training a model, not the cost to deploy it (which tends to be small relative to revenue anyway). Let f(x) map a training expense x to a "value per day per customer" of the trained model, under the assumption that the training makes efficient use of the cost. That is, a customer values using an AI model trained with x compute at $f(x) per day. I assume there are n identical customers here; of course, there are complexities where some customers value AI more than others, incentivizing price discrimination, but I'm abstracting this consideration out. (In general, variation in how much customers value a product will tend to increase consumer surplus while reducing revenue, as it makes it harder to charge customers just under the maximum amount they're willing to pay.) I'm also assuming there is only one company that trains closed models for profit. This assumption is flawed because there is competition between different companies that train closed models. However, perfect competition assumptions would tend to reduce the incentive to train models. Suppose two companies have closed models of equivalent expense x. They each want to charge slightly less than the minimum of f(x) and the competitor's price, per customer per day. If each competitor undercuts the other slightly, the cost will approach 0. See the Traveler's Dilemma for a comparison. The reasons why this doesn't happen have to do with considerations like differences in models' performance on different tasks, e.g. some models are better for programming than others. If models are sufficiently specialized (allowing this sort of niche-monopolization), each specialized type of model can be modeled independently as a monopoly. So I'll analyze the case of a closed model monopoly, noting that translation to the real world is more complex. Suppose the best open model has compute x and a company trains a closed model with compute y > x. Each customer will now spend up to f(y) - f(x) per day for the model; I'll assume the company charges f(y) - f(x) and the customers purchase this, noting that they could charge just below this amount to create a positive incentive for customers. So the company's revenue over m days is nm(f(y) - f(x)). Clearly, this is decreasing in x. So the better the open model is, the less expected revenue there is from training a closed model. But this is simply comparing doing nothing to training a model of a fixed cost y. So consider instead comparing expected revenue between two different model costs, y and z, both greater than x. The revenue from y is nm(f(y) - f(x)), and from z it is nm(f(z) - f(x)). The difference between the z revenue and the y revenue is nm(f(z) - f(y)). This is unaffected by x. This can model a case where the company has already trained a model of cost y and is considering upgrading to z. In this case, the open model doesn't affect the expected additional revenue from the upgrade. Things get more complex when we assume there will be a future improvement to the open model. Suppose that, for k days, the open model has training cost x, and for the remaining m-k days, it has training cost x' > x. Now suppose that the closed AI company has already trained a model of cost y, where x < y < x'. They are considering upgrading to a model of cost z, where z > x'. Suppose they do not upgrade. Then they get nk(f(y) - f(x)) revenue from the first k days and nothing thereafter. Suppose they do upgrade, immediately. Then they get nk(f(z) - f(x)) revenue from the first k days, an...

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: AXRP Episode 31 - Singular Learning Theory with Daniel Murfet, published by DanielFilan on May 7, 2024 on The AI Alignment Forum. What's going on with deep learning? What sorts of models get learned, and what do the learning dynamics? Singular learning theory is a theory of Bayesian statistics broad enough in scope to encompass deep neural networks that may help answer these questions. In this episode, I speak with Daniel Murfet about this research program and what it tells us. Topics we discuss: What is singular learning theory? Phase transitions Estimating the local learning coefficient Singular learning theory and generalization Singular learning theory vs other deep learning theory How singular learning theory hit AI alignment Payoffs of singular learning theory for AI alignment Does singular learning theory advance AI capabilities? Open problems in singular learning theory for AI alignment What is the singular fluctuation? How geometry relates to information Following Daniel Murfet's work In this transcript, to improve readability, first names are omitted from speaker tags. Filan: Hello, everybody. In this episode, I'll be speaking with Daniel Murfet, a researcher at the University of Melbourne studying singular learning theory. For links to what we're discussing, you can check the description of this episode and you can read the transcripts at axrp.net. All right, well, welcome to AXRP. Murfet: Yeah, thanks a lot. What is singular learning theory? Filan: Cool. So I guess we're going to be talking about singular learning theory a lot during this podcast. So, what is singular learning theory? Murfet: Singular learning theory is a subject in mathematics. You could think of it as a mathematical theory of Bayesian statistics that's sufficiently general with sufficiently weak hypotheses to actually say non-trivial things about neural networks, which has been a problem for some approaches that you might call classical statistical learning theory. This is a subject that's been developed by a Japanese mathematician, Sumio Watanabe, and his students and collaborators over the last 20 years. And we have been looking at it for three or four years now and trying to see what it can say about deep learning in the first instance and, more recently, alignment. Filan: Sure. So what's the difference between singular learning theory and classical statistical learning theory that makes it more relevant to deep learning? Murfet: The "singular" in singular learning theory refers to a certain property of the class of models. In statistical learning theory, you typically have several mathematical objects involved. One would be a space of parameters, and then for each parameter you have a probability distribution, the model, over some other space, and you have a true distribution, which you're attempting to model with that pair of parameters and models. And in regular statistical learning theory, you have some important hypotheses. Those hypotheses are, firstly, that the map from parameters to models is injective, and secondly (quite similarly, but a little bit distinct technically) is that if you vary the parameter infinitesimally, the probability distribution it parameterizes also changes. This is technically the non-degeneracy of the Fisher information metric. But together these two conditions basically say that changing the parameter changes the distribution changes the model. And so those two conditions together are in many of the major theorems that you'll see when you learn statistics, things like the Cramér-Rao bound, many other things; asymptotic normality, which describes the fact that as you take more samples, your model tends to concentrate in a way that looks like a Gaussian distribution around the most likely parameter. So these are sort of basic ingredients in understandi...

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: Probably Good launched a new job board!, published by Probably Good on May 7, 2024 on The Effective Altruism Forum. We're excited to share a new addition to our site: an impact-focused job board! We've considered launching a job board for some time, so we're happy to add this feature to the Probably Good site. The job board aims to: Help people find more promising job opportunities, including in cause areas that aren't as thoroughly covered by other impact-focused boards such as 80,000 Hours and Animal Advocacy Careers. Direct our audience to concrete opportunities that meet a high standard of impact. Reduce friction for people on our site to take the first step towards a career change, by providing opportunities to apply for or just exposing them to new options. As Animal Advocacy Careers have highlighted before, job boards are often the primary gateway to career advice sites, and so we hope the job board will also extend our content's reach and general impact. Why we're launching a job board We want more people to find genuinely impactful opportunities. Our target audience focuses on people who might not be familiar with EA but are motivated to do good with their careers. Many of our users might search for jobs on more general non-profit job boards, which do not aim to prioritize opportunities based on impact. We hope our job board will increase people's exposure to opportunities that have a higher chance of making a significant impact. We want to make the process of finding high-impact jobs approachable, across a variety of cause areas. Other scale-sensitive job boards we support and link to focus on a narrower set of cause areas, either in scope or in tone. We hope to complement these job boards with additional opportunities from multiple cause areas and make many of those opportunities accessible to a broader, or just different, audience. Several leaders of high-impact organizations in Global Health & Development and other cause areas have expressed interest in this kind of resource for future recruitment as well. We want to make it easy for readers on our site to find high-impact jobs that meet their constraints. We understand that different people have very different career constraints, such as being unable to move geographically, having family responsibilities, or needing to meet specific salary requirements. If someone is already on our site, we want to meet them where they're at by sharing opportunities that are both impactful and aligned with their individual values, interests, requirements, and current work experience. How you can help Start using the job board! If you're excited about it and find it useful, please share it with people who might be interested in this kind of resource. This is an initial product, so we're working to improve it and will continue to make changes that increase the value it provides users. We'll add more features and expand the scope of the board as we go, but we'd love to hear feedback from users. If you have any ideas for increased functionality or would like to suggest jobs or organizations to feature on the board, please get in touch! Final Notes If you've read this far and want to consult on career opportunities that are good for you and good for the world, we think you might be a good candidate for our free 1-1 advising service. If you're interested, please apply on our site - or send the page to anyone you think might benefit from an advising call. Additionally, if you want to know other ways Probably Good can support you (or your local group) further, feel free to reach out. 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: GDP per capita in 2050, published by Hauke Hillebrandt on May 6, 2024 on The Effective Altruism Forum. Abstract Here, I present GDP (per capita) forecasts of major economies until 2050. Since GDP per capita is the best generalized predictor of many important variables, such as welfare, GDP forecasts can give us a more concrete picture of what the world might look like in just 27 years. The key claim here is: even if AI does not cause transformative growth, our business-as-usual near-future is still surprisingly different from today. Latest Draft as PDF Results In recent history, we've seen unprecedented economic growth and rises in living standards. Consider this graph:[1] How will living standards improve as GDP per capita (GDP/cap) rises? Here, I show data that projects GDP/cap until 2050. Forecasting GDP per capita is a crucial undertaking as it strongly correlates with welfare indicators like consumption, leisure, inequality, and mortality. These forecasts make the future more concrete and give us a better sense of what the world will look like soon. Abstract thoughts about utopia generate little emotional energy; I find these forecasts more plastic and informative, because GDP/cap is highly predictive of welfare.[2] GDP/cap's generalized predictive power helps us paint a more vivid picture of what the world will look like soon. The business-as-usual near future suggested by the data below could be seen as a soft lower bound on how much the world will change. And yet, this world still seems radically different from today. Since the figures below are adjusted for purchasing power parity (PPP), you can compare the GDP/cap of a poorer country in 2050 with the GDP/cap of a richer country in 2020. For instance, between now and 2050, China's GDP/cap will go from $19k to $43k, which is similar to France's today. And so, by 2050, 1.3B Chinese people might enjoy a lifestyle not dissimilar to that of a typical French person today. These GDP/cap are ~3x[3] higher than the median (i.e. typical) income, due to income inequality. Instead of downward adjusting them in our head we can also read them as an approximation of the typical combined consumption three person household.[4] And so, we can still look at these figures and imagine the median household income in Japan today and in China in 2050 to be ~$40k. 2020 2030 2040 2050 Pop. GDP GDP/ cap Pop. GDP/ cap GDP Pop. GDP/ cap GDP Pop. GDP/ cap GDP US 334M $20T $60K 356M $66K $23T 374M $76K $28T 389M $88K $34T Netherlands 17M $1T $54K 18M $61K $1T 18M $71K $1T 18M $85K $1T S. Arabia 34M $2T $55K 39M $70K $3T 43M $86K $4T 46M $102K $5T Germany 80M $4T $52K 79M $59K $5T 77M $69K $5T 75M $82K $6T Australia 26M $1T $52K 28M $58K $2T 31M $67K $2T 33M $77K $3T S. Korea 51M $2T $42K 53M $50K $3T 52M $59K $3T 51M $70K $4T Canada 38M $2T $48K 40M $53K $2T 42M $61K $3T 44M $70K $3T UK 67M $3T $45K 70M $52K $4T 73M $61K $4T 75M $71K $5T France 66M $3T $44K 68M $50K $3T 70M $56K $4T 71M $66K $5T Japan 125M $5T $40K 120M $47K $6T 114M $54K $6T 107M $63K $7T Poland 38M $1T $31K 37M $40K $2T 35M $53K $2T 33M $63K $2T Spain 46M $2T $39K 46M $47K $2T 46M $53K $2T 45M $61K $3T Malaysia 32M $1T $32K 36M $42K $2T 39M $54K $2T 41M $69K $3T Italy 60M $2T $39K 59M $43K $3T 58M $48K $3T 57M $55K $3T Turkey 82M $2T $26K 88M $34K $3T 93M $43K $4T 96M $54K $5T Russia 143M $4T $28K 139M $34K $5T 133M $45K $6T 129M $55K $7T China 1403M $27T $19K 1416M $27K $38T 1395M $34K $47T 1348M $43K $58T Thailand 69M $1T $19K 68M $25K $2T 66M $34K $2T 62M $45K $3T Argentina 46M $1T $22K 49M $27K $1T 53M $34K $2T 55M $43K $2T Mexico 135M $3T $19K 148M $25K $4T 158M $32K $5T 164M $42K $7T Colombia 50M $1T $16K 53M $21K $1T 55M $28K $2T 55M $38K $2T Iran 83M $2T $21K 89M $27K $2T 91M $35K $3T 92M $42K $4T Indonesia 272M $4T $14K 295M $18K $5T 312M $25K $8T 322M $33K $11...

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: Rapid capability gain around supergenius level seems probable even without intelligence needing to improve intelligence, published by Towards Keeperhood on May 6, 2024 on LessWrong. TLDR: 1. Around Einstein-level, relatively small changes in intelligence can lead to large changes in what one is capable to accomplish. 1. E.g. Einstein was a bit better than the other best physi at seeing deep connections and reasoning, but was able to accomplish much more in terms of impressive scientific output. 2. There are architectures where small changes can have significant effects on intelligence. 1. E.g. small changes in human-brain-hyperparameters: Einstein's brain didn't need to be trained on 3x the compute than normal physics professors for him to become much better at forming deep understanding, even without intelligence improving intelligence. Einstein and the heavytail of human intelligence 1905 is often described as the "annus mirabilis" of Albert Einstein. He founded quantum physics by postulating the existence of (light) quanta, explained Brownian motion, introduced the special relativity theory and derived E=mc from it. All of this. In one year. While having a full-time job in the Swiss patent office. With the exception of John von Neumann, we'd say those discoveries alone seem more than any other scientist of the 20th century achieved in their lifetime (though it's debatable). Though perhaps even more impressive is that Einstein was able to derive general relativity. Einstein was often so far ahead of his time that even years after he published his theories the majority of physicists rejected them because they couldn't understand them, sometimes even though there was experimental evidence favoring Einstein's theories. After solving the greatest open physics problems at the time in 1905, he continued working in the patent office until 1908, since the universities were too slow on the uptake to hire him earlier. Example for how far ahead of his time Einstein was: Deriving the theory of light quanta The following section is based on parts of the 8th chapter of "Surfaces and Essences" by Douglas Hofstadter. For an analysis of some of Einstein's discoveries, which show how far ahead of his time he was, I can recommend reading it. At the time, one of the biggest problems in physics was the "Blackbody spectrum", which describes the spectrum of electromagnetic wavelengths emitted by a Blackbody. The problem with it was that the emitted spectrum was not explainable by known physics. Einstein achieved a breakthrough by considering light not just as a wave, but also as light quanta. Although this idea sufficiently explained the Blackbody spectrum, physicists (at least almost) unanimously rejected it. The fight between the "light is corpuscles" and "light is a wave" faction had been decided a century ago, with a clear victory for the "wave" faction. Being aware of these possible doubts, Einstein proposed three experiments to prove his idea, one of which was the photoelectric effect. In the following years, Robert Millikan carried out various experiments on the photoelectric effect, which all confirmed Einstein's predictions. Still, Millikan insisted that the light-quanta theory had no theoretical basis and even falsely claimed that Einstein himself did not believe in his idea anymore. From Surfaces and Essences (p.611): To add insult to injury, although the 1921 Nobel Prize in Physics was awarded to Albert Einstein, it was not for his theory of light quanta but "for his discovery of the law of the photoelectric effect". Weirdly, in the citation there was no mention of the ideas behind that law, since no one on the Nobel Committee (or in all of physics) believed in them! [1][...] And thus Albert Einstein's revolutionary ideas on the nature of light, that most fundamental and all-...

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's the latest estimate of charitable giving that targets poverty?, published by Garrison on May 6, 2024 on The Effective Altruism Forum. Does anyone know the latest estimate of what percentage of US/Western charity goes to poverty broadly and international poverty specifically? A 2013 Dylan Matthews piece in WaPo cites a 2007 estimate. Googling isn't helping much. I may be writing an opinion piece for a major outlet and would love a more up to date answer (I'm guessing it wouldn't change too much). Thanks for listening. To help us out with The Nonlinear Library or to learn more, please visit nonlinear.org