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: Announcing The New York Declaration on Animal Consciousness, published by Sofia Fogel on April 22, 2024 on The Effective Altruism Forum. The last ten years have witnessed rapid advances in the science of animal cognition and behavior. Striking results have hinted at surprisingly rich inner lives in a wide range of animals, driving renewed debate about animal consciousness. To highlight these advances, the NYU Mind, Ethics and Policy Program and NYU Wild Animal Welfare Program co-hosted a conference on the emerging science of animal consciousness on Friday April 19 at New York University. This conference also served as the launch event for The New York Declaration on Animal Consciousness. This short statement, signed by leading scientists who research a wide range of taxa, holds that all vertebrates (including reptiles, amphibians, and fishes) and many invertebrates (including cephalopod mollusks, decapod crustaceans, and insects) have a realistic chance of being conscious, and that their welfare merits consideration. We now welcome signatures from others as well. If you have relevant expertise (for example, a graduate education or the equivalent in science, philosophy, or policy), you can send an email to nydeclaration@gmail.com from your institutional email address, say that you wish to sign, and list your title and institution as they should appear. Day-one media coverage of the conference and declaration included articles at Nature, NBC, Quanta, The Hill, and The Times. We also recorded the event, and our team will post videos on the declaration website in the near future. If you have questions or comments, feel free to send an email to nydeclaration@gmail.com or sofia.fogel@nyu.edu. Thank you to NYU Animal Studies, the NYU Center for Bioethics, and the NYU Center for Mind, Brain, and Consciousness for supporting this event. 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: Time complexity for deterministic string machines, published by alcatal on April 21, 2024 on The AI Alignment Forum. This was a project conducted during MATS 5.0 under the mentorship of Vanessa Kosoy and supported by a grant from BERI. It builds off the String Machines framework (and depends on the linked post for certain definitions), which models category-theoretic generalizations of finite-state transducers. The framework as it previously existed did not have representation-independent ways of bounding (analogues of) time complexity, or natural guarantees that output size would not grow exponentially in input size. We introduce "filtered" transducers, which operate on categories enriched over filtered sets (sets equipped with a function to a partially ordered monoid, where morphisms are functions respecting order), and then, restricting our attention to transducers with a finite state space, prove constraints on the time complexity growth and expressivity of string machines. Parameterizing complexity in string machines Filtered transducers Definition 1. The category FiltSet of filtered sets is the category such that an object is a tuple (S,degS), where S is a set and degS:SN is a function, a morphism f:(S,degS)(T,degT) is a function ST such that degT(f(s))degS(s) for all sS. We will generally refer to objects in FiltSet solely by the symbol corresponding to the underlying set going forward. One can observe that the identity function on a set S by definition satisfies degS(idS(s))=degS(s) for all sS and is thus a morphism in FiltSet. One can also observe that given f:ST and g:TV, degV(g(f(s)))degT(f(s))degS(s) for all sS, and therefore gf is also a morphism in FiltSet. Therefore, FiltSet is indeed a category. Definition 2. Given two objects S,TOb(FiltSet), we define their filtered product ST to be the set ST equipped with the function degST:STN satisfying degST(s,t)=degS(s)+degT(t) for all (s,t)ST. Given a morphism f:SU and a morphism g:TV, we define the morphism fg:STUV to be the function fg. Indeed, we have that degUV(f(s),g(t))=degU(f(s))+degV(g(t))degS(s)+degT(t)=degST(s,t), so fg is a morphism in FiltSet. Due to the associativity and commutativity of addition, as well as the natural associativity and commutativity (up to isomorphisms which are still isomorphisms in FiltSet) of the cartesian product, is naturally associative and commutative up to isomorphism. Additionally, the one-element set 1 equipped with deg1()=0 and unitor maps which are the same as in Set (which are, by their definition, filtered morphisms) provides a left and right unit for , making FiltSet a symmetric monoidal category. Remark. Suppose filtered sets S,T,U and filtered morphisms f:ST and g:SU. Then, the unique factoring function STU defined by s(f(s),g(s)) is only a filtered morphism STU if degT(f(s))+degU(g(s))degS(s), which does not hold in general. Therefore, does not provide a product except for when at least one of the sets has degree uniformly zero. However, FiltSet does have finite products ST where degST(s,t):=max(degS(s),degT(t)). We will not be using this construction. Remark. The set-theoretic disjoint union, with its degree function being the canonical factoring map to N of its components' degree functions, provides all finite coproducts in FiltSet. Definition 3. A filtered-morphism category C is a locally small symmetric monoidal category enriched over FiltSet, using FiltSet's filtered product as its monoidal structure. This expresses the notion of morphisms having degrees which are subadditive under composition in a way that naturally extends to a complexity constraint on transducers. As the monoidal identity of FiltSet is the single-element set with degree zero, the arrows IFiltSetHomC(A,A) providing the identity morphism idA in the enrichment construction will ensure that...

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: Transfer Learning in Humans, published by niplav on April 22, 2024 on LessWrong. I examine the literature on transfer learning in humans. Far transfer is difficult to achieve, best candidate interventions are to practice at the edge of one's ability and make many mistakes, evaluate mistakes after one has made them, learn from training programs modeled after expert tacit knowledge, and talk about on one's strategies when practicing the domain. When learning, one would like to progress faster, and learn things faster. So it makes sense to search for interventions that speed up learning (effective learning techniques), enable using knowledge and knowledge patterns from one learned domain in a new domain if appropriate (transfer learning), and make it easier to find further learning-accelerating techniques (meta-learning). Summary I've spent ~20 hours reading and skimming papers and parts of books from different fields, and extracting the results from them, resulting spreadsheet here, google doc with notes here. I've looked at 50 papers, skimmed 20 and read 10 papers and 20% of a book. In this text I've included all sufficiently-different interventions I've found that have been tested empirically. For interventions tried by scientists I'd classify them into (ordered by how relevant and effective I think they are): Error-based learning in which trainees deliberately seek out situations in which they make mistakes. This has medium to large effect sizes at far transfer. Long Training Programs: These usually take the form of one- or two-semester long classes on decision-making, basic statistics and spatial thinking, and produce far transfer at small to medium effect sizes. Such programs take a semester or two and are usually tested on high-school students or university students. Effective Learning Techniques: Things like doing tests and exercises while learning, or letting learners generate causal mechanisms, which produce zero to or best small amounts of far transfer but speed up learning. OODA-loop-likes: Methods that structure the problem-solving process, such as the Pólya method or DMAIC. In most cases, these haven't been tested well or at all, but they are popular in the business context. Also they look all the same to me, but probably have the advantage of functioning as checklists when performing a task. Transfer Within Domains: Methods that are supposed to help with getting knowledge about a particular domain from an expert to a trainee, or from training to application on the job. Those methods have a high fixed cost since experts have to be interviewed and whole curricula have to be created, but they work very well at the task they've been created for (where training sometimes is sped up by more than an order of magnitude). Additionally, most of the research is on subjects which are probably not intrinsically motivated to apply a technique well (i.e. high school students, military trainees, and university students), so there is a bunch of selection pressure on techniques which still work with demotivated subjects. I expect that many techniques work much better with already motivated subjects, especially ones that are easy to goodhart. In general, the tension I was observing is that industry and the military are the ones who perform well/do non-fake things, but academia are the ones who actually measure and report those measures to the public. From when I've talked with people from industry, they don't seem at all interested in tracking per-employee performance (e.g. Google isn't running RCTs on their engineers to increase their coding performance, and estimates for how long projects will take are not tracked & scored). I also haven't seen many studies quantifying the individual performance of employees, especially high-earning white collar knowledge-workers. Recomme...

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 in a GWWC Pin?, published by JWS on April 21, 2024 on The Effective Altruism Forum. Earlier this year I got an email telling me that I could receive (if I wished) a GWWC pledge pin after completing the first year of being a pledger. So I sent off a form, and after waiting a little while longer, a small package arrived at my door with the pin inside. I wanted to write this post about that year much closer to the time I received it, but as always real life got in the way! But that extra time got me thinking more about what goes into the pin, and I think the resulting post is better. Physically The pin is actually a little bit larger than other charitable pins.[1] It's metal, mostly black with some gold colour that accents the edges of the pin and highlights the GWWC logo in the middle. It's a nice, minimal, slick design. And… that's kind of it from this perspective. Literally, that's what's in a GWWC pin. But reductionist physicalism is like, so false, and it can't give us good explanations of the world. The atoms that make up the pin, and the laws that governed their movement there matter much less than why the pin came to be in my possession, and what that means. Financially To qualify for a pledge pin, you have to:[2] Have been a GWWC Pledger for more than 1 year Be at 100% or more of the amount that you pledged to give (dated from when your Pledge began) In practice, you qualify if you donate 10% of your income for a year. In my case, I donated 10% of my pre-tax income, but GWWC also counts gift-aid so technically I donated 125% of my pledged amount,[3] but I don't really think of that when I make my donations. In practice, I found it was fairly easy for me to live as I was used to without making major changes while taking the pledge, but I am likely to find it easier than others given that I have no dependents, and am generally have a fairly prudent disposition towards personal spending. The major thing was probably lower savings, but even here I think I've done ok, though counterfactually less well off than I would have been otherwise from a purely monetary perspective. This is definitely something to bear in mind, and the commitments of the pledge aren't necessarily for everyone at all points of life. Furthermore, the 10% donation point is probably about where I've made all of the 'easy' trade-offs[4] regarding personal spending and lifestyle, and in practice I had sent up a recurring donation so it acted more like 'income forgone' than 'having to give away my hard-earned money', at least from pyschologically. If I were to start giving even more, I expect having to make more significant concessions and compromises, and I'm not sure I feel in a position to do that yet. But again, this still a purely behavioural explanation. I gave 10% of my income, filled out a form, and got sent a pin. We can still do better about finding out what's in the pin. Charitably An even better explanation might be where I decided to donate my 10%. So, thanks to the nifty new dashboard feature on the GWWC website, here's the % breakdown of my donations since the beginnings of my pledge: As you can see, there's definitely a spread of donations between causes and charities. I am a believer in moral pluralism and worldview diversification, and eschew EV-maxxing perspectives on my donations.[5] I feel unsure about what it means to do good or how to act under this uncertainty. I very much nodded along to the perspective Dustin shared in this article, but of course from many orders of magnitude less wealth with which to donate! This doesn't reflect any modal donation though. Instead, I mostly donate through a recurring donation that I split between the major cause areas, with Global Health & Development the top, followed by Animal Welfare, with Longtermism being the smallest amount. (F...

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: A couple productivity tips for overthinkers, published by Steven Byrnes on April 20, 2024 on LessWrong. 1. If you find that you're reluctant to permanently give up on to-do list items, "deprioritize" them instead I hate the idea of deciding that something on my to-do list isn't that important, and then deleting it off my to-do list without actually doing it. Because once it's off my to-do list, then quite possibly I'll never think about it again. And what if it's actually worth doing? Or what if my priorities will change such that it will be worth doing at some point in the future? Gahh! On the other hand, if I never delete anything off my to-do list, it will grow to infinity. The solution I've settled on is a priority-categorized to-do list, using a kanban-style online tool (e.g. Trello). The left couple columns ("lists") are very active - i.e., to-do list items that I might plausibly do today or tomorrow, with different columns for different contexts (e.g. "Deep work" items for when I have a block of time to concentrate, "Shallow work" items for when I don't, and before a trip I might temporarily add an "On the airplane" column, etc.). Then going off to the right, I have a series of lower- and lower-priority columns - "Within 1 week", "Within 2 weeks", "Within 1 month", "Within 2 months", "Within 6 months", "Someday / maybe", "Probably never". I don't take the column titles too literally; the important part is that if something doesn't seem that urgent or worthwhile, I find it very easy and satisfying to drag that task one or two columns to the right. I'm not giving up on it forever! But the further right we go, the less frequently I'll look at that column. So I get the benefit of a very manageable to-do list without needing to make the irreversible commitment of deleting items that I haven't done. (Following David Allen, I also have a "Waiting for…" column for items that someone else is supposed to do. I also have a "Done" column, which is arguably pointless as I just delete everything off the "Done" column every couple weeks, but the deleting ritual is nice because I get another chance to make sure I've really finished it, and is also an excuse to feel happy about my recent accomplishments.) 2. If you find that you're reluctant to delete (or heavily edit) a piece of text / slide that you worked hard on, copy it into a "graveyard" first I hate the idea of deleting something I wrote, because what if I change my mind and decide it's better as it is? I'd have to rewrite it, and maybe it wouldn't come out as good the second time! Gahh! (Granted, lots of text editors have affordances for going through a document's history to retrieve deleted text. But I find them a hassle to use.) Instead, whenever I'm deleting or rewriting more than a couple words, I simply copy-and-paste the current version into a disorganized "graveyard" of text snippets, paragraphs, sections, etc. at the end of the document (or in a separate sister document). Realistically, I almost never pull anything out of the "graveyard". But sometimes I do pull things out - not only in the course of whatever I'm writing, but also sometimes months after I finish. And more importantly, knowing that the graveyard is there and easily accessible makes me feel more comfortable "killing my darlings" in the first place. Ditto for editing slides and so on. 3. If you find that you're reluctant to throw out papers, make it fast and easy to file them Sometimes I get something in the mail that I probably will never need to look at, but I don't want to throw it out, because what if I'm wrong and I'll need it after all? Gahh! This is what a filing cabinet is for. In Getting Things Done, David Allen writes "If it takes longer than sixty seconds to file something, you won't file, you'll stack." (See here for his practical tips...

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 up with all the non-Mormons? Weirdly specific universalities across LLMs, published by mwatkins on April 20, 2024 on LessWrong. tl;dr: Recently reported GPT-J experiments [1 2 3 4] prompting for definitions of points in the so-called "semantic void" (token-free regions of embedding space) were extended to fifteen other open source base models from four families, producing many of the same bafflingly specific outputs. This points to an entirely unexpected kind of LLM universality (for which no explanation is offered, although a few highly speculative ideas are riffed upon). Work supported by the Long Term Future Fund. Thanks to quila for suggesting the use of "empty string definition" prompts, and to janus for technical assistance. Introduction "Mapping the semantic void: Strange goings-on in GPT embedding spaces" presented a selection of recurrent themes (e.g., non-Mormons, the British Royal family, small round things, holes) in outputs produced by prompting GPT-J to define points in embedding space randomly sampled at various distances from the token embedding centroid. This was tentatively framed as part of what appeared to be a "stratified ontology" (based on hyperspherical regions centred at the centroid). Various suggestions attempting to account for this showed up in the comments to that post, but nothing that amounted to an explanation. The most noteworthy consideration that came up (more than once) was layer normalisation: the embeddings that were being customised and inserted into the prompt template were typically out-of-distribution in terms of their distance-from-centroid: almost all GPT-J tokens are at a distance-from-centroid close to 1, whereas I was sampling at distances from 0 to 10000. This, as far as I could understand the argument, might be playing havoc with layer norm, thereby resulting in anomalous (but otherwise insignificant) outputs. That original post also presented circumstantial evidence, involving prompting for definitions of glitch tokens, that this phenomenon extends to GPT-3 (unfortunately that's not something that could have been tested directly). Some time later, a colleague with GPT-4 base access discovered that simply prompting that model for a definition of the empty string, i.e. using the prompt at temperature 0 produces "A person who is not a member of the clergy", one of the most frequent outputs I'd seen from GPT-J for random embeddings at various distances-from-centroid, from 2 to 12000. With the same prompt, but at higher temperatures, GPT-4 base produced other very familiar (to me) styles of definition such as: a small, usually round piece of metal; a small, usually circular object of glass, wood, stone, or the like with a hole through; a person who is not a member of a particular group or organization; a person who is not a member of one's own religion; a state of being in a state of being. Looking at a lot of these outputs, it seems that, as with GPT-J, religion, non-membership of groups, small round things and holes are major preoccupations. As well as indicating that this phenomenon is not a quirk particular to GPT-J, but rather something more widespread, the empty string results rule out any central significance of layer norm. No customised embeddings are involved here - we're just prompting with a list of eight conventional tokens. I would have predicted that the model would give a definition for emptiness, non-existence, silence or absence, but I have yet to see it do that. Instead, it behaves like someone guessing the definition of a word they can't see or hear. And in doing so repeatedly, statistically) it's perhaps tells us something completely unexpected about how its "understanding of the world" (for want of a better phrase) is organised. Models tested The same experiments were run on sixteen base models...

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: Thoughts on seed oil, published by dynomight on April 20, 2024 on LessWrong. A friend has spent the last three years hounding me about seed oils. Every time I thought I was safe, he'd wait a couple months and renew his attack: "When are you going to write about seed oils?" "Did you know that seed oils are why there's so much {obesity, heart disease, diabetes, inflammation, cancer, dementia}?" "Why did you write about {meth, the death penalty, consciousness, nukes, ethylene, abortion, AI, aliens, colonoscopies, Tunnel Man, Bourdieu, Assange} when you could have written about seed oils?" "Isn't it time to quit your silly navel-gazing and use your weird obsessive personality to make a dent in the world - by writing about seed oils?" He'd often send screenshots of people reminding each other that Corn Oil is Murder and that it's critical that we overturn our lives to eliminate soybean/canola/sunflower/peanut oil and replace them with butter/lard/coconut/avocado/palm oil. This confused me, because on my internet, no one cares. Few have heard of these theories and those that have mostly think they're kooky. When I looked for evidence that seed oils were bad, I'd find people with long lists of papers. Those papers each seemed vaguely concerning, but I couldn't find any "reputable" sources that said seed oils were bad. This made it hard for me to take the idea seriously. But my friend kept asking. He even brought up the idea of paying me, before recoiling in horror at my suggested rate. But now I appear to be writing about seed oils for free. So I guess that works? On seed oil theory There is no one seed oil theory. I can't emphasize this enough: There is no clear "best" argument for why seed oils are supposed to be bad. This stuff is coming from internet randos () who differ both in what they think is true, and why they think it. But we can examine some common arguments. We ate seed oil and we got fat. One argument is that for most of human history, nobody dieted and everyone was lean. But some time after the industrial revolution, people in Western countries started gaining weight and things have accelerated ever since. Here's BMI at age 50 for white, high-school educated American men born in various years: For the last few decades, obesity (BMI 30) has grown at around 0.6% per year. Clearly we are doing something wrong. We evolved to effortlessly stay at a healthy weight, but we've somehow broken our regulatory mechanisms. Anywhere people adopt a Western diet, the same thing happens. Of course, the Western diet is many things. But if you start reading ingredients lists, you'll soon notice that everything has vegetable oil in it. Anything fried, obviously, but also instant noodles, chips, crackers, tortillas, cereal, energy bars, canned tuna, processed meats, plant-based meat, coffee creamer, broths, frozen dinners, salad dressing, and sauces. Also: Baby food, infant formula, and sometimes even ice cream or bread. People eat a lot more vegetable oil than they used to (figure from Lee et al. (2022)): Many vegetable oils (and particularly seed oils) are high in linoleic acid. And guess what's making up a rapidly increasing fraction of body fat? (figure from Stephan Guyunet): Even many types of meat now have high linoleic acid levels, because the animals are now eating so much vegetable oil. It's plausible this is doing something to us. And seed oils are highly processed. Another common argument is that even if we can't identify exactly where the Western diet went wrong, we know that we spent almost our whole evolutionary history eating like hunter-gathers (and most of the rest eating like subsistence farmers). And hunter-gatherers are all thin. So maybe we should eat like they did? That sounds kind of fanciful, but consider the most conventional dietary advice, the thing tha...

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: Development through Economic Growth - a report commissioned by Open Philanthropy, published by Open Philanthropy on April 19, 2024 on The Effective Altruism Forum. Open Philanthropy commissioned a report from Stefan Dercon on economic growth as the main driver of poverty reduction. In the report, Dercon highlights a set of overlooked policies that can help boost economic growth in developing countries, as well as key reasons why these policies aren't always pursued. You can read the full report here. Thanks for listening. To help us out with The Nonlinear Library or to learn more, please visit nonlinear.org

Welcome to The Nonlinear Library, where we use Text-to-Speech software to convert the best writing from the Rationalist and EA communities into audio. This is: Healthier Hens Y2.5 update and new avenues, published by lukasj10 on April 19, 2024 on The Effective Altruism Forum. TL;DR Healthier Hens (HH) aims to improve cage-free hen welfare, focusing on key issues such as keel bone fractures (KBFs). In the last 6 months, we've conducted a vet training in Kenya, found a 42% KBF prevalence, and are exploring alternative promising interventions in collaboration with the Welfare Footprint Project, publishing transferrable findings along the way. Our staff satisfaction remains high, but concerns about operational capacity are on the rise. We're deciding on future strategies, considering funding and organizational changes. Our budget for Y3 ('23 Sep - '24 Sep) is $65k-$135k, with a $0-70k funding gap. In this post, we share key updates, lessons learned and our plans for immediate next steps. We hope others can benefit from what we're observing and our attempts to identify promising pathways towards improved hen health and welfare. We welcome feedback from the community. Key points Healthier Hens (HH) continues looking for ways to address major sources of cage-free egg-laying hen pain such as keel bone fractures (KBFs) through data collection and capacity building. Consider reading our introductory, 6M, 1Y, 1.5Y and 2Y update posts to learn more about our background, mission and updated approach. Due to funding constraints, we had downscaled last year and continue running lean, focusing on identifying promising avenues via mini-projects before piloting anything on the ground. Our Y3 (Sep '23 - Sep '24) budget will mainly comprise of outreach, staff, travel, and potential research activity costs. We raised $65k for this year's work. In our ongoing scoping efforts in Kenya we conducted a KBF baseline case study, a vet training workshop on hen welfare, and are exploring a certification scheme and alternative promising interventions. We continue publishing details on how we go about seeking impact for the hens, including outcomes and lessons learned for others to incorporate in their activities. Our mid-year staff survey revealed overall satisfaction with our culture and values but also some concerns about time commitments and task distribution. Although there is room for improvement, we all still agree that hen welfare is our priority. Our next strategic decision-making point is in May 2024, where we'll choose a scenario from a pool ranging from shutting down to trialling a couple of interventions, depending on the final impact and cost-efficacy estimates. Our funding gap is small and scenario-dependent. Our runway is flexible (3M max.). We will need $0-75k for the second half of 2024. HH team preparing for the vet training, Nairobi, November, 2023. End of vet training, Nairobi, November, 2023. Specific HH Y2.5 updates In-person veterinarian training ( full report here): We partnered with the University of Nairobi and held a two-day training for vet professionals, who visit farms routinely. The main purpose of the training was to provide knowledge on proper hen welfare and grow on-farm assessment capacity among active professionals. We had received over 70 applications and invited 22 veterinarians to attend, 18 of whom attended. Our training was approved by the Kenya Veterinary Board as a Continuing Professional Development (CPD) activity earning the participants 16 CPD points each - a factor that significantly incentivised the participants. Over two days, 9 experts from various fields provided class lectures and practical sessions in the veterinary surgery room. We concluded the training with a site visit to a cage-free farm to perform on-farm welfare indicator assessments. We conducted pre-, post- and 2-month-post surveys to evaluate the outcomes of the training. Additionally, a key informant survey was conducted six weeks after the ...

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 is looking for a CEO / Executive Director!, published by Probably Good on April 19, 2024 on The Effective Altruism Forum. Probably Good is searching for a CEO / executive director to take the reins and lead the organization in its mission to help more people use their careers to make a positive impact at scale. If you're not familiar, Probably Good's main activities are researching high-impact career paths, publishing online content, and providing one-on-one advising. Our reach has been growing rapidly, with more than 100,000 page views in 2023 and 449% year-on-year monthly organic (unpaid) traffic growth. Through ads, we can cause a new user to counterfactually read an entire article on the site (such as a career guide chapter or career profile) for a marginal cost of roughly $1. We believe this presents an exciting opportunity for impact, but our work so far has been bottlenecked by management capacity. As a result, we are excited to search for a CEO / executive director who can empower Probably Good to reach its full impact potential. We're looking for someone who can lead the organization to grow, improve, and help more people increase the positive impact of their careers. We believe that there's much more to be done and are looking for someone with a "founder's mindset" to think creatively about the next steps for Probably Good. This is a remote position reporting to the board of directors. If you think you might be a good fit or know someone who is, you can find more details and the application form here. Thanks for listening. To help us out with The Nonlinear Library or to learn more, please visit nonlinear.org