The Nonlinear Library

The Nonlinear Library

By The Nonlinear FundEducation
Download on the App Store

All The Nonlinear Library episodes

  • EA - How long till Brussels?: A light investigation into the Brussels Gap by Yadav
    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 long till Brussels?: A light investigation into the Brussels Gap, published by Yadav on December 26, 2022 on The Effective Altruism Forum.
    This is written quickly; I'm mostly looking to gauge whether this topic is important at all. As such, I'd love to hear your thoughts!
    Introduction
    This post quickly looks at the 'Brussels Gap', i.e. the time elapsed between enacting a Regulation within the EU and its emulation in other parts of the world (i.e. the Brussels Effect).
    The 'Brussels Gap' could be a relatively important phenomenon that is worth examining, given the development of a new Regulation on Artificial Intelligence proposed by the EU. If the Regulation comes to fruition (and it seems likely that it will), then having a better picture of the timescales at which the Brussels Effect operates seems crucial.
    In this post, I outline a primer on the Brussels Effect and the relationship it could play with AI development. Then I list out different case studies of the Brussels Effect and look at the time differences between the enactment of a Regulation within the EU and its replication in other countries outside Europe.
    Depending on your timelines for the development of Transformative AI, understanding the 'Brussels Gap' could be directly useful for weighing up the importance of the Brussels Effect.
    If you're already up to speed with Brussels Effect, I'd suggest skipping to the section titled "The 'Brussels Gap'".
    Background Information:
    What is the Brussels Effect?
    The "Brussels Effect" refers to the EU's unilateral ability to regulate the global marketplace. This ability allows legislation and policy adopted by the EU to diffuse to other jurisdictions.
    The General Data Protection Regulation (GDPR) is a salient example of this. The Regulation was passed by the European Parliament and the Council of the EU in 2016 and came into effect on the 25th of May, 2018.
    Introducing the GDPR led to a ripple effect; India, Japan, Brazil and elsewhere began emulating the Regulation.
    Types of Brussels Effect
    De Facto BE: This is where conduct under a particular area is done under the unilateral EU rules, even when other states continue to maintain their own rules. Anu Bradford describes this as the 'convergence' of a company's global production and conducts towards EU regulations.
    De Jure BE: Where states adopt EU regulations into their own legal body.
    Why does the Brussels Effect happen?
    Why do other countries choose to model after the EU? The two most substantial answers to this are (see Anu Bradford's work for more):
    1. Institutional Competence
    The EU's architecture is incredibly complex. Here is an outline of how each body interacts with one another.
    This complicated architecture has also had a 5.2% growth rate in all its bodies combined, with most of the staff being highly educated (usually possessing a master's degree).
    Both of these factors resulted in a signal of competence to other countries in the world, which results in a higher degree of trust in the EU's decisions and policies.
    2. Economic trade-offs
    The EU consists of 27 Member States, creating strong economic incentives for companies to operate within the EU so they can trade within these states. Those who want to participate in this internal market must comply with the Union's Regulations. Essentially, no one gets to keep their cake and eat it too. If you're going to get hold of the gains of trading within the EU, you'll have to obey the rules of the EU.
    In creating an EU-compliant product, companies have a choice: do they create separate products for different jurisdictions, or do they standardise their products across the world? The latter is usually more economically feasible, which unlocks the ability for the rules of the EU to spread.
    What does the Brussels Effect have to do with AI?
    The EU is likely to intro...
    9 min
  • LW - Are men harder to help? by braces
    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: Are men harder to help?, published by braces on December 26, 2022 on LessWrong.
    Of Boys and Men is a new book about how the outcomes of American men and boys are lagging in several ways. Most of it concerns aggregate trends that are irrefutable, such as the gender gap in college graduation. But books like these are often useful for testable hypotheses rather than reliable facts. Case in point: Reeves makes the odd claim that "men are hard to help":
    [A] startling number of social programs seem to work well for girls and women, but not for boys and men — among which are a student-mentoring scheme in Fort Worth, Texas; a school-choice program in Charlotte, North Carolina; an income boost to low-wage earners in New York City; and many more.
    The failure of these programs to help boys and men is a big problem, given that in many cases they are the ones who need the most help. But the problem rarely receives any attention, not least because almost nobody knows about it. (source)
    I was startled when I read this because it seemed like such a textbook example of cherry-picking. Of the many studies out there we should probably have priors that the effects are about even on average. Reeves cites around eight individual evaluations that found larger benefits for females compared to males, but he never points to any attempt at aggregation.
    The cited studies cover four broad areas:
    Free college tuition with other supports (Kalamazoo Promise and Stay the Course at Tarrant County College in Forth Worth, Texas)
    Pre-school programs (a study of the Abecedarian, Perry, and the early Training Project and one of Project READS in North Carolina)
    Mentoring programs and boarding schools (just mentions that they are in New Hampshire, Baltimore, and Washington, D.C.)
    Wage subsidies (the Paycheck Plus pilot)
    These all find the effect difference that Reeves claims to have discovered. The problem is that “men are harder to help” doesn’t seem to be true in general. For each of the four areas, I looked for meta-analyses or similar studies that estimated gender differences in the treatment effects:
    Free college tuition: a rigorous trial of the Susan Thompson Buffett Foundation, which gives large grants to Nebraska high schoolers planning to attend public colleges in the state, finds if anything larger impacts on men compared to women. A meta-analysis discussing this question finds a mix favoring men, women, or neither and concludes that there's not enough evidence to say.
    Pre-school programs: there was already a meta-analysis of this looking at all trials of early childhood interventions. There’s no clear advantage for girls. Papers citing the meta-analysis also show a tilt towards bigger effects for boys.
    Mentoring programs: a meta-analysis of 70 studies finds larger effects for those with a higher proportion of men. A smaller meta-analysis focused on disruptive behavior problems finds the same.
    Wage subsidies: Here he is perhaps right. Evaluations targeted at women tended to have larger effects in the most recent meta-analysis I could find, but by a small margin.
    This isn't meant to be an exhaustive evaluation of the question, but men do not seem to be harder to help and the evidence presented by Reeves is underwhelming. It’s a little disappointing that this easily checkable claim made it into the book.
    Thanks for listening. To help us out with The Nonlinear Library or to learn more, please visit nonlinear.org.
    4 min
  • EA - What are the best charities to donate to in 2022? by Luke Freeman
    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 are the best charities to donate to in 2022?, published by Luke Freeman on December 26, 2022 on The Effective Altruism Forum.
    This is a linkpost for GWWC's latest giving recommendations... and also a reminder to donate or pledge before the end of 2022!
    When you care about making a difference in the world, it's natural to ask: "Where can my donations do the most good?"
    The great news is that your donations can have an astonishing impact.
    But to maximise your impact you need to donate effectively. This is especially important given the best charities are often 10 times better than a typical charity within the same area and hundreds of times better than poorly performing charities. The worst charities can even do harm. That means if you donated $100 to the best charity, you could be doing even more good than someone who donates $1,000 to a typical charity, or $10,000 to a poorly performing one.
    How to donate effectively
    The point of charity is to help others. Donating effectively means that you’re taking steps to make sure you’re helping others the most for every dollar you give.We think the best way to do this is to:
    Identify a promising cause to support. The most promising causes are generally:
    Large in scale: they significantly impact many lives
    Neglected: they still need more funding and support
    Tractable: there are clear and practical ways of making progress
    Choose an excellent fund or charity working on the cause. Indicators of worthwhile organisations include: reliance on evidence, cost-effectiveness, transparency, room for more funding, and a strong track record. We generally recommend giving to expert-managed funds over charities, because they can allocate your donation where and when it is needed most.
    Pick an efficient way to donate. Try getting your donation matched by your employer or set up a recurring donation.
    In practice, choosing a cause and evaluating funds and charities can take a lot of time and effort, and many donors aren't able to work it into their busy schedules. To help you get started, we've put together a list of trustworthy, cost-effective funds and charities working on some of the most pressing causes.
    Our effective giving recommendations
    These recommendations are listed roughly in order of convenience and suitability for most donors. The right giving opportunity for you will depend on your particular values and worldview.
    Donate to expert-managed funds
    For most people, we recommend donating through an expert-led fund that is focused on effectiveness.
    Funds are both convenient and highly effective. They allow donors to pool their money so that they can support outstanding giving opportunities that are evaluated by expert grantmakers and trusted charity evaluators.
    This approach is similar to using an investment fund instead of trying to pick which individual stocks will be the best investments. The fund distributes your donation among multiple grantees with the goal of maximising your overall impact. Read more about the advantages of funds.
    Donors from the US, UK, and the Netherlands can make tax-deductible donations to the following funds using the Giving What We Can donation platform. For other countries, you can read our tax deductibility guide by country — but we still think these funds are likely to be your best option, even if they’re not tax deductible.
    Here are our top-rated funds grouped by cause area.
    Learn more about how we choose which charities and funds to recommend.
    Top-rated funds improving human wellbeing
    These funds generally give to organisations taking evidence-based approaches to improve and save lives of people currently alive. Donors who value their donations going to organisations either with a strong track-record or a promising and testable new approach to help people in the current generation can maximise the...
    11 min
  • EA - Slightly against aligning with neo-luddites by Matthew Barnett
    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: Slightly against aligning with neo-luddites, published by Matthew Barnett on December 26, 2022 on The Effective Altruism Forum.
    To summarize,
    When considering whether to delay AI, the choice before us is not merely whether to accelerate or decelerate the technology. We can choose what type of regulations are adopted, and some options are much better than others.
    Neo-luddites do not fundamentally share our concern about AI x-risk. Thus, their regulations will probably not, except by coincidence, be the type of regulations we should try to install.
    Adopting the wrong AI regulations could lock us into a suboptimal regime that may be difficult or impossible to leave. So we should likely be careful not endorse a proposal because it's "better than nothing" unless it's also literally the only chance we get to delay AI.
    In particular, arbitrary data restrictions risk preventing researchers from having access to good data that might help with alignment, potentially outweighing the (arguably) positive effect of slowing down AI progress in general.
    It appears we are in the midst of a new wave of neo-luddite sentiment.
    Earlier this month, digital artists staged a mass protest against AI art on ArtStation. A few people are reportedly already getting together to hire a lobbyist to advocate more restrictive IP laws around AI generated content. And anecdotally, I've seen numerous large threads on Twitter in which people criticize the users and creators of AI art.
    Personally, this sentiment disappoints me. While I sympathize with the artists who will lose their income, I'm not persuaded by the general argument. The value we could get from nearly free, personalized entertainment would be truly massive. In my opinion, it would be a shame if humanity never allowed that value to be unlocked, or restricted its proliferation severely.
    I expect most readers to agree with me on this point — that it is not worth sacrificing a technologically richer world just to protect workers from losing their income. Yet there is a related view that I have recently heard some of my friends endorse: that nonetheless, it is worth aligning with neo-luddites, incidentally, in order to slow down AI capabilities.
    On the most basic level, I think this argument makes some sense. If aligning with neo-luddites simply means saying "I agree with delaying AI, but not for that reason" then I would not be very concerned. As it happens, I agree with most of the arguments in Katja Grace's recent post about delaying AI in order to ensure existential AI safety.
    Yet I worry that some people intend their alliance with neo-luddites to extend much further than this shallow rejoinder. I am concerned that people might work with neo-luddites to advance their specific policies, and particular means of achieving them, in the hopes that it's "better than nothing" and might give us more time to solve alignment.
    In addition to possibly being mildly dishonest, I'm quite worried such an alliance will be counterproductive on separate, purely consequentialist grounds.
    If we think of AI progress as a single variable that we can either accelerate or decelerate, with other variables held constant upon intervention, then I agree it could be true that we should do whatever we can to impede the march of progress in the field, no matter what that might look like. Delaying AI gives us more time to reflect, debate, and experiment, which prima facie, I agree, is a good thing.
    A better model, however, is that there are many factor inputs to AI development. To name the main ones: compute, data, and algorithmic progress. To the extent we block only one avenue of progress, the others will continue. Whether that's good depends critically on the details: what's being blocked, what isn't, and how.
    One consideration, which has been pointed out by many before...
    8 min
  • AF - Can we efficiently distinguish different mechanisms? by Paul Christiano
    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: Can we efficiently distinguish different mechanisms?, published by Paul Christiano on December 27, 2022 on The AI Alignment Forum.
    (This post is an elaboration on “tractability of discrimination” as introduced in section III of Can we efficiently explain model behaviors? For an overview of the general plan this fits into, see Mechanistic anomaly detection and Finding gliders in the game of life.)
    Background
    We’d like to build AI systems that take complex actions to protect humans and maximize option value. Powerful predictive models may play an important role in such AI, either as part of a model-based planning algorithm or as part of a learned policy. Either way, this introduces the risk that we will select actions predicted to lead to good-looking outcomes without being able to distinguish predicted outcomes that are actually good from predicted outcomes where our measurements were corrupted. If tampering with sensors is easier than actually achieving our goals then we may inadvertently build very powerful systems taking creative actions to corrupt our measurements. If we iteratively improve and harden our measurements, this can lead to AI systems that work well for a long time before abruptly and catastrophically disempowering humanity.
    I consider this one of the conceptually cleanest alignment problems, and I expect similar dynamics to play a role in realistic alignment failures even if those failures aren’t this simple. ARC’s current work is focused on decisive solutions to this problem, though it looks like the same approach may also apply directly to identifying treacherous turns more generally.
    Are distinct mechanisms enough?
    ARC has been looking for training strategies that avoid this problem by leveraging only the fact that sensor tampering is “weird,” i.e. conceptually distinct from the normal mechanism giving rise to predictions of good-looking outcomes on the training distribution.
    More specifically, at training time our model predicts coherent sensor readings because it predicts that sensors reflect coherent structure in the world. But if someone tampers with sensors to show a convincing fiction, then the predicted observations are coherent because the fiction was designed to look coherent. This suggests that different mechanisms are responsible for (actions that lead to good-looking outcomes for the normal reasons) and (actions that lead to good-looking outcomes via sensor tampering). If we are able to detect that difference by looking at the internal behavior of a predictor, then we may be able to use that to avoid sensor tampering.
    It’s unclear if “distinct mechanisms” is a strong enough assumption to avoid sensor tampering. We hope that it is, and so we are trying to define formally what we mean by “distinct mechanisms” and show that it is possible to distinguish different mechanisms and that sensor tampering is always a distinct mechanism.
    If that fails, we will need to solve sensor tampering by identify additional structure in the problem, beyond the fact that it involves distinct mechanisms.
    Roadmap
    In this post I want to explore this situation in a bit more detail. In particular, I will:
    Describe what it might look like to have a pair of qualitatively distinct mechanisms that are intractable to distinguish.
    Discuss the plausibility of that situation and some reasons to think it’s possible in theory.
    Emphasize how problematic that situation would be for many existing approaches to alignment.
    Discuss four candidates for ways to solve the sensor tampering problem even if we can’t distinguish different mechanisms in general.
    Note that the existence of a pathological example of distinct-but–indistinguishable mechanisms may not be interesting to anyone other than theorists. And even for the theorists, it would still leave open many important questions...
    25 min
  • EA - Update on GWWC donation platform by Luke Freeman
    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: Update on GWWC donation platform, published by Luke Freeman on December 26, 2022 on The Effective Altruism Forum.
    We've recently received inquiries about funds donated through the donation platform hosted by Giving What We Can (GWWC), in light of the FTX bankruptcy.
    Both of the legal entities involved with GWWC (Effective Ventures Foundation and Centre for Effective Altruism USA) are financially solvent. These entities have funding sources outside of the FTX Foundation and other FTX-related entities/individuals. The GWWC related entities would be solvent even without the funds received from the FTX-related entities. Accordingly, our plan is to continue to accept and regrant donations.
    I apologise for any delays in addressing concerns about the FTX crisis. Coordinating across multiple entities and time zones is challenging, but we are committed to transparency and keeping our donors as informed as we can. As everyone can appreciate, this is an evolving situation, and we’re taking all necessary advice and making sure that we comply with all our legal duties. Part of this means that we cannot give as many details as we would like, however much we would like to.
    Thank you to all for your support and to everyone for all their hard work and dedication in responding to this incredibly difficult time.
    I am especially appreciative to all the donors who’ve stepped and will help us continue to broaden the base of donors given the funding shortfall experienced by many high-impact charities, nonprofits, and charitable funds.
    If anyone has any questions please don’t hesitate to ask – that’s what I’m here for.
    All my best,Luke
    In a comment here or by email. Although please bear in mind that it might take time to coordinate a response.
    Thanks for listening. To help us out with The Nonlinear Library or to learn more, please visit nonlinear.org.
    2 min
  • LW - Announcing: The Independent AI Safety Registry by Shoshannah Tekofsky
    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 Independent AI Safety Registry, published by Shoshannah Tekofsky on December 26, 2022 on LessWrong.
    Are you an independent AI Safety Researcher who would like to get in touch with other people in the same position? Fill out your location, work details, and collaboration interests, and you'll be added to a public google sheets file where Independent AI Safety Researchers can more easily get in touch with each other.
    Here you can browse the entries so far. In a couple of months I'll lock the file and the form, cause there is no updating mechanism for old entries. If the concept of this registry is wildly successful, I'll look into a more persistent format at that point.
    Enjoy!
    Form - So You Can Join
    File - So You Can Browse
    Thanks for listening. To help us out with The Nonlinear Library or to learn more, please visit nonlinear.org.
    2 min
  • LW - Slightly against aligning with neo-luddites by Matthew Barnett
    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: Slightly against aligning with neo-luddites, published by Matthew Barnett on December 26, 2022 on LessWrong.
    To summarize,
    When considering whether to delay AI, the choice before us is not merely whether to accelerate or decelerate the technology. We can choose what type of regulations are adopted, and some options are much better than others.
    Neo-luddites do not fundamentally share our concern about AI x-risk. Thus, their regulations will probably not, except by coincidence, be the type of regulations we should try to install.
    Adopting the wrong AI regulations could lock us into a suboptimal regime that may be difficult or impossible to leave. So we should likely be careful not endorse a proposal because it's "better than nothing" unless it's also literally the only chance we get to delay AI.
    In particular, arbitrary data restrictions risk preventing researchers from having access to good data that might help with alignment, potentially outweighing the (arguably) positive effect of slowing down AI progress in general.
    It appears we are in the midst of a new wave of neo-luddite sentiment.
    Earlier this month, digital artists staged a mass protest against AI art on ArtStation. A few people are reportedly already getting together to hire a lobbyist to advocate more restrictive IP laws around AI generated content. And anecdotally, I've seen numerous large threads on Twitter in which people criticize the users and creators of AI art.
    Personally, this sentiment disappoints me. While I sympathize with the artists who will lose their income, I'm not persuaded by the general argument. The value we could get from nearly free, personalized entertainment would be truly massive. In my opinion, it would be a shame if humanity never allowed that value to be unlocked, or restricted its proliferation severely.
    I expect most LessWrong readers to agree with me on this point — that it is not worth sacrificing a technologically richer world just to protect workers from losing their income. Yet there is a related view that I have recently heard some of my friends endorse: that nonetheless, it is worth aligning with neo-luddites, incidentally, in order to slow down AI capabilities.
    On the most basic level, I think this argument makes some sense. If aligning with neo-luddites simply means saying "I agree with delaying AI, but not for that reason" then I would not be very concerned. As it happens, I agree with most of the arguments in Katja Grace's recent post about delaying AI in order to ensure existential AI safety.
    Yet I worry that some people intend their alliance with neo-luddites to extend much further than this shallow rejoinder. I am concerned that people might work with neo-luddites to advance their specific policies, and particular means of achieving them, in the hopes that it's "better than nothing" and might give us more time to solve alignment.
    In addition to possibly being mildly dishonest, I'm quite worried such an alliance will be counterproductive on separate, purely consequentialist grounds.
    If we think of AI progress as a single variable that we can either accelerate or decelerate, with other variables held constant upon intervention, then I agree it could be true that we should do whatever we can to impede the march of progress in the field, no matter what that might look like. Delaying AI gives us more time to reflect, debate, and experiment, which prima facie, I agree, is a good thing.
    A better model, however, is that there are many factor inputs to AI development. To name the main ones: compute, data, and algorithmic progress. To the extent we block only one avenue of progress, the others will continue. Whether that's good depends critically on the details: what's being blocked, what isn't, and how.
    One consideration, which has been pointed out by many before, is that...
    8 min
  • AF - Avoiding perpetual risk from TAI by Stephen Casper
    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: Avoiding perpetual risk from TAI, published by Stephen Casper on December 26, 2022 on The AI Alignment Forum.
    Stephen Casper, [email protected]
    The goal of this post is to sort through some questions involving the difficulty of avoiding perpetual risk from transformative AI. Feedback in the comments is welcome!
    TL;DR
    Getting AI to go well means that at some point in time, the acute period of risk posed by the onset of transformative AI must end. Ending that period will require establishing a regime for transformative AI that is exclusive, benevolent, stable, and successful at alignment. This post argues that this type of regime may be very difficult to establish and that this will largely be an AI governance problem. If so, this gives an argument for emphasizing work on AI safety challenges other than just alignment.
    Intro
    Often–at least in conversation–we talk about how we need “a solution” to AI safety. And maybe AI safety is a problem that has a once-and-for-all solution. If we can avoid doom, maybe we can use aligned AI to solve our key challenges, avoid all other X-risks, and set ourselves on a sustainable course for the future. This seems possible, and it would be great. Thinking in these terms stems back at least to Nick Bostrom’s Superintelligence. Bostrom discusses the possibility that in the same way that extinction is an attractor state, meeting our cosmic endowment may be as well. If we can get AI right just once, maybe that’s the last key challenge we need to solve.
    From Superintelligence:
    However, achieving immortality would mean a lot of things would have to go right. Establishing a regime that avoids perpetual risk will be complex, and it will definitely involve more than just figuring out how to align AI systems with our goals.
    Four requirements for avoiding perpetual risk
    Suppose that highly transformative AI (TAI) technologies are someday developed and that they are powerful enough to do catastrophically dangerous things. By definition I will refer to whatever set of institutions that control the TAI and could cause catastrophic risks with it as the TAI regime. The regime could consist of human institutions, AI institutions, or both.
    To avoid perpetual AI risk, four things need to be true about the TAI regime. And to the extent that any are false, the probability of perpetual risk will rapidly increase. Note that this only applies within some sort of cosmic sphere of influence.
    Exclusivity: There will almost certainly exist bad actors who would cause major risks if they had the ability to create and deploy the TAI without being stopped at some point in the process. (For example, the same applies with nuclear weapons.) To avoid this, the regime needs to be effectively closed to those who would cause havoc if they joined it.
    Benevolence: The TAI regime needs to be one that will try to lock in a stable and good future instead of pursuing more evil or myopic goals.
    Stability: The regime needs to be stable over time and not be overthrown or degenerate into an unaligned one.
    Success at alignment: The TAI regime needs to solve outer alignment and inner alignment sufficiently well to ensure that their TAI does not itself cause major risks.
    It’s also worth adding that resilience against disasters is also really nice to have if risks become reality.
    Each of these four may be difficult
    Exclusivity:
    Developing TAI may offer substantial first-mover advantages that gives the first members of the regime a lot of influence. And there is some precedent for certain sectors that are really hard for new actors to break into and be competitive in such as the search engine industry. But exclusivity still seems hard.
    The research and development ecosystems for AI are competitive. Due in part to arms races and copycatting, there is a lot of precedent for different groups ma...
    9 min
  • EA - Announcing a subforum for forecasting & estimation by Sharang Phadke
    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 a subforum for forecasting & estimation, published by Sharang Phadke on December 26, 2022 on The Effective Altruism Forum.
    TL;DR: We’ve launched a dedicated space for discussing forecasting and estimation. Go explore it or join it if you’re interested!
    As mentioned before, the Forum team is piloting subforums that allow people interested in specific topics in effective altruism to keep up with them, have more casual conversations with a smaller group of people interested in the same topic, and engage more deeply with the topic. We’ve just kicked off our fourth subforum: forecasting & estimation!
    What goes in a “forecasting & estimation” subforum?
    Some discussions that we think people might have in this subforum:
    Predictions for 2023
    How to forecast, how to get better at forecasting, bottlenecks in forecasting
    Fermi estimation
    Specific forecasts and estimates
    And more!
    How subforums currently work — give us feedback!
    It’s early days for subforums, and we expect the features and structure to evolve significantly in the near future, so feedback is particularly useful right now. You can comment here or email us at [email protected].
    Here’s the current setup:
    All posts that are relevant to the subforum’s topic will appear in the subforum, even if they’re by someone who’s not part of the subforum.
    You can post directly to the subforum, so only people who have joined the space or are exploring it will see them. There are two approaches possible:
    Full posts in the subforum are like normal Forum posts
    Discussion threads in the subforum are a bit like shortform comments
    The default ordering of content in the subforum is by a combination of recency and upvotes (“magic”). If someone adds a comment to a post or thread in the subforum, the post will resurface at the top of the discussion.
    Every subforum has an “organizer.” You can see an outline of what they’ll be doing here.
    Notifications: if you join the subforum, you’ll have the choice to opt in to notifications of new posts and discussions. You can change this by clicking on the bell icon on the subforum page and setting your preferences accordingly (e.g. by removing “upweight on frontpage”).
    Thanks for listening. To help us out with The Nonlinear Library or to learn more, please visit nonlinear.org.
    3 min