The Nonlinear Library

The Nonlinear Library

By The Nonlinear FundEducation
Download on the App Store

All The Nonlinear Library episodes

  • LW - Plans Are Predictions, Not Optimization Targets by johnswentworth
    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: Plans Are Predictions, Not Optimization Targets, published by johnswentworth on October 20, 2022 on LessWrong.
    Imagine a (United States) high-schooler who wants to be a doctor. Their obvious high-level plan to achieve that goal is:
    Graduate high school and get into college
    Go to college, study some bio/chem/physiology, graduate and get into med school
    Go to med school, make it through residency
    Doctor!
    Key thing to notice about that plan: the plan is mainly an optimization target. When in high school, our doctor-to-be optimizes for graduating and getting into college. In college, they optimize for graduating and getting into med school. Etc. Throughout, our doctor-to-be optimizes to make the plan happen. Our doctor-to-be does not treat the plan primarily as a prediction about the world; they treat it as a way to make the world be.
    And that probably works great for people who definitely just want to be doctors.
    Now imagine someone in 1940 who wants to build a solid-state electronic amplifier.
    Building active solid-state electronic components in the early 1940’s is not like becoming a doctor. Nobody has done it before, nobody knows how to do it, nobody knows the minimal series of steps one must go through in order to solve it. At that time, solid-state electronics was a problem we did not understand; the field was preparadigmatic. There were some theories, but they didn’t work. The first concrete plans people attempted failed; implicit assumptions were wrong, but it wasn’t immediately obvious which implicit assumptions. One of the most confident predictions one might reasonably have made about solid-state electronics in 1940 was that there would be surprises; unknown unknowns were certainly lurking.
    So, how should someone in 1940 who wants to build a solid-state amplifier go about planning?
    I claim the right move is to target robust bottlenecks: look for subproblems which are bottlenecks to many different approaches/plans/paths, then tackle those subproblems. For instance, if I wanted to build a solid-state amplifier in 1940, I’d make sure I could build prototypes quickly (including with weird materials), and look for ways to visualize the fields, charge densities, and conductivity patterns produced. Whenever I saw “weird” results, I’d first figure out exactly which variables I needed to control to reproduce them, and of course measure everything I could (using those tools for visualizing fields, densities, etc). I’d also look for patterns among results, and look for models which unified lots of them.
    Those are strategies which would be robustly useful for building solid-state amplifiers in many worlds, and likely directly address bottlenecks to progress in many worlds. In our particular world, they might have highlighted the importance of high-purity silicon and dopants, or of surfaces between materials with different electrical properties, both of which were key rate-limiting insights along the path to active solid-state electronics.
    Now, when looking for those robust bottlenecks, I’d probably need to come up with a plan. Multiple plans, in fact. The point of “robust bottlenecks” is that they’re bottlenecks to many plans, after all. But those plans would not be optimization targets. I don’t treat the plans as ways to make the world be. Rather, the plans are predictions about how things might go. My “mainline plan”, if I have one, is not the thing I’m optimizing to make happen; rather, it’s my modal expectation for how I expect things to go (conditional on my efforts).
    My optimization targets are, instead, the robust bottlenecks.
    When reality throws a brick through the plans, I want my optimization target to have still been a good target in hindsight. Thus robust bottlenecks: something which is still a bottleneck under lots of different assumptions is more likely to b...
    8 min
  • LW - Covid 10/20/22: Wait, We Did WHAT? by Zvi
    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: Covid 10/20/22: Wait, We Did WHAT?, published by Zvi on October 20, 2022 on LessWrong.
    The what, in this case, was the type of thing that could cause another pandemic:
    This has not made enough people very angry, and is insufficiently widely regarded as a bad move. It happened, and we are paying for similar things to happen again.
    There’s a section on that below (you can Ctrl-F for Gain of Function).
    It was a surprisingly news-filled week for Covid in other ways as well. None of the other developments should scare you.
    Executive Summary
    We did Gain of Function research. Such research must stop.
    New variants are coming, unclear how big a wave might follow.
    CDC adds Covid-19 to childhood vaccination schedule.
    Let’s run the numbers.
    The Numbers
    Predictions
    Prediction from last week: 234k cases (+0%) and 2,500 deaths (+2%).
    Result: 223k cases (-5%) and 2,319 deaths (-6%).
    Prediction for next week: 218k cases (-3%) and 2,160 deaths (-7%).
    Now that the decline sustained itself past the holiday and its aftermath, we can be confident that existing strains continue to be on their way down. The question now shifts to the new variants, and when, whether and to what extent they will cause trouble. See the variant section for further discussion of this. Once again, the pandemic is ‘resetting’ in this fashion.
    Deaths
    Indiana didn’t report deaths or cases last week and this week likely filled in the backlog, which is probably most of why the Midwest went up instead of down.
    Cases
    The south continues to decline, half of which is Mississippi and North Carolina not reporting cases. I chose not to adjust.
    Variants
    Paper says BA.2.75.2 ‘exhibits extensive escape from neutralizing antibodies,’ which is not news.
    Which new variant will take over next?
    In the equilibrium model, any reproduction advantage is scary, since it will continue to compound until behaviors adjust or enough people get infected. What we have learned over the last few years is that there is a big reproduction advantage to being the new variant and starting from a small base. It takes a huge edge during growth to be the kind of game changer that Alpha, Delta and Omicron were.
    Here is a graph for America.
    It is clear that BA.5 has peaked and is in decline, with BQ.1.1 looking most likely to be next and looking like it has a higher reproductive advantage than Moritz is suggesting. That is enough relative growth to be somewhat worrying. It looks like more than the ‘10% advantage’ I see quoted in places like this Andy Slavitt thread. If I was convinced the advantage was only ~10%, I would be at least as confident as the 80% Slavitt reports from scientists that the winter will be highly manageable.
    As an alternative perspective, looking at variant groups as levels rather than looking at individual variants.
    As we saw last week, this is very concentrated in Germany and Austria, and to a lesser extent France. Graphs like this are effectively somewhat misleading in isolation.
    He gives a variety of other charts, including many variants of the one above. The variations for other nations all look broadly similar.
    An underappreciated fact is that case counts effectively fully reset from scratch every time a new variant takes over. If Western Europe (or at least Germany) is experiencing a surge of Level 5 cases now, and the big winter wave is going to be Level 6, then the current case counts don’t matter for the future local path of the pandemic. At all.
    Thus, we can make statements like this, which are relatively clear.
    Cornelius Roemer says on October 13 that variants like BA.4.6 are not yet impacting case counts much. This means that current case counts are still mostly irrelevant to future case counts.
    (Global cases count in terms of the mutation rate, and past cases matter for spread.)
    Think about the history of the pandemic in...
    26 min
  • LW - Introduction to abstract entropy by Alex Altair
    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: Introduction to abstract entropy, published by Alex Altair on October 20, 2022 on LessWrong.
    This post, and much of the following sequence, was greatly aided by feedback from the following people (among others): Lawrence Chan, Joanna Morningstar, John Wentworth, Samira Nedungadi, Aysja Johnson, Cody Wild, Jeremy Gillen, Ryan Kidd, Justis Mills and Jonathan Mustin. Illustrations by Anne Ore.
    Introduction & motivation
    In the course of researching optimization, I decided that I had to really understand what entropy is. But there are a lot of other reasons why the concept is worth studying:
    Information theory:
    Entropy tells you about the amount of information in something.
    It tells us how to design optimal communication protocols.
    It helps us understand strategies for (and limits on) file compression.
    Statistical mechanics:
    Entropy tells us how macroscopic physical systems act in practice.
    It gives us the heat equation.
    We can use it to improve engine efficiency.
    It tells us how hot things glow, which led to the discovery of quantum mechanics.
    Epistemics (an important application to me and many others on LessWrong):
    The concept of entropy yields the maximum entropy principle, which is extremely helpful for doing general Bayesian reasoning.
    Entropy tells us how "unlikely" something is and how much we would have to fight against nature to get that outcome (i.e. optimize).
    It is relevant to the fate of the universe.
    And it's also a fun puzzle to figure out!
    I didn't intend to write a post about entropy when I started trying to understand it. But I found the existing resources (textbooks, Wikipedia, science explainers) so poor that it actually seems important to have a better one as a prerequisite for understanding optimization! One failure mode I was running into was that other resources tended only to be concerned about the application of the concept in their particular sub-domain. Here, I try to take on the task of synthesizing the abstract concept of entropy, to show what's so deep and fundamental about it. In future posts, I'll talk about things like:
    How abstract entropy can be made meaningful on continuous spaces
    Exactly where the "second law of thermodynamics" comes from, and exactly when it holds (which turns out to be much broader than thermodynamics)
    How several domain-specific types of entropy relate to this abstract version
    Many people reading this will have some previous facts about entropy stored in their minds, and this can sometimes be disorienting when it's not yet clear how those facts are consistent with what I'm describing. You're welcome to skip ahead to the relevant parts and see if they're re-orienting; otherwise, if you can get through the whole explanation, I hope that it will eventually be addressed!
    But also, please keep in mind that I'm not an expert in any of the relevant sub-fields. I've gotten feedback on this post from people who know more math & physics than I do, but at the end of the day, I'm just a rationalist trying to understand the world.
    Abstract definition
    Entropy is so fundamental because it applies far beyond our own specific universe, the one where something close to the standard model of physics and general relativity are true. It applies in any system with different states. If the system has dynamical laws, that is, rules for moving between the different states, then some version of the second law of thermodynamics is also relevant. But for now we're sticking with statics; the concept of entropy can be coherently defined for sets of states even in the absence of any "laws of physics" that cause the system to evolve between states. The example I keep in my head for this is a Rubik's Cube, which I'll elaborate on in a bit.
    The entropy of a state is the number of bits you need to use to uniquely distinguish it.
    Some useful things t...
    30 min
  • EA - Growing the US tofu market - a roadmap by George Stiffman
    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: Growing the US tofu market - a roadmap, published by George Stiffman on October 20, 2022 on The Effective Altruism Forum.
    This post is a roadmap to growing the US tofu market. There is a ton of delicious impact to be had, and I hope more of you in the alt protein space will consider joining me.
    (Thank you to Kris Chari and Nick Corvino for helpful feedback.)
    A yuba-based Napoleon pastry from my forthcoming book, Broken Cuisine. Photo credit: Ryan Tanaka
    Why should I care about tofu?
    When I first visited China, I assumed tofu was just a pasty, white block.
    I never would’ve imagined that tofus could melt. or that some varieties tasted like aged cheese. or that others had a bread-like crumb!
    Chinese tofus are some of the most culinarily exciting plant-based proteins in existence. Yet, hardly anyone outside the Chinese community knows they exist!
    Yunnan juicy tofu, grilled over charcoal with seasoned chili powder.
    The average American eats just 1 block of conventional tofu per year. Yet, even this tiny consumption, added up, is enough to displace millions of animal lives from factory farms each year.
    While I personally love firm, soft, and silken tofu, these ingredients are poor fits for western cooking styles and taste preferences. In contrast, several Chinese tofus are much, much, much better fits. If we can find the "avocado toast" of these tofus, expand ingredient supply, and successfully promote, the market for these varieties could thus be many times larger, someday saving 10s or 100s of millions of animal lives each year.
    Beyond directly reducing demand for meat, promoting Chinese tofus could have several other benefits:
    Chinese tofus could make plant-based cooking more appealing to chefs in ways that meat analogs (plant-based meats, cultivated meat, etc.) cannot. If true, this could turn some of our biggest stalwarts to plant-based progress into allies and evangelists, shifting the culture around dining. Shifting culture could have more impact than merely substituting meat for mock meat or tofu.
    Better tasting vegan food seems necessary for moral circle expansion. Chinese tofus are one of the best opportunities to expand vegan cooking, as their unique culinary properties can do things that meat and vegetables cannot. (This seems especially pressing if you have short AI timelines and worry about value lock-in, but there don't seem to be many interventions currently being pursued.)
    Making tofu a high prestige food in the States could possibly raise it's status in China. After living in China for several years and studying the Chinese farmed animal welfare space throughout 2020, I am pretty pessimistic about near-term prospects for animals. Promoting tofu in the west, to raise its status in China, could, however, be a straightforward and significant way to contribute.
    Because traditional plant protein advocacy is so neglected within the alt protein space, there could be a lot of information value in trying to promote Chinese tofus.
    Beyond being massive in scale and very neglected, I am hopeful that this opportunity is tractable, as you'll see below...
    Wait! What is Chinese tofu?
    China is the birthplace of tofu and has over 20 distinct varieties. While some of these varieties are common across Asia, the majority are specific to Chinese cooking. There are a few, in particular, that seem to be great fits for western cooking styles:
    Shanghai tofu ( shanghaisuji) - ultra high in protein, rich/eggy/custardy flavor, bready structure that can be cooked into cakes, soup dumplings, "crostini", protein crumbles, etc.
    Spongy tofu ( qianyedoufu) - a fishcake-like tofu, pre-seasoned (NOT bland), delicious blended, braised, or grilled, can be dense and chewy or light and silky.
    Fermented tofu ( furu) - a funky, cheesy, umami seasoning that's blended into sauces or rubbed onto brea...
    46 min
  • EA - A couple of pressing jobs in biosecurity at the moment (Oct 2022) by Clifford
    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 of pressing jobs in biosecurity at the moment (Oct 2022), published by Clifford on October 20, 2022 on The Effective Altruism Forum.
    There are a lot of jobs out there. I have a hunch that it would be useful to get a sense of what field leaders thought were the most pressing roles to fill to motivate people to apply or share with people they know.
    I asked Chris Bakerlee (Senior Program Associate for Biosecurity and Pandemic Preparedness at Open Philanthropy) for three biosecurity roles he is excited to see filled right now. I was excited to get his view as I feel like he has a good overview of the opportunities that are available within biosecurity and which seem highly impactful to fill. He highlighted the following two roles.
    For more information on why people prioritise Global Catastrophic Biological Risk, I recommend reading 80000 hours profile.
    This post is (in part) a test. Please give me feedback on this post here, in a DM, or in the comments. I’d be particularly interested to know if you found this post useful and/or if this was the first time you’d heard about any of these roles.
    Open Philanthropy, Biosecurity and Pandemic Preparedness, Executive Assistant
    Flexible location (within the US and US time zones). The team is currently split between Boston and the Bay Area.
    Who is it for?
    Someone with project management, organization, and prioritization skills.
    What’s the role in a nutshell?
    The executive assistant role would unlock valuable time that will be spent on research, grantmaking, and other essential activities. The role would focus on providing administrative and operational support for the team. Responsibilities include:
    providing executive assistant-style support to Senior Program Officer, Andrew Snyder-Beattie
    managing and organizing team documents
    doing ad hoc research or organizational tasks (e.g. proofreading documents),
    Read more about the impact case for working in executive assistant roles here.
    Read full job description.
    Why this organisation?
    Open Philanthropy is one of the biggest funders in effective altruism. They have given over $131 million for projects aimed at reducing biological risks. You can read more about Open Philanthropy here.
    Learn more
    Senior Program Officer / Senior Director, Global Biological Policy and Programs (NTI | bio)
    Washington D.C.
    Who is this for?
    Someone with 7-10 years of policy, research, and/or project management experience, including in biology/biotechnology, international security, global health, or international affairs.
    What’s the role in a nutshell?
    This role presents an excellent chance to work alongside Jaime Yassif and the Nuclear Threat Initiative’s ‘Bio’ team to advance creative solutions to challenging problems in biotech governance and bioweapons nonproliferation. Listen to Jaime’s episode on the 80,000 hours podcast here which describes some of the problems they are tackling.
    Responsibilities include:
    overseeing research on the development of practical, innovative risk-reduction approaches;
    managing projects, budgets, and staff;
    convening international events and managing relationships with global stakeholders;
    drafting internal leadership memos and briefing documents;
    Read full job description
    Why this organisation?
    The Nuclear Threat Initiative (NTI) works with governments, scientists, and citizens to prevent catastrophic attacks with weapons of mass destruction and disruption—nuclear, biological, radiological, chemical and cyber. You can read more about NTI here.
    Learn more
    Thanks for listening. To help us out with The Nonlinear Library or to learn more, please visit nonlinear.org.
    4 min
  • LW - Should we push for requiring AI training data to be licensed? by ChristianKl
    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: Should we push for requiring AI training data to be licensed?, published by ChristianKl on October 19, 2022 on LessWrong.
    We frequently speak about AI capability gain being bad because it shortens the timeframe for AI safety research. In that logic, taking steps to decrease AI capability would be worthwhile.
    At the moment the large language models are trained with a lot of data without the company, that trains the language model, licensing the data. If there would be a requirement to license the required data, that would severely reduce the available data for language models and reduce their capabilities.
    It's expensive to fight lawsuits in the United States. Currently, there are artists who feel like their rights are violated by Dalle 2 using their art as training data. Similar to how Thiel funded the Gawker lawsuits, it would be possible to support artists in a suit against OpenAI to require OpenAI to license images for training Dalle 2. If such a lawsuit is well-funded it will be much more likely that a precedent for requiring data licensing gets set which would slow down AI development.
    I'm curious about what people who think more about AI safety than myself think about such a move. Would it be helpful?
    Thanks for listening. To help us out with The Nonlinear Library or to learn more, please visit nonlinear.org.
    2 min
  • AF - Notes on "Can you control the past" by Nate Soares
    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: Notes on "Can you control the past", published by Nate Soares on October 20, 2022 on The AI Alignment Forum.
    The following is a (lightly edited version of a) series of notes I sent Joe Carlsmith about his essay, Can you control the past?. It's addressed to Joe, but it seems worth publishing here while I'm on the topic of decision theory. I’ve included some of his comments, and my replies, below.
    I only recently skimmed Can you control the past?, and have a couple notes that you may or may not be interested in. (I'm not under the impression that this matters a ton, and am writing this recreationally.)
    First: this is overall a great review of decision theories. Better than most I've seen. Nice.
    Now, onto some more substansive points.
    Who am I?
    I think a bunch of your sense of oddness about the "magic" that "you can write on whiteboards light-years away" is stemming from a faulty framing you have. In particular, the part where the word "you" points to a single physical instantiation of your algorithm in the universe. I'd say: insofar as your algorithm is multiply instantiated throughout the universe, there is no additional fact about which one is really you.
    For analogy, consider tossing a coin in a quantum-mechanical universe, and covering it with your hand. The coin is superpositioned between heads and tails, and once you look at it, you'll decohere into Joe-who-saw-heads and Joe-who-saw-tails, both of whom stem from Joe-who-hasn't-looked-yet. So, before you look, are you Joe-who-saw-heads or Joe-who-saw-tails?
    Wrong question! These two entities have not yet diverged; the pasts of those two separate entities coincide. The word "you", at the time before you split, refers to ~one configuration. The time-evolution splits the amplitude on that configuration between ~two distinct future configurations, and once they've split (by making different observations), each will be able to say "me" in a way that refers to them and not the other, but before the split there is no distinction to be made, no extra physical fact, and no real question as to whether pre-split Joe "is" Joe-who-will-see-heads versus Joe-who-will-see-tails.
    (It's also maybe informative to imagine what happens if the quantum coin is biased. I'd say, even when the coin is 99.99999% biased towards heads, it's still the case that there isn't a real question about whether Joe-who-has-not-looked-at-the-coin is Joe-who-will-see-heads versus Joe-will-see-tails. There is a question of to what degree Joe-who-has-not-looked becomes Joe-who-saw-heads versus Joe-who-saw-tails, but that's a different sort of question.)
    One of my most-confident guesses about anthropics is that being multiply-instantiated in other ways is analogous. For instance, if there are two identical physical copies of you (in physical rooms that are identical enough that you're going to make the same observations for the length of the hypothetical, etc.), then my guess is that there isn't a real question about which one is you. They are both you. You are the pattern, not the meat.
    This person may become multiple people in the future, insofar as they see different things in different places-that-embed-them. But before the differing observations come in, they're both you. You can tell because the situation is symmetric: once you know all the physical facts, there's no additional bit telling you which one is "you".
    From this perspective, the "magic" is much less mysterious: whenever you are multiply-instantiated, your actions are also multiply-instantiated. If you're multiply-instantiated in two places separated by a 10-light-year gap, then when you act, the two meat-bodies move in the same way on each side of the gap. This is all much less surprising once you acknowledge that "you" refers to everything that instantiates you(-who-have-seen-what-you-have...
    33 min
  • AF - Scaling Laws for Reward Model Overoptimization by leogao
    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: Scaling Laws for Reward Model Overoptimization, published by leogao on October 20, 2022 on The AI Alignment Forum.
    TL;DR: Reward model (RM) overoptimization in a synthetic-reward setting can be modelled surprisingly well by simple functional forms. The coefficients also scale smoothly with scale. We draw some initial correspondences between the terms of the functional forms and the Goodhart Taxonomy. We suspect there may be deeper theoretical reasons behind these functional forms, and hope that our work leads to a better understanding of overoptimization.
    Some other results:
    We compare two different methods of optimization (RL, best-of-n); RL consumes much more KL distance than best-of-n for the same amount of optimization.
    We show that using KL distance between the initial and optimized policies is not a reliable measure of optimization power when comparing different methods. We also find that penalizing based on the KL distance in RL does not change the KL distance--gold reward frontier in our setting.
    We find some very preliminary evidence that at our scales, scaling the policy does not substantially increase the amount of optimization pressure placed on the RM. Further study of this effect could be relevant to some models of inner optimization.
    With a few additional assumptions, our functional form also makes some predictions about iterated RLHF (that it will reduce Extremal Goodhart but not Regressional Goodhart).
    This setup only captures the effect of overoptimizing a learned RM relative to using the ground truth directly. In particular, this setup does not directly capture any mismatch between the ground truth labels and the human intent, which plausibly contains a majority of the difficulty of outer alignment
    If you're interested in the intersection of alignment theory and empirical research, we're hiring! We want to gain insight on things like Goodhart's Law, ELK, and inner alignment via experiments on large language models. Shoot me (leogao) a DM if you're interested.
    Select figures from the paper
    Thanks for listening. To help us out with The Nonlinear Library or to learn more, please visit nonlinear.org.
    3 min
  • EA - Announcing Squigglepy, a Python package for Squiggle by Peter Wildeford
    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 Squigglepy, a Python package for Squiggle, published by Peter Wildeford on October 19, 2022 on The Effective Altruism Forum.
    Squiggle is a "simple programming language for intuitive probabilistic estimation". It serves as its own standalone programming language with its own syntax, but it is implemented in JavaScript.
    I like the features of Squiggle and intend to use it frequently, but I also frequently want to use similar functionalities in Python, especially alongside other Python statistical programming packages like Numpy, Pandas, and Matplotlib.
    The squigglepy package here implements many Squiggle-like functionalities in Python.
    The package also has useful utility functions for Bayesian networks (using rejection sampling), pooling forecasts (via weighted geometric mean of odds and others), laplace (including the time-invariant version), and kelly betting.
    The package and documentation are available on GitHub.
    The package can be downloaded from Pypi using pip install squigglepy.
    This package is unofficial and supported by myself and Rethink Priorities. It is not affiliated with or associated with the Quantified Uncertainty Research Institute, which maintains the Squiggle language (in JavaScript).
    This package is also new and not yet in a stable production version, so you may encounter bugs and other errors. Please report those so they can be fixed. It's also possible that future versions of the package may introduce breaking changes.
    This package is available under an MIT license.
    Thanks for listening. To help us out with The Nonlinear Library or to learn more, please visit nonlinear.org.
    2 min
  • EA - Announcing VIVID: A new EA organization aspiring to scale effective self-improvement & reflection by GidiKadosh
    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 VIVID: A new EA organization aspiring to scale effective self-improvement & reflection, published by GidiKadosh on October 19, 2022 on The Effective Altruism Forum.
    Summary
    We’re excited to announce VIVID - a new organization aspiring to scale effective self-improvement & reflection. With an investment of $1.75M from the Future Fund regranting program, we’re building an app that helps individuals find the most effective ways to remove internal obstacles.
    Over the course of the next two years, VIVID will focus mainly on EAs. Our new app is available here. The app is soon coming out of beta phase, so we’re looking for feedback, positive and negative! You can send feedback easily by taking a screenshot anywhere on the app.
    Overview
    There is an inherent problem with self-change advice. On the one hand, it has to be generic because it must be applicable to a wide range of subjects and individuals. On the other hand, individuals can’t really apply the generic advice as it is - they need to complete the last mile of customization in order to apply it.This is the point of failure of many good theories and best practices that are found in self-help books, academic articles, and even advice from therapists - they are often lost in adaptation.
    There are two possible solutions to this problem. The first is to make customization easier. The second is to assume that we don’t need to create all of the possible customized self-change plans that could ever exist, but rather just find the most effective ones for common problems. For instance, find the five most effective plans for the specific context of impostor syndrome within the specific context of EA.
    How can we find them? And do so for thousands of other common use cases? This is why we started VIVID.
    The VIVID app helps individuals find the most effective ways to remove internal obstacles, by allowing extensive customization and self-testing, and by empowering wellbeing professionals to assist in this process.With VIVID, individuals continuously self-improve by working on multiple plans - each plan addressing a mindset that underlies an internal obstacle.
    One of VIVID's cofounders is Dr. Tal Ben-Shahar - a leading expert in positive psychology, who taught two of Harvard's most popular courses of all time. Ben-Shahar is the author of several worldwide best-selling books, and his upcoming book discusses the methodology of VIVID.
    Theory of change
    Our theory of change is divided into two main parts:
    Short-term: Assist promising individuals in overcoming internal obstacles. This means faster progress on the world's most pressing problems when used by high-impact individuals like EAs. VIVID is meant to grow better at this point over time - the main idea behind its design is that it learns over time which self-change methods are most effective for each category (e.g. self-esteem, guilt, friendship, etc.), and for which user profiles.
    Long-term: The a priori chances of projects such as this becoming widespread are low (but we’re optimistic!). However, the upsides are stellar. We could:
    Broadly promote the value of self-reflection. We believe that this point is of high importance, particularly in terms of mitigating risks associated with value lock-in (this idea is explained in WWOTF, ch. 3, and we will share a more thorough write-up about this soon).
    Accelerate progress in the wellbeing field and align it with large-scale evidence. In other words, we may discover scalable methods to improve wellbeing and individual potential in dozens of common use cases, relevant both to high-income and low-income countries. There are very few interventions that aim to improve wellbeing at scale, which is generally important - but depending on your assumptions regarding the current baseline of wellbeing, it might be even more important.
    Fig...
    9 min