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: EA-Aligned Impact Investing: Mind Ease Case Study, published by Brendon Wong on November 15, 2021 on The Effective Altruism Forum.

Summary

We use the term “impact investing” to refer to any investment made for the dual purpose of having a positive impact and generating a financial return

Impact investing has received relatively little attention from an EA perspective

We review an investment case study produced by the Total Portfolio Project (TPP) leveraging research from Lionheart Ventures, Let’s Fund, and Rethink Priorities

The case study uses TPP’s Total Portfolio Return (TPR) framework, a rigorous framework which can be used to assess the value of an impact investment relative to other ways of generating impact

The case study estimates that an impact investment into Mind Ease averts DALYs at $150 per dollar invested in expectation, comparing favorably to GiveWell top charities before financial returns are taken into account

We show how the TPR framework can be used to determine that, for a donor who is already making the best possible use of donations and investing to give, investing in Mind Ease appears better than either additional donations or more investing to give

Given the lack of prior research in this area and the potential for impact investing to significantly benefit many cause areas, we recommend that the EA community direct more research and attention to this area.

We highlight several ways to contribute including: dialogue and movement building, evaluations, research, entrepreneurship, and shifting money towards impact investing.

Background and Implications

Impact investing has largely been neglected by the EA community. This is perhaps due to the perception that most impact investing is relatively low impact per dollar invested. In turn, EA principles are largely neglected within impact investing. We believe there could be benefits on both sides to exploring the potential of impact investing to produce effective results.

Complicating things is the fact that the term “impact investing” has many possible interpretations and connotations. It generally refers to one or more strategies within a disparate range of financial strategies that are associated (correctly or not) with “impact,” from buying (or not buying) certain stocks, to providing loans to nonprofits and disadvantaged communities, to funding early-stage startups with a specific social mission.

We only want to distinguish an investment as impactful if results are generated that would not occur in expectation without the investment. Among the various interpretations of the term impact investing, we believe that only some of these approaches offer potential for impact. Thus, we will use the term “impact investing” in this article to refer to any investment that is expected to have a positive impact on the world and generate financial returns. Such investments could take the form of anything from early stage startup investments to ESG-related strategies in liquid stock markets (but note that ESG does not imply impact).

There are many parallels between the impact investing space at present and the donation space before EA emerged. Most EAs believe the impact of charitable donations has a heavy-tailed distribution, meaning a small number of donations/charities have a much higher impact (sometimes orders of magnitude more) than donations/charities on average. We think impact investing likely has a distribution of impact similar to or even more extreme than donations (see here for a technical discussion). Analogous to the early days of EA, there is no evaluator like GiveWell using rigorous evaluation techniques to find the most impactful investing opportunities.

In this article, we share what we think is the first work in the EA community attempting to rigorously evaluate a specific impact investment ...

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: New Effective Thesis services and opportunities to get involved, published by DavidJanku on November 15, 2021 on The Effective Altruism Forum.

What is Effective Thesis?

Effective Thesis is an organisation that supports students to begin research careers that significantly improve the world. We do this primarily by helping students identify important problems where further research could have a big impact. Previously, our main service was individual coaching aiming to help students with their final thesis topic choices (see more details in our 2020 post on updates from 2019). Recently, we have launched a few more services with the same goal that we want to introduce here.

Coaching

This is the core service that we have been providing for the last few years. We assist students individually with the general process of choosing their thesis/PhD topic and connect them with relevant resources, other students with the same focus and more experienced domain experts. You can apply for our coaching here.

Our current team consists of one Head of Coaching Management, one Head of Coaching Development and three volunteer coaches. As our organisation grows, we are continuously looking for promising new volunteer coaches to complement our team. See the Get involved section below for more information.

Help with finding a supervisor

Finding a good supervisor is often a struggle for students planning to pursue research careers. At the same time, we believe that choice of supervisor can make a significant difference to a student’s development as a researcher. We have decided to create resources to support students searching for supervisors and help them make better decisions.We are compiling, on an ongoing basis, a list of academics who we believe could be a good fit to act as undergraduate/masters/PhD thesis supervisors for the students interested in doing research in one of the research directions we recommend. We don't match and introduce students to potential supervisors ourselves, but we give students instructions on how to communicate with potential supervisors.

You can apply for access to this database here.We are also building a series of posts covering how to find potential supervisors (with an emphasis on choosing PhD supervisors), how to choose between multiple potential supervisors, and other key questions students should ask themselves before embarking on a PhD. If you can supervise students interested in some of the research directions we recommend and would like to be added to our database, please fill in this form. Similarly, please fill in the form if you know anyone who you think would be a good supervisor in any of the research directions mentioned.

Sharing early career research-related opportunities

We are running an opportunity search service for students looking for research opportunities in some of our prioritised research directions. The opportunities we cover include jobs, conferences, scholarships, internships, awards, collaboration offers, calls for papers, seminars, reading groups and funding. We only share research opportunities and focus on opportunities suited to students and early career researchers with up to 2 years of professional experience. You can sign up for this service here.We try to cover as large a space of opportunities as possible and thus rely on multiple people contributing to our database of opportunities whenever they find out about something relevant. If you’d like to help us search for these opportunities in a given research direction you're interested in/have networks in, we would be very happy to have you join our team.

See the Get involved section below for more information.If you know about a one-off opportunity that could fit our focus and you would like us to share it with other students (e.g. you run research internships, are hiri...

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: New Effective Thesis services and opportunities to get involved, published by DavidJanku on November 15, 2021 on The Effective Altruism Forum.

What is Effective Thesis?

Effective Thesis is an organisation that supports students to begin research careers that significantly improve the world. We do this primarily by helping students identify important problems where further research could have a big impact. Previously, our main service was individual coaching aiming to help students with their final thesis topic choices (see more details in our 2020 post on updates from 2019). Recently, we have launched a few more services with the same goal that we want to introduce here.

Coaching

This is the core service that we have been providing for the last few years. We assist students individually with the general process of choosing their thesis/PhD topic and connect them with relevant resources, other students with the same focus and more experienced domain experts. You can apply for our coaching here.

Our current team consists of one Head of Coaching Management, one Head of Coaching Development and three volunteer coaches. As our organisation grows, we are continuously looking for promising new volunteer coaches to complement our team. See the Get involved section below for more information.

Help with finding a supervisor

Finding a good supervisor is often a struggle for students planning to pursue research careers. At the same time, we believe that choice of supervisor can make a significant difference to a student’s development as a researcher. We have decided to create resources to support students searching for supervisors and help them make better decisions.We are compiling, on an ongoing basis, a list of academics who we believe could be a good fit to act as undergraduate/masters/PhD thesis supervisors for the students interested in doing research in one of the research directions we recommend. We don't match and introduce students to potential supervisors ourselves, but we give students instructions on how to communicate with potential supervisors.

You can apply for access to this database here.We are also building a series of posts covering how to find potential supervisors (with an emphasis on choosing PhD supervisors), how to choose between multiple potential supervisors, and other key questions students should ask themselves before embarking on a PhD. If you can supervise students interested in some of the research directions we recommend and would like to be added to our database, please fill in this form. Similarly, please fill in the form if you know anyone who you think would be a good supervisor in any of the research directions mentioned.

Sharing early career research-related opportunities

We are running an opportunity search service for students looking for research opportunities in some of our prioritised research directions. The opportunities we cover include jobs, conferences, scholarships, internships, awards, collaboration offers, calls for papers, seminars, reading groups and funding. We only share research opportunities and focus on opportunities suited to students and early career researchers with up to 2 years of professional experience. You can sign up for this service here.We try to cover as large a space of opportunities as possible and thus rely on multiple people contributing to our database of opportunities whenever they find out about something relevant. If you’d like to help us search for these opportunities in a given research direction you're interested in/have networks in, we would be very happy to have you join our team.

See the Get involved section below for more information.If you know about a one-off opportunity that could fit our focus and you would like us to share it with other students (e.g. you run research internships, are hiri...

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: Sleep: effective ways to improve it, published by Ben Williamson on November 15, 2021 on The Effective Altruism Forum.

Introduction

Our research found that melatonin supplementation, cognitive behavioural therapy for insomnia (CBT-I), light regulation, mindfulness-based stress reduction (MBSR), and improved night-time air circulation could all be effective ways for you to increase the quality and quantity of your sleep.

This is a summary of research conducted into the most worthwhile practices for sleeping better. The post is the first in a series looking into the most effective ways people can improve their wellbeing, aiming to present this information as simply and practically as possible.

Thanks to the EA Infrastructure Fund for financing this project. If you find this information useful and/ or take up any of the suggestions, please let me know in the comments or a personal message!

Important note: None of the following constitutes professional medical advice. Some of the interventions suggested have risks of negative side effects that are discussed below. We encourage you to experiment with these practices but please be cautious in doing so and take any risks seriously.

Top takeaway

Our principal recommendations for improving sleep quality are (in order):

Melatonin supplements: 0.3mg (300mcg) daily taken two hours before bed.

CBT-I: a six-to-seven-week, self-guided course in cognitive behavioural therapy for insomnia, accessed through an app.

Light therapy: greatly increasing exposure to bright light during the day, either through building a lumenator or purchasing a SAD lamp.

Improved night-time air circulation: opening a window to reduce overnight CO2 accumulation.

Mindfulness-based stress reduction: mindfulness training through an app with a focus on sleep.

Key Findings

Following a broad search of possible interventions, this review evaluated the effectiveness of 11 practices for improving sleep quality and quantity using an adapted weighted factor model.[1]

The five most promising interventions are listed below along with a brief explanation. More detailed assessments of all 11 practices are provided further down.

This research compared interventions across six criteria: strength of evidence, quality of evidence, ease of implementation, risk, externalities, and novelty. Full results from the model, along with reasoning for the metrics used and their respective weightings, can be found here.

Melatonin

Ranking: 1st

Intervention: 0.3mg daily supplementation (e.g. Options A, B and C)

Summary: Melatonin is a natural sleep hormone that can improve sleep quality and quantity when taken in small, daily doses. Melatonin supplementation appears highly practical, with a low risk of notable side effects as well as possible spillover benefits for other conditions.

CBT-I

Ranking: 2nd

Intervention: Completion of an app-based CBT-I course (e.g. Dozy; CBT-I Coach)

Summary: Cognitive Behavioural Therapy for Insomnia (CBT-I) involves a combination of reframing negative thoughts around sleep, improving sleep hygiene, and implementing sleep restriction. We found good evidence that completing a self-guided CBT-I course via an app can be an effective way to sleep better.

Light therapy

Ranking: 3rd

Intervention: Increasing indoor lighting brightness, preferably up to 10,000 lux or more (e.g. a ‘lumenator’ or a Seasonal Affective Disorder [SAD] lamp).

Summary: Multiple studies have found that significantly increased exposure to bright morning light improves sleep quality and quantity. We found negligible risks to this practice as well as possible benefits to mood and alertness, though this does require some time and money to set up.

Improved night-time air circulation

Ranking: 4th

Intervention: Leaving a window or internal door open while sleeping.

Summary: There is reasonable evide...

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: Sleep: effective ways to improve it, published by Ben Williamson on November 15, 2021 on The Effective Altruism Forum.

Introduction

Our research found that melatonin supplementation, cognitive behavioural therapy for insomnia (CBT-I), light regulation, mindfulness-based stress reduction (MBSR), and improved night-time air circulation could all be effective ways for you to increase the quality and quantity of your sleep.

This is a summary of research conducted into the most worthwhile practices for sleeping better. The post is the first in a series looking into the most effective ways people can improve their wellbeing, aiming to present this information as simply and practically as possible.

Thanks to the EA Infrastructure Fund for financing this project. If you find this information useful and/ or take up any of the suggestions, please let me know in the comments or a personal message!

Important note: None of the following constitutes professional medical advice. Some of the interventions suggested have risks of negative side effects that are discussed below. We encourage you to experiment with these practices but please be cautious in doing so and take any risks seriously.

Top takeaway

Our principal recommendations for improving sleep quality are (in order):

Melatonin supplements: 0.3mg (300mcg) daily taken two hours before bed.

CBT-I: a six-to-seven-week, self-guided course in cognitive behavioural therapy for insomnia, accessed through an app.

Light therapy: greatly increasing exposure to bright light during the day, either through building a lumenator or purchasing a SAD lamp.

Improved night-time air circulation: opening a window to reduce overnight CO2 accumulation.

Mindfulness-based stress reduction: mindfulness training through an app with a focus on sleep.

Key Findings

Following a broad search of possible interventions, this review evaluated the effectiveness of 11 practices for improving sleep quality and quantity using an adapted weighted factor model.[1]

The five most promising interventions are listed below along with a brief explanation. More detailed assessments of all 11 practices are provided further down.

This research compared interventions across six criteria: strength of evidence, quality of evidence, ease of implementation, risk, externalities, and novelty. Full results from the model, along with reasoning for the metrics used and their respective weightings, can be found here.

Melatonin

Ranking: 1st

Intervention: 0.3mg daily supplementation (e.g. Options A, B and C)

Summary: Melatonin is a natural sleep hormone that can improve sleep quality and quantity when taken in small, daily doses. Melatonin supplementation appears highly practical, with a low risk of notable side effects as well as possible spillover benefits for other conditions.

CBT-I

Ranking: 2nd

Intervention: Completion of an app-based CBT-I course (e.g. Dozy; CBT-I Coach)

Summary: Cognitive Behavioural Therapy for Insomnia (CBT-I) involves a combination of reframing negative thoughts around sleep, improving sleep hygiene, and implementing sleep restriction. We found good evidence that completing a self-guided CBT-I course via an app can be an effective way to sleep better.

Light therapy

Ranking: 3rd

Intervention: Increasing indoor lighting brightness, preferably up to 10,000 lux or more (e.g. a ‘lumenator’ or a Seasonal Affective Disorder [SAD] lamp).

Summary: Multiple studies have found that significantly increased exposure to bright morning light improves sleep quality and quantity. We found negligible risks to this practice as well as possible benefits to mood and alertness, though this does require some time and money to set up.

Improved night-time air circulation

Ranking: 4th

Intervention: Leaving a window or internal door open while sleeping.

Summary: There is reasonable evide...

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: My understanding of the alignment problem, published by Daniel Dewey on November 15, 2021 on The AI Alignment Forum.

I've been clarifying my own understanding of the alignment problem over the past few months, and wanted to share my first writeups with folks here in case they're useful:

The site currently has 3 pages:

The case for risk: how deep learning could become very influential, training problems that could lead models to behave in systematically harmful ways, and what I think we should do about it. Inspired mainly by What failure looks like.

Fermi estimate of future training runs: a short AI timelines estimate inspired by Forecasting transformative AI.

Applications of high-capability models: some notes on how high-capability models could actually be trained, and how their behavior could become highly influential.

None of the ideas on the site are particularly new, and as I note, they're not consensus views, but the version of the basic case I lay out on the site is very short, doesn't have a lot of outside dependencies, and is put together out of nuts-and-bolts arguments that I think will be useful as a starting point for alignment work. I'm particularly hoping to avoid semantic arguments about "what counts as" inner vs outer alignment, optimization, agency, etc., in favor of more mechanical statements of how models could behave in different situations.

I think some readers on this forum will already have been thinking about alignment this way, and won't get a lot new out of the site; some (like me) will find it to be a helpful distillation of some of the major arguments that have come out over the past ~5 years; and some will have disagreements (which I'm curious to hear about).

I thought about posting all of this directly on the Alignment Forum / LessWrong, but ultimately decided I wanted a dedicated home for these ideas.

Out of everything on the site, the part I'm most hoping will be helpful to you is my (re)statement of two main problems in AI alignment. These map roughly onto outer and inner alignment, though different people use those terms differently, so not everyone will agree:

As models become more capable, it looks like currently known training methods will run into fundamental safety problems, and become increasingly likely to produce models that behave in systematically harmful ways:

1. Evaluation breakdown: As a model’s behavior becomes more sophisticated, it will reach a point where an automated reward function or human evaluator will not be able to fully understand its behavior. In many domains, it will then become possible for models to get good evaluations by producing adversarial behaviors that systematically hide bad outcomes, maximize the appearance of good outcomes, and generally seek to control the information flowing to the evaluator instead of achieving the desired results.

Evaluation breakdown would produce high-capability models that appear to work as intended, but that will behave in arbitrarily harmful ways when that behavior is useful for producing good evaluations; this would be broadly analogous to a company using its advantages in resources, personnel, and specialized knowledge to keep regulators and the public in the dark about harms.

2. High-level distribution shift: Even if evaluation breakdown is avoided, a model may behave arbitrarily badly when its input distribution is different from its training distribution. Especially harmful behavior could occur under “high-level” distribution shifts – shifts that leave the low-level structure of the domain unchanged (e.g. causal patterns that allow prediction of future observations or consequences of actions), but change some high-level features of the broader situation the model is operating in. Since the basic structure of the domain is unchanged, a model could continue to behave ...

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: My understanding of the alignment problem, published by Daniel Dewey on November 15, 2021 on The AI Alignment Forum.

I've been clarifying my own understanding of the alignment problem over the past few months, and wanted to share my first writeups with folks here in case they're useful:

The site currently has 3 pages:

The case for risk: how deep learning could become very influential, training problems that could lead models to behave in systematically harmful ways, and what I think we should do about it. Inspired mainly by What failure looks like.

Fermi estimate of future training runs: a short AI timelines estimate inspired by Forecasting transformative AI.

Applications of high-capability models: some notes on how high-capability models could actually be trained, and how their behavior could become highly influential.

None of the ideas on the site are particularly new, and as I note, they're not consensus views, but the version of the basic case I lay out on the site is very short, doesn't have a lot of outside dependencies, and is put together out of nuts-and-bolts arguments that I think will be useful as a starting point for alignment work. I'm particularly hoping to avoid semantic arguments about "what counts as" inner vs outer alignment, optimization, agency, etc., in favor of more mechanical statements of how models could behave in different situations.

I think some readers on this forum will already have been thinking about alignment this way, and won't get a lot new out of the site; some (like me) will find it to be a helpful distillation of some of the major arguments that have come out over the past ~5 years; and some will have disagreements (which I'm curious to hear about).

I thought about posting all of this directly on the Alignment Forum / LessWrong, but ultimately decided I wanted a dedicated home for these ideas.

Out of everything on the site, the part I'm most hoping will be helpful to you is my (re)statement of two main problems in AI alignment. These map roughly onto outer and inner alignment, though different people use those terms differently, so not everyone will agree:

As models become more capable, it looks like currently known training methods will run into fundamental safety problems, and become increasingly likely to produce models that behave in systematically harmful ways:

1. Evaluation breakdown: As a model’s behavior becomes more sophisticated, it will reach a point where an automated reward function or human evaluator will not be able to fully understand its behavior. In many domains, it will then become possible for models to get good evaluations by producing adversarial behaviors that systematically hide bad outcomes, maximize the appearance of good outcomes, and generally seek to control the information flowing to the evaluator instead of achieving the desired results.

Evaluation breakdown would produce high-capability models that appear to work as intended, but that will behave in arbitrarily harmful ways when that behavior is useful for producing good evaluations; this would be broadly analogous to a company using its advantages in resources, personnel, and specialized knowledge to keep regulators and the public in the dark about harms.

2. High-level distribution shift: Even if evaluation breakdown is avoided, a model may behave arbitrarily badly when its input distribution is different from its training distribution. Especially harmful behavior could occur under “high-level” distribution shifts – shifts that leave the low-level structure of the domain unchanged (e.g. causal patterns that allow prediction of future observations or consequences of actions), but change some high-level features of the broader situation the model is operating in. Since the basic structure of the domain is unchanged, a model could continue to behave ...

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: "Summarizing Books with Human Feedback" (recursive GPT-3), published by gwern on November 15, 2021 on The AI Alignment Forum.

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: "Summarizing Books with Human Feedback" (recursive GPT-3), published by gwern on November 15, 2021 on The AI Alignment Forum.

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: Rethink Priorities - 2021 Impact and 2022 Strategy, published by Peter Wildeford on November 15, 2021 on The Effective Altruism Forum.

Summary

Our purpose and people

Rethink Priorities is a research organization that conducts research to inform policymakers and major foundations about how to best help people and nonhuman animals in both the present and the long-term future. Our work informs key stakeholders of the effective altruism community regarding decisions around funding, interventions, and research time allocation worth millions of dollars every year.

This year we hired 14 new people. We now have a staff of 28 people, corresponding to 24.75 full-time equivalents (with 19.75 FTE focused on research and 5 FTE on operations). We’ll have spent about $2.1M USD in 2021.

Our 2021 impact

In 2021, we achieved some tangible impact from our work, such as:

Improving animal welfare strategies in the European Union (moving several million dollars to what we believe are more effective approaches).

Setting up an ambitious project to study the capacity for welfare of different species.

Starting to staff up a team to address AI Governance and Strategy.

Investigating lead reduction as a possible intervention competitive with GiveWell’s top charities.

Running an intern program that successfully got new researchers involved in effective altruism and led to some interns getting permanent jobs in organizations like the Centre for Effective Altruism, Founders Pledge, and here at Rethink Priorities.

Our plans for 2022

Among other projects in 2022, we’re especially excited to:

Begin work with our newly expanded AI Governance and Strategy team and Global Health and Development team, both of which are directly focused on identifying high-impact giving opportunities.

Build a larger longtermist research team to explore longtermist work and interventions more broadly.

Tentatively conclude our intensive work on interspecies comparisons of moral weight, which could help us better prioritize across many cause areas.

Help solve the funding overhang in EA and unlock tons of impact by identifying interventions across cause areas that can take lots of money while still meeting a high bar for cost-effectiveness.

Our funding goals

If better funded, we would be able to do more high-quality work and employ more talented researchers than we otherwise would.

Currently, our goal is to raise $5,435,000 by the end of 2022. This consists of gaps of:

$2,230,000 for animal-focused research

$1,140,000 for longtermism research

$1,275,000 for EA movement research and surveying

$520,000 for global health and development research

However, we believe that if we were maximally ambitious and expanded as much as is feasible, we could effectively spend the funds if we raised up to $12,900,000 in 2022.

If you’d like to support our work, you can donate to us as part of Facebook’s donation matching on Giving Tuesday on November 30, or donate directly to us here. We do accept and track restricted funds by cause area if that is of interest. If you have questions about tax-deductibility in your country or are interested in making a major gift, please contact our Director of Development Janique Behman.

Ask us more

We’re running an AMA on the EA Forum this Friday, November 19. Ask us any questions you may have!

Our path to impact

Rethink Priorities achieves impact by improving the decisions made by grantmakers and on-the-ground organizations, multiplying the impact of their work. The effective altruism movement currently allocates hundreds of millions of dollars and millions of hours of work every year — we aim to improve that allocation.

We work independently to uncover new insights while also collaborating with existing groups and funders to ensure the most effective actions are taken based on rigorous research.

Our organiza...