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 To Get Into Independent Research On Alignment/Agency, published by johnswentworth on November 19, 2021 on The AI Alignment Forum.

I’m an independent researcher working on AI alignment and the theory of agency. I’m 29 years old, will make about $90k this year, and set my own research agenda. I deal with basically zero academic bullshit - my grant applications each take about one day’s attention to write (and decisions typically come back in ~1 month), and I publish the bulk of my work right here on LessWrong/AF. Best of all, I work on some really cool technical problems which I expect are central to the future of humanity.

If your reaction to that is “Where can I sign up?”, then this post is for you.

Background Models

Independence

First things first: the “independent” part of “independent research” means self-employment, and everything that goes with it. It means the onus is on you to figure out what to do, how to provide value, what to prioritize, and what to aim for. In practice, it also usually means “independent” in a broader sense: you won’t have a standard template or agenda to follow. If you go down this path, assume that you will need to chart your own course - in particular, your own research agenda.

For the sort of person this post is aimed at, that will be a very big upside, not a downside.

Disclaimer: there are ways to get into alignment research which don’t involve quite so much figuring-it-all-out-on-your-own. Some people receive mentorship from existing researchers. Some people go work for alignment research organizations. Either of those paths can involve “independent research” in the sense that you are technically self-employed, but those paths aren’t “independent” in the broader sense of the word, and they’re not the main topic of this post.

Preparadigmicity

As a field, the study of alignment and agency is especially well-suited to independent research, because they center around problems we don’t understand. It’s not just that we don’t have the answers; we don’t even have the right frames for thinking about the problems. Agency is an area where we are fundamentally confused. AI alignment is largely a problem which hasn’t happened yet, on technology which hasn’t been invented yet, which we nonetheless want to solve in advance. Figuring out the right frames - the right paradigm - is itself a central part of the job.

The field needs people who are going to come up with new frames/approaches/models/paradigms/etc, because we’re pretty sure the current frames/approaches/models/paradigms/etc aren’t enough. Thus the great fit for independent research: as an independent researcher, you’re not beholden to some existing agenda based on existing frames. Coming up with your own idea of what the key problems are, how to frame them, what tools to apply. that sort of thing is exactly what we need, and it requires people who aren’t committed to the strategies of existing senior researchers and organizations. It requires people who have an independent high-level understanding of the field and different angles of looking at, and can pick out the key problems and paths from that perspective.

Again, for the sort of person this post is aimed at, that will be a very big upside.

. but it comes with some trade-offs. As a historical example of preparadigmatic research, here’s Kuhn talking about optics before Newton:

Being able to take no common body of belief for granted, each writer on physical optics felt forced to build his field anew from its foundations. In doing so, his choice of supporting experiment and observation was relatively free, for there was no standard set of methods or of phenomena that every optical writer felt forced to employ and explain. Under these circumstances, the dialogue of the resulting books was often directed as much to the members of other...

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 To Get Into Independent Research On Alignment/Agency, published by johnswentworth on November 19, 2021 on The AI Alignment Forum.

I’m an independent researcher working on AI alignment and the theory of agency. I’m 29 years old, will make about $90k this year, and set my own research agenda. I deal with basically zero academic bullshit - my grant applications each take about one day’s attention to write (and decisions typically come back in ~1 month), and I publish the bulk of my work right here on LessWrong/AF. Best of all, I work on some really cool technical problems which I expect are central to the future of humanity.

If your reaction to that is “Where can I sign up?”, then this post is for you.

Background Models

Independence

First things first: the “independent” part of “independent research” means self-employment, and everything that goes with it. It means the onus is on you to figure out what to do, how to provide value, what to prioritize, and what to aim for. In practice, it also usually means “independent” in a broader sense: you won’t have a standard template or agenda to follow. If you go down this path, assume that you will need to chart your own course - in particular, your own research agenda.

For the sort of person this post is aimed at, that will be a very big upside, not a downside.

Disclaimer: there are ways to get into alignment research which don’t involve quite so much figuring-it-all-out-on-your-own. Some people receive mentorship from existing researchers. Some people go work for alignment research organizations. Either of those paths can involve “independent research” in the sense that you are technically self-employed, but those paths aren’t “independent” in the broader sense of the word, and they’re not the main topic of this post.

Preparadigmicity

As a field, the study of alignment and agency is especially well-suited to independent research, because they center around problems we don’t understand. It’s not just that we don’t have the answers; we don’t even have the right frames for thinking about the problems. Agency is an area where we are fundamentally confused. AI alignment is largely a problem which hasn’t happened yet, on technology which hasn’t been invented yet, which we nonetheless want to solve in advance. Figuring out the right frames - the right paradigm - is itself a central part of the job.

The field needs people who are going to come up with new frames/approaches/models/paradigms/etc, because we’re pretty sure the current frames/approaches/models/paradigms/etc aren’t enough. Thus the great fit for independent research: as an independent researcher, you’re not beholden to some existing agenda based on existing frames. Coming up with your own idea of what the key problems are, how to frame them, what tools to apply. that sort of thing is exactly what we need, and it requires people who aren’t committed to the strategies of existing senior researchers and organizations. It requires people who have an independent high-level understanding of the field and different angles of looking at, and can pick out the key problems and paths from that perspective.

Again, for the sort of person this post is aimed at, that will be a very big upside.

. but it comes with some trade-offs. As a historical example of preparadigmatic research, here’s Kuhn talking about optics before Newton:

Being able to take no common body of belief for granted, each writer on physical optics felt forced to build his field anew from its foundations. In doing so, his choice of supporting experiment and observation was relatively free, for there was no standard set of methods or of phenomena that every optical writer felt forced to employ and explain. Under these circumstances, the dialogue of the resulting books was often directed as much to the members of other...

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 11/18: Paxlovid Remains Illegal, published by Zvi on November 18, 2021 on LessWrong.

I had already written most of this week’s post, and also spun out a way-too-long way-too-deep dive on what happened with a retracted FLCCC paper that that I was planning to publish today but I can’t yet because the rabbit hole is still getting deeper, when Scott Alexander posted Ivermectin: Much More Than You Wanted To Know.

My reaction to that was a deep sense of gratitude, as I was worried I’d have to write something because someone had to and no one else would. Then Scott Alexander wrote it. So now I didn’t have to. I don’t agree with every decision he made, I definitely ‘have thoughts’ and I’ll probably end up writing about that, but it can wait for another week.

There’s a variety of other stuff going on as well, including a rise in cases that is unlikely to be going away any time soon, none of it terribly surprising. Paxlovid remains illegal and somehow this may not be easy, various mandates were imposed, children noticed it was nice not to wear masks, and those against vaccination have their latest crazy nonsense where they think they can ‘detox’ and undo the vaccination shot. What a time to be alive.

Executive Summary

Ivermectin: Much More Than You Wanted To Know.

Rise in cases is accelerating. Winter is coming.

Paxlovid remains illegal.

Let’s run the numbers.

The Numbers

Predictions

Prediction from last week (accidentally made on Monday 11/15, so it doesn’t count for prediction scoring purposes, but I didn’t get meaningful new data before then and it still attempts to represent my expectations at the time): 507k cases (+8%) and 7,690 deaths (-5%)

Results: 554k cases (+18%) and 7,427 deaths (-8%).

Prediction for next week: 654k cases (+18%) and 7,500 deaths (+1%).

Prediction for two weeks from now: We’ll see less deaths than normal due to the Thanksgiving holiday messing up reporting, and the counterfactual Zvi that didn’t put in this note likely would have forgotten to adjust for it.

Winter is coming. Boosters and child vaccinations will help a nonzero amount but they’re not sufficiently powerful to put a stop to this, and there’s very little appetite or tolerance for trying to shut things down again. There is no reason to think cases are going to stop rising any time soon, and having seen the acceleration this past week there’s a chance that the rise will be even larger in percentage terms over the next month. That means cases are likely to more than double before we peak again and it would be unsurprising if our case numbers peaked similarly to where they peaked a year ago.

Deaths

Deaths continued to decline because they lag cases. With cases rising again, this trend should reverse shortly, and death counts will begin rising again within two weeks. The only question is when, as there’s no reason to think the IFR/CFR will change much until they approve Paxlovid or (to a lesser and less certain but still probably noticeable extent) if we got more widespread use of Fluvoxamine than we currently have.

Cases

When something similar happened last year I was terrified. This year, I notice almost all of my worry is about the reaction to the coming wave. I don’t want there to be several times as much Covid-19 running around, I’m sad that more people are going to die, I’m sad about those who for medical reasons can’t get boosters, and I’m mildly worried about hospital overloads, but mostly I am worried about all the Covid-19 prevention that will result from this because almost everyone has the opportunity to get booster shots, with a booster protection is robust, and during the worst part of the wave we should have Paxlovid available, although we might still be short on supply.

It is an interesting time, in terms of intertemporal substitution. In the short term, risk is going to...

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 11/18: Paxlovid Remains Illegal, published by Zvi on November 18, 2021 on LessWrong.

I had already written most of this week’s post, and also spun out a way-too-long way-too-deep dive on what happened with a retracted FLCCC paper that that I was planning to publish today but I can’t yet because the rabbit hole is still getting deeper, when Scott Alexander posted Ivermectin: Much More Than You Wanted To Know.

My reaction to that was a deep sense of gratitude, as I was worried I’d have to write something because someone had to and no one else would. Then Scott Alexander wrote it. So now I didn’t have to. I don’t agree with every decision he made, I definitely ‘have thoughts’ and I’ll probably end up writing about that, but it can wait for another week.

There’s a variety of other stuff going on as well, including a rise in cases that is unlikely to be going away any time soon, none of it terribly surprising. Paxlovid remains illegal and somehow this may not be easy, various mandates were imposed, children noticed it was nice not to wear masks, and those against vaccination have their latest crazy nonsense where they think they can ‘detox’ and undo the vaccination shot. What a time to be alive.

Executive Summary

Ivermectin: Much More Than You Wanted To Know.

Rise in cases is accelerating. Winter is coming.

Paxlovid remains illegal.

Let’s run the numbers.

The Numbers

Predictions

Prediction from last week (accidentally made on Monday 11/15, so it doesn’t count for prediction scoring purposes, but I didn’t get meaningful new data before then and it still attempts to represent my expectations at the time): 507k cases (+8%) and 7,690 deaths (-5%)

Results: 554k cases (+18%) and 7,427 deaths (-8%).

Prediction for next week: 654k cases (+18%) and 7,500 deaths (+1%).

Prediction for two weeks from now: We’ll see less deaths than normal due to the Thanksgiving holiday messing up reporting, and the counterfactual Zvi that didn’t put in this note likely would have forgotten to adjust for it.

Winter is coming. Boosters and child vaccinations will help a nonzero amount but they’re not sufficiently powerful to put a stop to this, and there’s very little appetite or tolerance for trying to shut things down again. There is no reason to think cases are going to stop rising any time soon, and having seen the acceleration this past week there’s a chance that the rise will be even larger in percentage terms over the next month. That means cases are likely to more than double before we peak again and it would be unsurprising if our case numbers peaked similarly to where they peaked a year ago.

Deaths

Deaths continued to decline because they lag cases. With cases rising again, this trend should reverse shortly, and death counts will begin rising again within two weeks. The only question is when, as there’s no reason to think the IFR/CFR will change much until they approve Paxlovid or (to a lesser and less certain but still probably noticeable extent) if we got more widespread use of Fluvoxamine than we currently have.

Cases

When something similar happened last year I was terrified. This year, I notice almost all of my worry is about the reaction to the coming wave. I don’t want there to be several times as much Covid-19 running around, I’m sad that more people are going to die, I’m sad about those who for medical reasons can’t get boosters, and I’m mildly worried about hospital overloads, but mostly I am worried about all the Covid-19 prevention that will result from this because almost everyone has the opportunity to get booster shots, with a booster protection is robust, and during the worst part of the wave we should have Paxlovid available, although we might still be short on supply.

It is an interesting time, in terms of intertemporal substitution. In the short term, risk is going to...

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: Competition for "Fortified Essays" on nuclear risk, published by MichaelA on November 17, 2021 on The Effective Altruism Forum.

This post announces an initiative that Metaculus and I have put together and that I'd be excited for members of the EA community to participate in. The text is copied (with slight modifications) from a page Metaculus and I co-wrote. To participate, follow that link.

Prize Pool: $1,000.00

End Date: Dec. 6, 2021

To begin drafting your first essay click this button. You will be able to save and edit your piece until you officially submit.

About

In this essay contest, we open the floor for investigations of long-term nuclear risks and what might be done to reduce them.

As of November 12, 2021, through the Nuclear Risk Forecasting Tournament and Nuclear Risk Horizons Project, 151 Metaculus users have provided a total of 3503 forecasts on 113 questions. These will be used by Rethink Priorities and other stakeholders to guide decisions about funding, research, and policy, hopefully helping to secure and improve the long-term future of humanity.

But the aggregate forecasts are in themselves silent about what mechanisms are driving the probabilities up and down, what actions should be taken to affect the probabilities or in light of the probabilities, and what lines of reasoning or sources may be worth considering. Forecasters sometimes address those topics in comments, and this is sometimes highly valuable, but it’s not directly incentivized, it’s not common, and it rarely goes into substantial depth.

This Fortified Essay competition is intended to fill that void - to leverage and demonstrate forecasters’ knowledge and intellectual labour in ways that the forecasting questions themselves don’t. We seek essays that reference long-range nuclear risk forecasting questions (and, optionally, other forecasting questions) while putting them into context and discussing mechanisms, recommendations, and reasoning. These essays can then be used by stakeholders to inform their decisions, and to help them better understand the nuclear risk forecasts themselves and how much weight to put on them.

Three additional goals of this competition are to:

Identify and reward insightful thinkers and forecasters on the topic of nuclear risk

Provide a probabilistically and quantitatively minded arena for discussion and exchange of ideas regarding nuclear risk

Provide exemplars of a style of writing that’s quite distinct from most writing on nuclear risk (and other topics), in that that writing, despite many strengths, often leaves its predictions about the future too unclear to act on or falsify

Evaluation criteria

An expert panel of judges will be instructed to rate the entries based on the extent to which the essay:

clearly and coherently addresses one of the essay prompts (see below)

synthesizes at least 3 of the forecasts made in the Nuclear Risk Horizons Project and draws out their key implications. Referencing more forecasts is only desirable if it meaningfully contributes to the analysis

demonstrates a solid understanding of both forecasting and relevant aspects of nuclear risk

provides relevant evidence and reasoning and has high reasoning transparency

is well written, so that it is comprehensible, enjoyable to read, and accessible to a well-educated but non-specialist audience (aiming in the range between the level of Scientific American and a review article in Science or Nature)

Essay Topics

Each essay must address one of the topics below:

What is/are the most likely way(s) a nuclear war involving hundreds of detonations could end up occurring by 2050? That is, how might the war start, and why would it either (a) start out involving hundreds of detonations or (b) escalate to that point? Justify the view that this/these way(s) are the most likely.

If there’s a nucle...

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: Competition for "Fortified Essays" on nuclear risk, published by MichaelA on November 17, 2021 on The Effective Altruism Forum.

This post announces an initiative that Metaculus and I have put together and that I'd be excited for members of the EA community to participate in. The text is copied (with slight modifications) from a page Metaculus and I co-wrote. To participate, follow that link.

Prize Pool: $1,000.00

End Date: Dec. 6, 2021

To begin drafting your first essay click this button. You will be able to save and edit your piece until you officially submit.

About

In this essay contest, we open the floor for investigations of long-term nuclear risks and what might be done to reduce them.

As of November 12, 2021, through the Nuclear Risk Forecasting Tournament and Nuclear Risk Horizons Project, 151 Metaculus users have provided a total of 3503 forecasts on 113 questions. These will be used by Rethink Priorities and other stakeholders to guide decisions about funding, research, and policy, hopefully helping to secure and improve the long-term future of humanity.

But the aggregate forecasts are in themselves silent about what mechanisms are driving the probabilities up and down, what actions should be taken to affect the probabilities or in light of the probabilities, and what lines of reasoning or sources may be worth considering. Forecasters sometimes address those topics in comments, and this is sometimes highly valuable, but it’s not directly incentivized, it’s not common, and it rarely goes into substantial depth.

This Fortified Essay competition is intended to fill that void - to leverage and demonstrate forecasters’ knowledge and intellectual labour in ways that the forecasting questions themselves don’t. We seek essays that reference long-range nuclear risk forecasting questions (and, optionally, other forecasting questions) while putting them into context and discussing mechanisms, recommendations, and reasoning. These essays can then be used by stakeholders to inform their decisions, and to help them better understand the nuclear risk forecasts themselves and how much weight to put on them.

Three additional goals of this competition are to:

Identify and reward insightful thinkers and forecasters on the topic of nuclear risk

Provide a probabilistically and quantitatively minded arena for discussion and exchange of ideas regarding nuclear risk

Provide exemplars of a style of writing that’s quite distinct from most writing on nuclear risk (and other topics), in that that writing, despite many strengths, often leaves its predictions about the future too unclear to act on or falsify

Evaluation criteria

An expert panel of judges will be instructed to rate the entries based on the extent to which the essay:

clearly and coherently addresses one of the essay prompts (see below)

synthesizes at least 3 of the forecasts made in the Nuclear Risk Horizons Project and draws out their key implications. Referencing more forecasts is only desirable if it meaningfully contributes to the analysis

demonstrates a solid understanding of both forecasting and relevant aspects of nuclear risk

provides relevant evidence and reasoning and has high reasoning transparency

is well written, so that it is comprehensible, enjoyable to read, and accessible to a well-educated but non-specialist audience (aiming in the range between the level of Scientific American and a review article in Science or Nature)

Essay Topics

Each essay must address one of the topics below:

What is/are the most likely way(s) a nuclear war involving hundreds of detonations could end up occurring by 2050? That is, how might the war start, and why would it either (a) start out involving hundreds of detonations or (b) escalate to that point? Justify the view that this/these way(s) are the most likely.

If there’s a nucle...

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: Sasha Chapin on bad social norms in rationality/EA, published by Kaj Sotala on November 17, 2021 on LessWrong.

So, I’ve noticed that a significant number of my friends in the Rationalist and Effective Altruist communities seem to stumble into pits of despair, generally when they structure their lives too rigidly around the in-group’s principles. Like, the Rationalists become miserable by trying to govern their entire lives through nothing but rationality, and the EAs feel bad by holding themselves to an impossible standard of ethics. [...]

I’ve tried to figure out why this happens, and I’ve tried to write about it several times, batting around complex notions that, as I examine them, reveal themselves to be fake models that make me sound smart but don’t explain anything.

But today I realized that it’s generally much simpler than I thought previously. Most of it is just toxic social norms. These groups develop toxic social norms. In the Rationalist community, one toxic norm is something like, “you must reject beliefs that you can’t justify, sentiments that don’t seem rational, and woo things.” In the EA community, one toxic norm is something like, “don’t ever indulge in Epicurean style, and never, ever stop thinking about your impact on the world.”

Generally, toxic social norms don’t develop intentionally, nobody wants them to happen, they’re not written down, and nobody enforces them explicitly. (The intentional development of toxic social norms is otherwise known as founding a cult.) What happens is that there are positive social norms, like, “talking about epistemics and being curious about beliefs is cool,” or “try to be intentional about the positive impact you can have on the world.” These norms are great! But then, the group acts like a group, which is to say, people confer status depending on level of apparent adherence to values. This leads insecure people who completely depend on the group to over-identify with the set of values, to the extent that even slightly contrary actions become forbidden. Not forbidden in the like “we’ll arrest you” way, but in the like “everyone in the room immediately looks at you like you’re being rude if you talk about spirituality” way.

And then the second, more sinister stage occurs—the point at which these toxic norms are internalized such that they apply to you when you’re in a room alone. As Wittgenstein noted, it’s hard to tell where aesthetics end and ethics begin; it can start to feel unethical, like, dirty, to perform behaviors your peers would think distasteful. Toxic norms eventually pervade completely, to the point where you don’t even want to think bad thoughts.

Sometimes—often—these forbidden thoughts/actions aren’t even contrary to the explicit values. They just don’t fit in with the implied group aesthetic, which is often a much stricter, more menacing guideline, all the more so because it’s a collective unwritten fiction. “Rationality is cool” becomes “rationality is the best framework” becomes “Rationalist and Rationalist-flavored stuff is a better use of your time than anything else” becomes “it’s uncool if you want to spend a lot of time doing stuff that has nothing to do with testable beliefs, or our favorite issues.” This is all unintentional and implicit. No Rationalist has ever said, to my knowledge, that you shouldn’t write poetry, but a few Rationalists have told me that they feel like they shouldn’t make weird art because it’s dumb and un-Rationalist to do so—they feel they ought to produce useful thoughts instead, even though their hearts are trying to steer them somewhere else.

I point out to them that Scott Alexander wrote a fantasy novel for fun, but somehow this isn’t persuasive enough.

Here, I should probably stop and define toxic norms. I think a toxic norm is any rule where following it makes yo...

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: Sasha Chapin on bad social norms in rationality/EA, published by Kaj Sotala on November 17, 2021 on LessWrong.

So, I’ve noticed that a significant number of my friends in the Rationalist and Effective Altruist communities seem to stumble into pits of despair, generally when they structure their lives too rigidly around the in-group’s principles. Like, the Rationalists become miserable by trying to govern their entire lives through nothing but rationality, and the EAs feel bad by holding themselves to an impossible standard of ethics. [...]

I’ve tried to figure out why this happens, and I’ve tried to write about it several times, batting around complex notions that, as I examine them, reveal themselves to be fake models that make me sound smart but don’t explain anything.

But today I realized that it’s generally much simpler than I thought previously. Most of it is just toxic social norms. These groups develop toxic social norms. In the Rationalist community, one toxic norm is something like, “you must reject beliefs that you can’t justify, sentiments that don’t seem rational, and woo things.” In the EA community, one toxic norm is something like, “don’t ever indulge in Epicurean style, and never, ever stop thinking about your impact on the world.”

Generally, toxic social norms don’t develop intentionally, nobody wants them to happen, they’re not written down, and nobody enforces them explicitly. (The intentional development of toxic social norms is otherwise known as founding a cult.) What happens is that there are positive social norms, like, “talking about epistemics and being curious about beliefs is cool,” or “try to be intentional about the positive impact you can have on the world.” These norms are great! But then, the group acts like a group, which is to say, people confer status depending on level of apparent adherence to values. This leads insecure people who completely depend on the group to over-identify with the set of values, to the extent that even slightly contrary actions become forbidden. Not forbidden in the like “we’ll arrest you” way, but in the like “everyone in the room immediately looks at you like you’re being rude if you talk about spirituality” way.

And then the second, more sinister stage occurs—the point at which these toxic norms are internalized such that they apply to you when you’re in a room alone. As Wittgenstein noted, it’s hard to tell where aesthetics end and ethics begin; it can start to feel unethical, like, dirty, to perform behaviors your peers would think distasteful. Toxic norms eventually pervade completely, to the point where you don’t even want to think bad thoughts.

Sometimes—often—these forbidden thoughts/actions aren’t even contrary to the explicit values. They just don’t fit in with the implied group aesthetic, which is often a much stricter, more menacing guideline, all the more so because it’s a collective unwritten fiction. “Rationality is cool” becomes “rationality is the best framework” becomes “Rationalist and Rationalist-flavored stuff is a better use of your time than anything else” becomes “it’s uncool if you want to spend a lot of time doing stuff that has nothing to do with testable beliefs, or our favorite issues.” This is all unintentional and implicit. No Rationalist has ever said, to my knowledge, that you shouldn’t write poetry, but a few Rationalists have told me that they feel like they shouldn’t make weird art because it’s dumb and un-Rationalist to do so—they feel they ought to produce useful thoughts instead, even though their hearts are trying to steer them somewhere else.

I point out to them that Scott Alexander wrote a fantasy novel for fun, but somehow this isn’t persuasive enough.

Here, I should probably stop and define toxic norms. I think a toxic norm is any rule where following it makes yo...

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: Some readings & notes on how to do high-quality, efficient research, published by MichaelA on November 17, 2021 on The Effective Altruism Forum.

Here is the doc version of this, which has some additional comments and which I'll continue adding to occasionally without bothering to edit this post.

You might also be interested in my similar collections on how to do high-impact research, write/communicate well, manage/mentor researchers, and/or get useful input from busy people.

I made all of these things in a personal capacity, and they don't necessarily represent anyone else's views.

How to use this doc

I suggest reading everything in bold

Other than that, you can skim or skip around as much as you want

Feel very free to add comments/suggestions, including regarding which resources/tips seem particularly useful to you, which seem not useful, and what other things it might be worth adding

I put the purpose and epistemic status of this doc at the bottom

What this doc focuses on

How to increase the accuracy, informativeness, insightfulness, etc. of your research outputs

Not taking into account how important the original scope/questions/goals were, or whether anyone actually consumes and understands your outputs

How to get research done in fewer hours of work and less calendar time

To some extent, how to (rapidly) “get up to speed” on a topic/field

Some things this doc doesn’t focus on are listed in the following footnote.[1]

Readings - misc

Scholarship & Learning - tag on LessWrong

I haven’t read most of the posts with that tag, though I vaguely yearn to do so

Some posts I especially recommend:

The Neglected Virtue of Scholarship

Scholarship: How to Do It Efficiently

Costs and Benefits of Scholarship

The Best Textbooks on Every Subject

I didn’t personally find this very useful since most of the recommended textbooks were on subjects that aren’t very relevant to my work and weren’t available as audiobooks (I find it easier to make time to listen to books than to read them).

But this was somewhat useful even for me, and would likely be very useful for people interested in more of the relevant fields and people who don’t have a preference for audiobooks over regular books.

Literature Review For Academic Outsiders: What, How, and Why

Research methods - tag on EA Forum

Some posts I liked:

How to succeed as an early-stage researcher: the “lean startup” approach

How to generate research proposals - EA Forum

Research advice from Carl Shulman (doc created by Claire Zabel in 2016, according to this page)

Effective Thesis's collection of Resources

Advice for New Researchers - A collaborative EA doc

Resources to learn how to do research

(and the pages linked to under “OUR RESEARCH PROCESS”)

Research Skills Toolbox - CERI

Collection of Resources on Doing Impactful Research Well - SERI

The doc version of Readings and notes on how to do high-impact research, which contains things that were somewhat out of scope there and are arguably more in-scope here

Semi-relevant: Notes on EA-related research, writing, testing fit, learning, and the Forum - EA Forum

Maybe Independent research - EA Forum

Maybe some stuff in Readings and notes on how to do high-impact research

Maybe Action: Write a literature review or meta-analysis

Notes - misc

I suggest implementing Luke M’s advice from The Neglected Virtue of Scholarship and Scholarship: How to Do It Efficiently

But there’s one (fortunate!) way Luke’s advice hasn’t aged well:

The EA community now has far more money (I think tens of billions more) than it did around 2011, when Luke wrote those articles.

We’re now much more constrained by people with the necessary skills and motivations to do various important types of work than by funding.

So in most cases, if someone is doing valuable EA-aligned work or is on track to do so, I would be quite sad if ...

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: Some readings & notes on how to do high-quality, efficient research, published by MichaelA on November 17, 2021 on The Effective Altruism Forum.

Here is the doc version of this, which has some additional comments and which I'll continue adding to occasionally without bothering to edit this post.

You might also be interested in my similar collections on how to do high-impact research, write/communicate well, manage/mentor researchers, and/or get useful input from busy people.

I made all of these things in a personal capacity, and they don't necessarily represent anyone else's views.

How to use this doc

I suggest reading everything in bold

Other than that, you can skim or skip around as much as you want

Feel very free to add comments/suggestions, including regarding which resources/tips seem particularly useful to you, which seem not useful, and what other things it might be worth adding

I put the purpose and epistemic status of this doc at the bottom

What this doc focuses on

How to increase the accuracy, informativeness, insightfulness, etc. of your research outputs

Not taking into account how important the original scope/questions/goals were, or whether anyone actually consumes and understands your outputs

How to get research done in fewer hours of work and less calendar time

To some extent, how to (rapidly) “get up to speed” on a topic/field

Some things this doc doesn’t focus on are listed in the following footnote.[1]

Readings - misc

Scholarship & Learning - tag on LessWrong

I haven’t read most of the posts with that tag, though I vaguely yearn to do so

Some posts I especially recommend:

The Neglected Virtue of Scholarship

Scholarship: How to Do It Efficiently

Costs and Benefits of Scholarship

The Best Textbooks on Every Subject

I didn’t personally find this very useful since most of the recommended textbooks were on subjects that aren’t very relevant to my work and weren’t available as audiobooks (I find it easier to make time to listen to books than to read them).

But this was somewhat useful even for me, and would likely be very useful for people interested in more of the relevant fields and people who don’t have a preference for audiobooks over regular books.

Literature Review For Academic Outsiders: What, How, and Why

Research methods - tag on EA Forum

Some posts I liked:

How to succeed as an early-stage researcher: the “lean startup” approach

How to generate research proposals - EA Forum

Research advice from Carl Shulman (doc created by Claire Zabel in 2016, according to this page)

Effective Thesis's collection of Resources

Advice for New Researchers - A collaborative EA doc

Resources to learn how to do research

(and the pages linked to under “OUR RESEARCH PROCESS”)

Research Skills Toolbox - CERI

Collection of Resources on Doing Impactful Research Well - SERI

The doc version of Readings and notes on how to do high-impact research, which contains things that were somewhat out of scope there and are arguably more in-scope here

Semi-relevant: Notes on EA-related research, writing, testing fit, learning, and the Forum - EA Forum

Maybe Independent research - EA Forum

Maybe some stuff in Readings and notes on how to do high-impact research

Maybe Action: Write a literature review or meta-analysis

Notes - misc

I suggest implementing Luke M’s advice from The Neglected Virtue of Scholarship and Scholarship: How to Do It Efficiently

But there’s one (fortunate!) way Luke’s advice hasn’t aged well:

The EA community now has far more money (I think tens of billions more) than it did around 2011, when Luke wrote those articles.

We’re now much more constrained by people with the necessary skills and motivations to do various important types of work than by funding.

So in most cases, if someone is doing valuable EA-aligned work or is on track to do so, I would be quite sad if ...