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 From a Pledger, published by Justis on April 30, 2022 on The Effective Altruism Forum.

This post is a response to Jeff's post here, and the genre of post it represents. I think Jeff's post is valid and valuable, and that thinking through this sort of thing is a good idea. I also think that necessarily, personal testimonies are going to be more common from more engaged EAs. So I'd like to give the perspective of a GWWC pledge taker more toward the periphery. N=1, so have your salt shaker ready!

EA status is unimportant at a certain distance

I've done freelance editing work for EAs and EA-related organizations for over five years, including (at various times) CEA, LessWrong, AI Impacts, BERI, PIE, and probably others I'm forgetting. There have been times - the longest was a little under a year - when this editing work was my primary source of income. I read a lot of EA content and occasionally wade into the discourse. So among non-inner-ring EAs, I think I'm probably unusually engaged by most metrics, perhaps among the most engaged 10% in the reference class of people who might skim the EA newsletter now and then.

All this being said, my relative status as an EA is just not very important to me at all. I took the Giving What We Can pledge several years ago. I've been donating 10% of my income to the Against Malaria Foundation since then. I care approximately 0 if EA bigwigs think I'm a bit dim for this decision, or if they think "ok well, a direct worker is worth about 30 Justis equivalents on the margin." My grandparents were religious and tithed; it feels nice to "do my part" not in an abstract EA community cred sense but in a vague, "being a morally decent person" sense, and no amount of focus on direct work by people across the country/world is likely to make me feel inadequate or rethink this.

Really! I will maintain my pledge with a similar level of pride and joy no matter what the official recommendation to current Yale students in EA student groups is (not to undermine them - they are very important, just approximately irrelevant to my life). I am in Florida. Most of my friends work at restaurants or for the state government. Sure, it feels nice when people across the country want to include me or rank me well, but it's not in a crucial spot on my hierarchy of needs. I suspect the same is true for most pledge-takers.

It's possible to regard EA roles as desirable without regarding them as morally necessary

I've applied for a few jobs in EA over the years. I didn't get them. This was painful. In one case I was doing 25+ hours a week freelance work for an org for several months, it went really well, they put up a job posting with precisely my current duties and invited me to apply, then hired someone else. This was very painful, and strongly discouraged me from applying to full time EA roles in the future.

However, at no point did I think "oh no, now I can't have all the impact I might have had counterfactually, so I'm a bad/worthless person."

Here I think it's time for another interlude: I've got moderate OCD and a lot of that manifests as scrupulosity. My most common negative emotion is guilt. I'm confident that I worry about being somehow ineffably "bad" significantly more than the average human being. I read Peter Singer in high school and it rocked my world - I got on the phone with Oxfam to donate something, anything, but realized I didn't actually have any money I'd earned myself yet and hung up. I'm confident that I am among the most vulnerable people to feeling distress from the notion that I've not lived up to an ethical obligation. But no, none of my EA-role-failure pain has really been in that direction. I am giving 10% of my income. This is much more than basically anyone I know, including virtually all my friends and all my family. I will proba...

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: Big List of Cause Candidates: January 2021–March 2022 update, published by Leo on April 30, 2022 on The Effective Altruism Forum.

Since Big List of Cause Candidates was posted, there have been many other posts proposing new neglected or less-discussed cause areas. This post aims to record and organize any such causes. The original post will be shortly updated, but I thought it could be useful to list them separately.

I didn't focus on judging the ideas. I just tried to add as many as I could find, giving short summaries for potential readers. I thought it was exciting that some of the proposed causes are currently being funded by the Future Fund, for example “biological weapons shelters” and “better PPE”, among other biosecurity projects.

Notes: 1) Being a continuation of the original post, this one follows the same categorization. 2) The number next to the author's name is the post's current karma. 3) This post focuses on higher level “cause areas”, not on lower-level “interventions”. For instance, Mosul Dam Could Kill 1 Million Iraqis, while high-quality, is not included in this compilation.

Animal Welfare and Suffering

1. Animal-Free Proteins

Animal-free proteins: A bright outlook and a to-do list (BCG report) (@FGH) +38

The report describes what needs to happen to get to 11%, and further to 22% of meat, seafood, eggs and dairy eaten globally every day. Current technology must be refined and scaled, and in some areas, step changes are needed. For instance, optimized protein crops for human consumption need to be bred, and microorganisms as well as animal cells grown on low-cost feedstocks. Regulatory support, such as carbon taxes on meat or subsidies for farmers who are shifting from animal agriculture to alternative proteins, could further boost growth.

2. Herbivorizing Predators

Should we herbivorize predators? (@Stijn) +4

This post puts forth a moral argument with the intention to open discussion about considering herbivorizing predators as a cause area. The author argues that we should start doing scientific research to look for new technologies that would make it possible.

3. Improving plant-based diets

The Case for Rare Chinese Tofus (@George Stiffman) +180

In order to improve vegan alternatives, this post proposes to create new types of plant-based food crossing rare Chinese tofu with traditional western cooking methods. The author analyzes the idea in detail, and answers possible objections.

4. Welfare of Specific Animals

Baboons: Urban wildlife in South Africa - Cape baboons (@ajmfisher) +9

The aim of this post is to catalogue existing methods for managing the population of Cape chacma baboons living in the Cape peninsula, with a focus on welfare impacts for the baboons.

Chickens: [Question] New EA cause area: Breeding really dumb chickens (@Sam Enright) +34

This post poses some questions about the idea of “breeding chickens (and other farm animals) to be less intelligent as a way to reduce the suffering caused by factory farming”.

Cleaner Fish: Cleaner Fish: A Neglected Issue Within A Neglected Issue (@Martine Klock Fleten) +50

Despite poor evidence, the use of cleaner fish is common practice among salmon farmers to control sea lice, leading to their suffering and death. The post proposes to improve cleaner fish welfare, or to put an end to this practice.

Mice and Rats: [Question] Are mice or rats (as pests) a potential area of animal welfare improvement? (@Louis_Dixon) +17

This post hints that the suffering of mice and rats in cities might be a possible cause area. The answers give several clues about how to tackle the issue.

Silkworms: Silk production: global scale and animal welfare issues (@abrahamrowe) +71

The post examines if the suffering of silkworms used in silk production could be an area to be prioritized, but concludes that available resource...

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: The availability bias in job hunting, published by Vaidehi Agarwalla on April 30, 2022 on The Effective Altruism Forum.

TL:DR; EA talent is currently suboptimally allocated, resulting in lower impact for the movement. There are many reasons for this, and in this post I talk about one reason: attention misallocation - and specifically, availability bias around certain careers and paths.

Linguistic note: I will refer to career paths that are given the most attention in EA “highlighted paths”.

The availability bias is “a mental shortcut that relies on immediate examples that come to a given person's mind when evaluating a specific topic, concept, method or decision.” Many factors cause some facts to be recalled more easily than others (e.g. if the fact is recent, frequent, extreme, vivid, negative etc.)

In the EA context, people exhibit the availability bias towards a small subset of very competitive, geography-limited, low absorbency career paths and roles (henceforth highlighted paths). How certain are we that this is actually happening? One proxy for understanding attention is to look at job applications. These are pretty strong signals of intent on the part of community members. Jobs at some EA-aligned organisations are notoriously hard to get and getting rejected from such opportunities comes with costs. The mismatch of attention and opportunities illustrated in the stylized sankey plot below:

How does the availability bias work?

Highlighted paths get disproportionate attention and status from the community, but cannot absorb enough people. The way that these opportunities are perceived and experienced creates the conditions for availability bias. Additionally, there are limited resources and support for other, higher absorbency paths so the lack of information and support around these paths further exacerbates the condition.

It makes highlighted paths...

socially acceptable

EA changes peoples’ perceptions of different opportunities. In particular, it raises the bar for what counts as effective. People consider a smaller set of opportunities to be acceptable and may feel like a given opportunity needs endorsement from the EA community to be acceptable. Because there are so few opportunities that are defined as clearing this bar, they get disproportionate attention.

(socially) desirable

As people get more involved with EA and deepen their engagement, the more likely they are to want direct impact as well. It actually is more desirable to work for an EA organisation because EA jobs provide scarce non-monetary goods such as:

Social status

Meaning-making / life orientation

A sense of having a near-maximal impact

Being part of a value-aligned group

easier to access (not to actually get)

Job opportunities for highlighted paths at existing EA organisations, feel more accessible because they are publicised repeatedly, concrete, and convenient.

Repetition: You are much likely to apply if you hear about a role at a careers fair, on job boards, on social media and the Forum, and by referrals from group organisers and friends.

Concrete: work at an org vs create your own role (e..g in policy this can happen)

Convenient: Some publicity methods like job boards or career fairs make finding some opportunities very convenient. Curated job boards present you with relevant and up-to-date openings, are convenient to search, and appear comprehensive (sometimes listing hundreds of vacancies). Careers fairs give you the ability to network with staff from the represented organisations which both increasing the likelihood of people applying and may appear to (and actually) improve their chances of getting the job. It also makes it easy to anchor on such methods, because the alternative to a job board is to search up lots of organisations yourself, and the alternative to a careers fair is 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: There are currently more than 100 open EA-aligned tech jobs, published by Sebastian Becker on April 30, 2022 on The Effective Altruism Forum.

TL:DR; There are currently more than 100 open EA-aligned tech jobs, including roles such as software engineering, ML engineering, data science, product management, and UI/UX design.Thanks to Vaidehi Agarwalla, Yonatan Cale, Patrick Gruban and David Mears for feedback on drafts of this post.

EA’s tech needs in 2018

Flashback to EAG London 2018: Having just come out of a web development bootcamp, I got the impression that the only use for software engineers within the EA community was earning-to-give. There were very few software engineering jobs at EA organizations, and they didn’t seem to be particularly exciting. As one conference participant said to me: “There are only so many EA orgs for which websites must be built”. Sure, there was AI safety research, but your “standard” web developer does not have the required skill set.

EA’s software needs in 2022

Fast forward 3.5 years, this has changed. Probably partly due to EA growing significantly in funding, more and more tech jobs are advertised every day. The diversity of jobs has also increased, with roles ranging from app and web development, to data science, ML engineering or product management, as well as varying in seniority levels.

Some examples:

Anthropic is looking for a senior software engineer to build large scale ML systems to do AI alignment work.

Momentum is hiring a frontend engineer to help build their donation app.

Our World in Data is searching for a Head of Product and Design.

IDInsight has open positions for junior data scientists and engineers.

At the most recent EAG London, I was talking about this to someone working at a big tech company, and they asked me: “But how many EA-aligned tech jobs are there? Surely not a 100?”, to which I replied without much thinking, “No, 100 sounds about right!”

Are there really more than 100 open EA-aligned tech jobs?

Curious whether I was correct, I’ve compiled this list, and, lo and behold, my estimate seems rather spot on. I could find 115 openings for EA-aligned tech jobs, at least if we interpret the relevant terms as follows.

What I counted as a tech job

I consider the following roles a tech job:

Software engineer/developer

Data scientist

Data engineer

IT admin

Product manager

UI/UX designer

Research engineer

Machine Learning engineer

What I didn't count as a tech job

I did not include AI research jobs, like this one at DeepMind. Under some interpretation of the term “tech”, you might want to add such jobs, but the target audience for this post are people working in “ordinary” industry jobs that don’t require a PhD in Computer Science. I have met many EAs who have a data/software background, but aren’t drawn to AI research work as they aren't qualified. I did, however, include any tech jobs from the above list in organizations working on AI alignment, such as this one at OpenAI.

I excluded internships, as calling these “jobs” seems to be somewhat misleading.

What did I count as EA-aligned?

As EA-aligned, I considered jobs that satisfy any of the following conditions:

Posted on the 80k job board

A job at an organization that is on this list of EA-aligned organizations

[EDIT] To avoid confusion, it is worth pointing out that some of the jobs on the 80k job board are listed there not because they are directly impactful, but because they help people build up career capital to be impactful later. As Niel Bowerman from 80k writes in a comment to this post: "I'm keen for people to take these "career capital" roles so that in the future they can contribute more directly to making the development of powerful AI systems go well." So "EA-aligned" shouldn't be read as "directly impactful towards EA goals". [EDIT 2] Some have argued ...

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: [Linkpost] New multi-modal Deepmind model fusing Chinchilla with images and videos, published by p.b. on April 30, 2022 on LessWrong.

Seems to be flying under the radar so far. Maybe because it looks more like incremental progress at first glance, similar to what, for example, Aleph Alpha has done continuing the Frozen approach.

However, with the (possibly cherry-picked) examples, it looks to me a lot like the image/video/text-GPT-4 many are expecting.

Blogpost here. Paper here.

Thanks for listening. To help us out with The Nonlinear Library or to learn more, please visit nonlinear.org.

Welcome to The Nonlinear Library, where we use Text-to-Speech software to convert the best writing from the Rationalist and EA communities into audio. This is: How do we create a culture of ambition without deteriorating the community’s mental health?, published by Max Görlitz on April 30, 2022 on The Effective Altruism Forum.

Being more ambitious has become a very popular meme within the EA community and “a culture of ambition” was the unofficial motto of the recent EAG London conference.

I understand the argument that we should strive for opportunities which have a small chance of success but huge payoff in impact if they work out. If most of us were to act this way, then the aggregate impact of the whole community will be bigger and what we care about is more overall good in the world, disregarding who brought it about.

Most of the people taking this attitude will fail at what they are trying to achieve, because the things they are working on are very hard. That is an intrinsic property of this “high risk, high reward” style of doing good, but it feels bad to fail.

Here is the idea in the words of SBF on the recent 80k podcast:

“So I think there are really compelling reasons to think that the “optimal strategy” to follow is one that probably fails — but if it doesn’t fail, it’s great. But as a community, what that would imply is this weird thing where you almost celebrate cases where someone completely craps out — where things end up nowhere close to what they could have been — because that’s what the majority of well-played strategies should end with. I don’t think that we recognize that enough as a community, and I think there are lots of specific instances as well where we don’t incentivize that. ” (emphasis mine)

Supporting the argument intellectually is one thing, but coping with failure is another. How can we make it feel more rewarding for everyone who did the right thing and tried the ambitious project, but ultimately didn’t succeed?

Anecdotally, I recently found myself in a career situation where I was trying to build up skills which me and others consider valuable to have more of in the EA community. Unfortunately I started noticing that the situation wasn’t ideal for me and my mental health was getting worse. While making the decision to stop pursuing this particular path, I had thoughts like “Oh man, I really don’t enjoy doing this, but it would sure be useful for me to build up skill X so that in five years I can realize this really ambitious project Y. Therefore maybe I should continue anyways”.

Surely others have been in a similar place. A few data points of people mentioning the bad feeling of failure to the point of burnout (!) are in this recent thread of EAs failing in high risk, high reward projects.

Some people will thrive in an environment with a high risk of failure and the thought of potentially achieving something incredible will be enough motivation for them. Many others will find such an environment difficult and I worry that they will be put off from EA.

In short: How do we create a culture of ambition without deteriorating the community’s mental health?

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: Discussion on Thomas Philippon's paper on TFP growth being linear, published by Arjun Yadav on April 30, 2022 on The Effective Altruism Forum.

Note: This post was written quite quickly and I'm not well versed in this subject matter. Thomas' paper here and Dylan Matthews' excellent write-up on it here.I would love to spark some discussion on this: total factor productivity growth being linear in many developed countries, not exponential, could potentially be very scary.Of course, as Dylan mentioned, TFP has issues. I believe the main critique is that, due to its simplicity, it can sometimes remain the same even after changes in technology and productivity.

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: Is it still hard to get a job in EA? Insights from CEA’s recruitment data, published by Aadil Kara on April 29, 2022 on The Effective Altruism Forum.

How hard is it to get a job in an EA-aligned organisation? CEA can contribute to this conversation by sharing some insights from our recruitment data.

In this post we look at the recruitment process for 12 vacancies, which were recruited for between January 2021 and April 2022.

Summary

The 12 roles were recruited for in two categories: CEA Core roles, which represent the main rounds, and Expressions of Interest (EOIs), which we were open to hiring but not actively focused on.

Total number of applications: CEA Core roles received an average of 54 applications each; EOIs received an average of 53 applications each.

Proportion of applicants invited to the first stage: On average 48% of applicants to CEA Core roles moved forward to the first stage of the evaluation process. 21% of EOIs were advanced to the first stage.

Proportion of candidates who were ultimately hired: For CEA Core roles, we ultimately hired 4.7% of candidates who reached the first stage, and 9.5% of candidates who reached the second stage. There were no successful hires for the EOIs.

Expected time spent on the recruitment process: Of candidates who advanced beyond the initial screen, the time spent on recruitment activities was (excluding the initial application):

for CEA Core roles: 1.9 hours (estimated range: 0.8 to 4.6)

for EOIs: 1.3 hours (estimated range: 0.7 to 3.6)

Comparing recruitment at CEA to non-EA organisations: We compared our Product Manager hiring round to industry benchmark data. We found that the probability of an applicant receiving an offer was similar to industry averages.

Background and data used

The Centre for Effective Altruism (CEA) nearly doubled its headcount in 2021 and continues to make a substantial number of hires each quarter.

Most roles have a multi-stage recruitment process. First, applications undergo an initial screening. Subsequently there may be trial tasks, interviews, and work trials, finally culminating in an offer of employment. The CEA website gives more information about the hiring process.

Since January 2021, CEA has maintained an internal system to manage recruitment. The analysis in this post concerns data extracted from the system in April 2022, which records information for 34 roles.

From this analysis we have excluded about two-thirds of the roles for which information is recorded:

5 roles within the Operations team, which spun out from CEA Core in 2021

8 roles where teams separately tracked the hiring process

3 roles where recruitment was merged into other roles

1 role where, due to the nature of the role, applicants were each recruited in different ways

10 roles where applications are still open or recruitment is ongoing

The remaining 12 positions are listed in this footnote and broadly fit into one of two categories:

CEA Core roles represent the main full recruitment rounds. CEA usually conducts lots of outreach for these roles, and proactively encourages people to apply.

Expressions of Interest (EOIs) represent roles which we are interested in eventually hiring for, but are not actively focused on at present. The website signposts that these roles have a higher bar for progression, and that CEA carries out much less outreach for them, but aims to make applying as easy as possible.

In the analysis below, it is worth noting the following:

Candidates exited the application process for a variety of reasons, and no distinction is made between those reasons; e.g. some were rejected while others chose to withdraw voluntarily (although the number of voluntary withdrawals is relatively small).

Some applicants were fast-tracked through the process and skipped some stages; in these cases the analysis still counts...

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: Salvage Epistemology, published by jimrandomh on April 30, 2022 on LessWrong.

A funny thing happens with woo sometimes, in the rationality community. There's a frame that says: this is a mix of figurative stuff and dumb stuff, let's try to figure out what the figurative stuff is pointing at and salvage it. Let's call this "salvage epistemology". Unambiguous examples include the rationality community's engagement with religions, cold-reading professions like psychics, bodywork, and chaos magic. Ambiguous examples include intensive meditation, Circling, and many uses of psychedelics.

The salvage epistemology frame got locally popular in parts of the rationality community for awhile. And this is a basically fine thing to do, in a context where you have hyper-analytical programmers who are not at risk of buying into the crazy, but who do need a lens that will weaken their perceptual filters around social dynamics, body language, and muscle tension.

But there's a bad thing happens when you have a group that are culturally adjacent to the hyper-analytical programmers, but who aren't that sort of person themselves. They can't, or shouldn't, take for granted that they're not at risk of falling into the crazy. For them, salvage epistemology disarms an important piece of their immune system.

I think salvage epistemology is infohazardous to a subset of people, and we should use it less, disclaim it more, and be careful to notice when it's leading people in over their heads.

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: Covid 4/28/22: Take My Paxlovid, Please, published by Zvi on April 28, 2022 on LessWrong.

Many have made the observation that if you had told people two years ago that we would have a cure (not a vaccine, but a cure!) for Covid-19 that was free, safe and effective, but that no one wanted to take it, and Congress wasn’t willing to fund further purchases, people would not have believed you. And yet here we are.

The White House is trying to fix the situation. I’ll cover the details below.

We haven’t learned much new about China. When we do, it will go into its own post. I did get private supporting evidence that the situation in Shanghai is being characterized roughly correctly.

Executive Summary

Paxlovid is available, no one wants it.

Case counts up now that we’re past Easter.

Cases hit local peaks including in NYC, indicating BA.2 wave won’t be too bad.

Also, did you hear? Elon Musk is buying Twitter.

Let’s run the numbers.

The Numbers

Predictions

Prediction from last week: 277,800 cases (+20%) and 4,100 deaths (+71%!?)

Results: 328,762 cases (+41%) and 2,485 deaths (+2%!)

Prediction for next week: 400,000 cases (+22%) and 2,720 deaths (+10%?)

On deaths this is one of the biggest misses I’ve had, and it’s not one of the ones I look back on and instantly feel was simply stupid – often it’s obvious in hindsight why I messed up, and mostly this isn’t one of those cases

There was a huge drop for Easter, and it was sustained even as cases popped back up even more than I expected. I don’t understand it, but the distributions involved look real. I can only interpret this as ‘we had inflated death numbers for a while that weren’t real, and now we have less of that’ because I can’t think of any other way to make sense of this. I guess I’m predicting a small increase next week here, but I mostly notice I am confused.

I believe a lot of this jump in cases is related to Easter, and the details suggest that we are going to be reaching another local peak soon especially in the northeast.

Deaths

That is not many deaths. I don’t know how to interpret the drop being sustained in a way that makes the drop ‘real’ which is why I’m assuming this means that the death counts were previously artificially high.

This in turn makes me wonder if death counts have been artificially high at other times as well.

Cases

New York City cases have plateaued and are falling. This is a big deal because New York was ‘ahead of the curve’ on the current BA.2 wave, and there is uncertainty about how far things are likely to go before they stabilize. If NYC is stabilizing at this low level, it strongly suggests there is nothing to worry about.

The Northeast echoes this, with only a small increase. The larger increases in other areas seem like they have a lot to do with Easter, and involve being ‘farther along’ and/or different weather conditions. Yes, cases are up quite a bit, but the upside risks have diminished, which is the most important thing.

Physical World Modeling

Elizabeth finds Home Antigen tests useless in terms of keeping your gathering safe, based on them failing to find Covid cases. Cue tests were most effective although potentially overly sensitive and definitely expensive. $250 fixed cost for the reader plus $65 per test is not a reasonable additional cost to impose in most circumstances. If my job asked for such testing daily (as did the workplace in the post) and I wasn’t paid at least mid six figures, I’d be pretty upset. She did note that the antigen tests are still good at knowing if you had gotten over Covid yet and could end your isolation, which is a different use case.

Alas, public health information in many places continues to say the opposite, and even cancelling rapid test appointments for people with recent Covid.

For any ‘normal’ gathering, at this point in the pandemic, I do not ...