Welcome to The Nonlinear Library, where we use Text-to-Speech software to convert the best writing from the Rationalist and EA communities into audio. This is: What’s the chance a smart London resident dies of a Russian nuke in the next month?, published by DanielFilan on March 10, 2022 on LessWrong.

Credit allocation: the structure of this estimate, and some of the numbers, are basically lifted from a talk Nuño Sempere gave, but most of the numbers aren’t.

Epistemic status: I spent maybe an hour putting this together. I spent more time thinking about the bit I wrote about more, and less time thinking about the bit I wrote about less. For everything but the prior, you might want to check out the Samotsvety Group’s forecast here.

tl;dr About 0.01%.

As I write this, there’s a lot of tension between the West and Russia over Russia’s invasion of Ukraine. If you’re living in a major city like London, you might wonder whether you should move to the countryside in case your city is the target of a nuclear strike. Here’s my estimate of the chances of that, as well as some context for how bad those odds are.

First of all: what’s the prior odds for this sort of thing? Let’s say we’re talking about a NATO nuclear exchange with Russia. Since the US and Russia have co-existed as nuclear powers for about 70 years, you would think that Laplace’s law of succession would give you about 70:1 odds against a nuclear strike in a normal year. That being said, I think this is conservative: Laplace’s law assumes that in the absence of historical data, you’d have even odds on a nuclear strike. But if I didn’t know any history, I’d think the odds of intense conflict in a given year would be like 20:1 against, not 1:1, since countries don’t want to have this sort of conflict and have ways of avoiding it. The way we adjust for that is that we add that 20 to the 70 years to get a prior odds of 90:1 - see this Wikipedia page on pseudocounts.

I then think we should update down a bit because there’s a relatively wide class of intense NATO-Russia conflicts, only some of which would lead to nuclear war. I’m going to say that that’s a 1.5:1 update against (a mild update because it’s easy to see how intense NATO-Russia conflict leads to a nuclear exchange), and then get 135:1 odds against, on priors. Note that this is a different kind of update to adding pseudocounts that produces different effects, and I’m not totally sure it’s legit, but I’m going with it.

Now: we should probably update on the fact that Russia’s invading Ukraine, and the West is sanctioning Russia over that. The question is, how big an update is that? Bayes’ rule says we multiply the odds by the likelihood ratio: that is, the ratio between the probability of something like the current conflict happening given nuclear escalation this year, and the probability of something like the current conflict happening given no nuclear escalation this year. I’ll treat those two separately.

So: what’s the probability of something like the Ukraine situation given a nuclear exchange this year? I’d actually expect a nuclear exchange to be precipitated by somewhat more direct conflict, rather than something more proxy-like. For instance, maybe we’d expect Russia to talk about how Estonia is rightfully theirs, and how it shouldn’t even be a big deal to NATO, rather than the current world where the focus has been on Ukraine specifically for a while. So I’d give this conditional probability as 1/3, which is about 3/10.

What’s the probability of something like the Ukraine situation given no nuclear exchange this year? Luckily, we can actually empirically estimate this, by looking at all the years NATO and Russia haven’t had a nuclear exchange, and seeing how many of them had something like this Ukraine situation. I’d count the NATO bombing of Yugoslavia, the initial invasion of Ukraine, the Cuban missile crisis, and the Russian invasion of Afghanistan. Let’s say Yugoslavia counts for 1 year, Ukraine 1 counts for...

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: Companies with the most EAs and those with the biggest potential for new Workplace Groups, published by High Impact Professionals on March 11, 2022 on The Effective Altruism Forum.

Introduction

EA Workplace and Professional Groups make up an important part of the EA landscape. Building effective altruism is a top cause area according to 80K and EA surveys indicate that something like 50% of EAs are professionals not working for EA organizations. These facts together imply that community building with workplace and professional groups has a large potential for impact.

As part of our work at High Impact Professionals (HIP), we’ve been conducting research into the companies where EA professionals work. A better understanding this will allow high-impact professionals to:

better determine where to start workplace groups

join existing workplace groups they might not have known existed

decide which companies to join in the first place

This is the first in a series of posts about EAs and workplaces and presents data about companies with the most EAs.

Thanks to Adam Steinberg, Sunnie Huang and Jona Glade for their feedback, as well as Soon Khen Ow Yong for his research. All errors are our own.

Companies with the Most EAs

The diagram below shows the 30 organizations with the most in-house EAs, according to our research, along with the number of EAs we were able to identify. The colors represent the stage of each workplace group.

Organizations in green have an active workplace group; we plan to support these groups and help them thrive. For red and yellow organizations, we are looking to help create or revive a group by recruiting a group organizer whose efforts we can support. Finally, blue represents organizations where groups are currently being created.

If you are interested in more, you can also check out our more comprehensive list of organizations and the EAs working for them.

Actions Professionals Can Take

If you are a working professional interested in having more impact, here are some actions you can take.

Start a workplace group

As a working professional, you can have significant impact by creating a workplace group at your organization, by getting colleagues interested in EA, running workplace initiatives, and keeping members value-aligned to EA principles, just to name a few. If you are interested in learning more about what setting up a group could look like and whether this might be a good fit for you, please book some time with us.

If you happen to work for one of the companies listed in the bar graph above, so much the better, as these companies have a critical mass of EAs who can help get your group off to an especially solid start. That being said, even if you work for a company that isn’t listed we’d still love to talk to you.

Help us find high-impact leaders

If you are not in the position to start a group yourself, chances are good you know someone who is, so please also reach out if you know someone who works for an organization who might be a good candidate for starting a group. Making connections that otherwise wouldn’t have been made can be an easy way to be impactful.

Connect with an existing group

You can also have an impact and build out your network by joining an existing group - there are something like 40-50 active groups, so the ones above represent only a fraction of what’s out there. Watch our website for a list of all active EA workplace and professional groups. We will let you know when this goes live, but until then you can check out this post or look on the EA hub.

Methodology In Brief

To gather this data, we analyzed EA group membership on LinkedIn as an indicator of the number of EA-aligned professionals at different organizations. Employees who declared themselves a member of any of a number of EA LinkedIn groups were counted as an EA at...

Welcome to The Nonlinear Library, where we use Text-to-Speech software to convert the best writing from the Rationalist and EA communities into audio. This is: A Longlist of Theories of Impact for Interpretability, published by Neel Nanda on March 11, 2022 on The AI Alignment Forum.

I hear a lot of different arguments floating around for exactly how mechanistically interpretability research will reduce x-risk. As an interpretability researcher, forming clearer thoughts on this is pretty important to me! As a preliminary step, I've compiled a list with a longlist of 19 different arguments I've heard for why interpretability matters. These are pretty scattered and early stage thoughts (and emphatically my personal opinion than the official opinion of Anthropic!), but I'm sharing them in the hopes that this is interesting to people

(Note: I have not thought hard about this categorisation! Some of these overlap substantially, but feel subtly different in my head. I was not optimising for concision and having few categories, and expect I could cut this down substantially with effort)

Credit to Evan Hubinger for writing the excellent Chris Olah's Views on AGI Safety, which was the source of several of these arguments!

Force-multiplier on alignment research: We can analyse a model to see why it gives misaligned answers, and what's going wrong. This gets much richer data on empirical alignment work, and lets it progress faster

Better prediction of future systems: Interpretability may enable a better mechanistic understanding of the principles of how ML systems and work, and how they change with scale, analogous to scientific laws. This allows us to better extrapolate from current systems to future systems, in a similar sense to scaling laws.

Eg, observing phase changes a la induction heads shows us that models may rapidly gain capabilities during training

Auditing: We get a Mulligan. After training a system, we can check for misalignment, and only deploy if we're confident it's safe

Auditing for deception: Similar to auditing, we may be able detect deception in a model

This is a much lower bar than fully auditing a model, and is plausibly something we could do with just the ability to look at random bits of the model and identify circuits/features - I see this more as a theory of change for 'worlds where interpretability is harder than I hope'

Enabling coordination/cooperation: If different actors can interpret each other's systems, it's much easier to trust other actors to behave sensibly and coordinate better

Empirical evidence for/against threat models: We can look for empirical examples of theorised future threat models, eg inner misalignment

Coordinating work on threat models: If we can find empirical examples of eg inner misalignment, it seems much easier to convince skeptics this is an issue, and maybe get more people to work on it.

Coordinating a slowdown: If alignment is really hard, it seems much easier to coordinate caution/a slowdown of the field with eg empirical examples of models that seem aligned but are actually deceptive

Improving human feedback: Rather than training models to just do the right things, we can train them to do the right things for the right reasons

Informed oversight: We can improve recursive alignment schemes like IDA by having each step include checking the system is actually aligned

Note: This overlaps a lot with 7. To me, the distinction is that 7 can be also be applied with systems trained non-recursively, eg today's systems trained with Reinforcement Learning from Human Feedback

Interpretability tools in the loss function: We can directly put an interpretability tool into the training loop to ensure the system is doing things in an aligned way

Ambitious version - the tool is so good that it can't be Goodharted

Less ambitious - The could be Goodharted, but it's expensive, and this shifts the inductive biases to favour aligned cognition

Norm setting: If interpretability is easier, there may be ...

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 3/10/22: We Have a Plan, published by Zvi on March 10, 2022 on LessWrong.

Covid situation continues to improve, although Covid remains a thing. There’s a new executive plan that I did my best to read with incomplete success, that checks off lots of boxes and doesn’t have anything horribly wrong with it.

What’s less great is that Congress passed an omnibus bill that cut out all pandemic funding and most already allocated funds have been spent, so we are failing to pay for our Paxlovid purchases and threaten to run into a wide range of other problems as well if this isn’t addressed, such as us no longer paying for the uninsured to get treatment and running out of monoclonal antibodies. It would be quite the disaster, although scaled to the extent of cases at the time.

In the meantime, slowly but surely less and less masking is required and vaccine cards are checked less, and people start to see each other more often the way people used to. We presume that like most such things the snafus with pandemic funding will get sufficiently worked out that it won’t actually cause us to lose access to treatments, although I wouldn’t fully count on it. Life returns, and we can shift our doom scrolling to the invasion of Ukraine.

On that front, writing continues to be difficult. I’ve written a post focused on what options are available, both on an individual and collective level, on the continued theory of ‘this is the piece of the puzzle I know how to write right now’ and writing each piece even out of order makes other pieces easier.

Meanwhile, back to the Covid show.

Executive Summary

Conditions are safe and one can return to living life.

Congress withholds pandemic funding, endangering treatment if not fixed.

New executive plan for Covid seems fine, so long as they get their funding.

Let’s run the numbers.

The Numbers

Predictions

Prediction from last week: 247k cases (-25%) and 9.200 deaths (-19%).

Results: 226k cases (-32%) and 9,138 deaths (-20%).

Prediction for next week: 170k cases (-25%) and 7,050 deaths (-22%).

The decline in deaths finally picked up a bit, which I expect to accelerate a bit more. Cases pleasantly surprised, as it seems behavioral changes haven’t slowed the decline much yet, although I will continue to predict that those changes will soon have some impact. I definitely see substantial changes taking place around me.

Deaths

I found the log scale button, so here’s the new deaths graph.

Some excellent data journalism indeed, relative death rates over time edition:

Cases

Now in log scale.

The decline in cases continues to be steady in log/percentage terms. I do expect that to slow somewhat soon as behavior adjusts, but no sign of it so far.

I Read The Plan So You Don’t Have To

You’re welcome.

Well, all right, by read the plan I mean ‘read the executive summary carefully then skim the rest to see if any of it isn’t content-free’ since that already took hours and most of the sentences are content-free, but hey, you still don’t have to and that’s what counts.

First I accidentally started reading the January 2021 version of the plan by accident, and you know what? Once I realized it was from January 2021, it was pretty good. It had way too many different priorities and way too much bluster and made some major conceptual errors, and everything useful was kind of obvious, sure, but other than that it seemed like it was mostly was a good plan.

How’s the new plan?

I nitpick a bunch throughout here, but mostly I was actually pretty impressed. There’s a lot of thought to logistics and thinking ahead in ways that could physically matter. Yes, there’s also a lot of bragging, both justified and otherwise, and all that, and a lot of stuff that’s content-free especially when it’s talking about equity, but there was no avoiding that. I’m not quite as high on it as Scot...

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: EAGxBoston: Updates and Info from the Organizing Team, published by Kaleem on March 11, 2022 on The Effective Altruism Forum.

TLDR: All you need to know about EAGxBoston - It's going to be awesome !

Also: If you would like to nominate a friend or colleague to present at EAGxBoston, or have presented at an EAG(x) before and would like to do so at EAGxBoston, please email us at [email protected]

(edit: the email address above only got corrected at 1pm EST on the 11th of March - if you were trying to email us before that, we apologize and kindly ask you to resend your email to the corrected address)

Intro

EAGxBoston will be taking place from 6pm on Friday the 1st of April until 8pm on Sunday the 3rd of April. We’re very excited about how big, exciting, and high quality this conference is shaping up to be—we are prepared to welcome up to 1,000 attendees and it’s looking possible that we’ll actually be doing that. If this is your first time attending an EA conference (or even if you have attended one before but perhaps didn’t take preparation seriously) we suggest reading these excellent posts about how to get ready for an EA conference. We’re also still eager to expand our team of pre-event and on-the-day volunteers (it’ll be fun and impactful) ! There are lots of other ways to engage with other attendees before the event: there is an #eagxboston-2022 channel on the Groups Slack, there is this WhatsApp group, there are EA Facebook pages, and the Swapcard app (launching soon!). The applications for EAGxBoston close on the 21st of March.

Dates and times

In case you missed this somehow, the conference is happening from the 1st to 3rd of April 2022. We hope to make the detailed schedule of sessions at the conference available soon, through the Swapcard app. In the meantime, we can tell you generally that there will be an opening reception and career fair/internship fair/academic program fair from 6-9pm on the 1st of April, and that the conference will be open from 8am-8pm on the 2nd and 3rd, but we will try to have the majority of programming fall between 10am and 6pm.

Location

The conference will be happening at the Boston Convention and Exhibition Center (BCEC) located at 415 Summer St, Boston, MA 02210, in Boston’s Seaport District. There are multiple ways to get to the venue: You can use the MBTA (Massachusetts Bay Transportation Authority)’s Red Line or Silver Line, or the 4, 7, or 11 bus, depending where you are starting from. If you’re using the red line, South Station is the closest station to BCEC (~10min / ~1.1km/ ~0.7mile walk). Broadway station is also close (~12min / ~1.4km / ~0.9mile walk). If you use the silver line, the SL1, SL2, SL3, and SLW all stop right next to BCEC.

Venue info

If you’ve looked at BCEC online or via maps, you’d have noticed that it's basically the size of a small airport. We will be using the entirety of one side of the second floor—that is to say, we have loads of space. In addition to the massive amount of meeting (and eating) space, you can also walk around the rest of the interior of BCEC, or around its exterior perimeter.

What there’ll be in the venue

Coat check

Quiet space

2 large one-on-one spaces with >100 2-person tables

General session space

Dining space for ~800 people at a time

7 ~60 person workshop/meeting spaces

First aid

Security at the entrances

What we recommend you bring with you

Refillable water bottle

Rain jacket/umbrella

Laptop/notebook

Laptop and phone chargers

Things to help you de-stress or relax during off time, such as a book or headphones or an eye mask if you want to take a nap

Necessary medications

Covid policy

We will not mandate the wearing of masks. However, we encourage people to think seriously about their decision to wear/not wear one with respect to other attendees’ safety. We will have ~1,000 KN95 m...

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: Replacing micromarriages, published by Richard Ngo on March 11, 2022 on LessWrong.

tl;dr micromarriages are ill-defined, and not analogous to micromorts. I recommend using QAWYs (Quality-adjusted Wife Years) instead, where 1 QAWY is an additional year of happy marriage.

I once compiled a list of concepts which I’d discovered were much less well-defined than I originally thought. I’m sad to say that I now have to add Chris Olah’s micromarriages to the list. In his words: “Micromarriages are essentially micromorts, but for marriage instead of death. A micromarriage is a one in a million chance that an action will lead to you getting married, relative to your default policy.”

It’s a fun idea, and sometimes helpful in deciding when to go to various events. But upon thinking about it more, I’ve realised that the analogy doesn’t quite work. The key difference is that micromorts are a measure of acute risk - i.e. immediate death. For activities like skydiving, this is the main thing to worry about, so it’s a pretty good metric. But most actions we’d like to measure using micromarriages (going to a party, say, or working out more) won’t lead you to get married immediately - instead they flow through to affect marriages that might happen at some later point.

So how can we measure the extent to which an action affects your future marriages, even in theory? One option is to track how it changes the likelihood you’ll get married eventually. But this is pretty unhelpful. By analogy, if micromorts measured an action’s effect on the probability that you’d die eventually, then all actions would have almost zero micromorts (with the possible exception of some life-extension work during the last few decades). Similarly, under this definition the micromarriages you gain from starting a new relationship could be mostly cancelled out by the fact that this relationship cuts off other potential relationships.

An alternative is to measure actions not by how much they change the probability that you’ll get married eventually, but by how much you expect them to causally contribute to an eventual marriage. The problem there is that many actions can causally contribute to a marriage (meeting someone, asking them out, proposing, etc) and there’s no principled way of splitting the credit between them. I won’t go into the details here, but the basic problem is the same as one which arises when trying to allocate credit to multiple contributors to a charitable intervention. E.g. if three different funders are all necessary for getting a project off the ground, in some sense they can all say that they “caused” the project to happen, but that would end up triple-counting their total impact.

(In this case, we can use Shapley values to allocate credit - but the boundaries between different “actions” are much more arbitrary than the boundaries between different “agents”, making it harder to apply Shapley values to the micromarriage case. Should we count the action “skipping meeting someone else” as a contributor to the marriage? Or the action “turning your head to catch sight of them”? This is basically a rabbit-hole without end - and that’s not even getting into issues of marriage identity across possible worlds.)

Fortunately, however, there’s another approach which does work. When thinking about mortality, the medical establishment doesn’t just measure acute risks, but also another category of risk: chronic risks, like smoking. When smoking, you don’t get a binary outcome after each cigarette, but rather a continual degradation of health. So chronic risks are instead measured in terms of the expected decrease in your lifespan - for example, with units of microlives, where one microlife is one millionth of an adult lifespan (about half an hour); or with quality-adjusted life years (QALYs), to adjust fo...

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: Lessons and results from workplace giving talks, published by Jack Lewars on March 10, 2022 on The Effective Altruism Forum.

This is the third in a series of posts about effective giving in 2022, from One for the World and Giving What We Can. You can read part 1 on why we should promote effective giving here, and part 2 on longtermist donation opportunities here. Thanks to Luke Freeman for his help with the draft.

It also follows two previous articles on ‘How we promoted EA at a large tech company’.

Tl;dr

One for the World has now conducted ~50 corporate talks on effective giving since October 2020.

These have raised nearly $600k in one-off donations to GiveWell charities; and $131k in recurring donations. The expected value of the latter is ~$400k over 5 years.

This is still a small dataset but there are some signals about what makes a corporate talk effective at raising donations.

There are now several organisations with the materials, skills and experience to deliver effective corporate talks - most notably One for the World, High Impact Professionals, Giving What We Can and the EA and Consulting Network. Jona Glade is also running point on workplace outreach at CEA.

We believe that this is a neglected source of donations, both small and major, which can help to diversify EA funding positively; but also potentially a source of talent, power and technical expertise.

Call to action

CTA: Fill out this form if you’d like support in promoting effective giving at your workplace! This can include:

Requesting an expert speaker

Getting trained to deliver effective giving talks yourself

Getting ready-made materials such as presentations, graphics and advertising copy

Getting a checkout page/site set up for your company

Results

~50 talks feels enough to start tentatively forming lessons on what works best. The results of these talks are as follows.

Disclaimers:

Corporate giving platforms are extremely slow to disburse and report, so numbers are often revised. However, these are very unlikely to change significantly, and if they do they are most likely to improve, as we have been sceptical in putting them together.

Money listed here is ‘real’, i.e. it has moved through One for the World’s bank account or been demonstrated with donation receipts. Tracking donations and pledges inspired by Giving What We Can is much harder than tracking those processed by One for the World, so it is likely that we are undercounting pledges and donations not made via One for the World.

Recurring and one-off donations include matching from host companies but not from outside sources such as Double Up Drive (so $1200/year from Microsoft is most probably $600 from the donor and $600 from the company).

These figures only reflect talks delivered by One for the World or One for World & Giving What We Can (indicated by an asterisk), not other EA Groups/organisations.

Companies are based in a variety of locations - mainly the US, but also the UK, Canada, Australia and in Europe. US talks produce the most money, reflecting historical giving trends.

All figures are USD.

Company# of talksRecurring donations (annual value)Average annual run rate per talkOne-off donationsNotesBridgewater+$28,400 from Double Up DriveAdobe MicrosoftIncludes $412k from a single donorFuturice Bain BJSS Mott Macdonald Plaid Miller HullSome, but not yet reported SimpliSafe Blackrock Google Accenture DSGMBH FacebookSingle recurring donor; includes $20k matchFacebookDonations are from a Giving What We Can pledgeBridgespanHarder to track donations/pledges inspired by GWWCBigCommerceHarder to track donations/pledges inspired by GWWCEconicHarder to track donations/pledges inspired by GWWCFlatiron HealthHarder to track donations/pledges inspired by GWWC; includes $25,750 matchQualcommHarder to track donations/pledges inspired by GWWCTot...

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: Participate in the Next Generation for Biosecurity Competition, published by jtm on March 10, 2022 on The Effective Altruism Forum.

This week, the 6th iteration of the Next Generation for Biosecurity Competition was announced by the Nuclear Threat Initiative, Next Generation for Global Health Security Network, the iGEM Foundation, SynBio Africa, and the Global Health Security Network.

Along with Harshu Musunuri and Jonas Sandbrink, I participated in the 2020 version of the competition (and won!) , and I would highly recommend students and early-career folks focused on biosecurity to participate this year.

I think our team spent about 45-65 hours writing up the paper between the three of us (including meetings), over the course of 2-4 weeks and I think that’s fairly representative of what you could expect.

The competition

From the announcement:

“The competition aims to cultivate the next generation of global leaders in biosecurity. We seek innovative and creative papers focused on verification measures for the Biological Weapons Convention (BWC).

Winners of the Next Generation for Biosecurity Competition will receive:

Online publication of their paper on the NTI website

The opportunity to travel, attend, and present during a side event at a prestigious global health security event, such as the Global Health Security Conference 2022.

ELIGIBILITY

Teams must have three participants and include members from two or more countries and/or regions. Multisectoral teams are strongly encouraged.

Applicants must be currently enrolled in an academic institution or have less than five years of professional experience.

DEADLINE

Submission deadline is April 18, 2022 at 11:59PM ET. Submissions should be sent to [email protected]. Winners will be announced in June 2022.

NEED A TEAM?

Participants can find teammates by:

Completing the Next Generation for Biosecurity Interest Form. Once you have completed this form, you may find potential teammates here.

Becoming a member of the Next Generation GHS Network. Members can find teammates through the NextGen GHS Network Hub, which is available to all members after registration. To become a member, click here."

Some good reasons for participating

I was reluctant about participating in the competition during a busy summer but am really happy that I did. Below are some of the things that I got out of it that I think you could, too.

I should note that people's experience probably depends a lot on how well they get along with their team.

Working with awesome peers. The competition can be a terrific opportunity to meet new peers from across the world with a shared interest in biosecurity, or to strengthen your bonds with someone you already know but haven't had the chance to work with before.

Making invaluable connections in the biosecurity field. The competition is sponsored by some of the coolest organisations in the biosecurity field and is organised, I think, largely because they want to connect with promising young people. I still interact and work with several of the organisers and judges I met in 2020. The winning team will have especially good chances to connect with people via presentations at global events, but I can imagine it's possible to make some good connections if you're motivated, regardless of whether you win.

Starting a paper that you can expand, and maybe publish, after the competition. After the competition, my team and I continued working on our paper, and ultimately got to publish (a heavily revised version of) it in the journal mBio. Of course, this took a lot of extra work and some topics are undoubtedly easier to publish than others, but I'd say there's a decent chance for this to work out for highly motivated and creative teams, even if they don't win the competition.

Contributing to an important open question. Verification...

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: I left Russia on March 8, published by avturchin on March 10, 2022 on LessWrong.

I moved to Bishkek in Kyrgyzstan, a small country in Central Asia. I am travelling with family and a toddler. The town is lovely and has many nice cafes.

International travel from Russia is still possible here, as many work migrants in Russia are from Kyrgyzstan. So planes are flying and money sending services seems to work.

There was some kind of interview on the Russian side of the border: when you will return? why your vacation is so long? Some people said that their phones were checked and border guards read their message history. My message history was clean.

Note that any non-official opinion about the current Russian operation could be punished by 15 years in jail according to the new law. It is also illegal to call it "war". So far no one was jailed for 15 years, but some people were fired or arrested for a shorter time after protests.

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: Update from Open Philanthropy’s Longtermist EA Movement-Building team, published by ClaireZabel on March 10, 2022 on The Effective Altruism Forum.

Summary

Open Philanthropy’s Longtermist EA Movement-Building team aims to grow and support the pool of people who are well-positioned to work on longtermist priority projects.

This post outlines our recent work and strategic updates as a team, and isn’t meant to represent the work or views of other teams at Open Phil.

We think this is a very promising space, and we’re hiring for several roles so that we can move faster and deploy more funding.

Over time, we have become more confident in the value of the grants we’ve already made, since our mostly grantees seem to be bringing in promising people to work on longtermist projects at a good rate.

This has led us to begin spending our time differently:

Less time evaluating opportunities (since we’ve come to think that most of the things we want to fund will probably be above our “bar” for impact)

More time trying to generate additional opportunities (e.g. by creating different programs where people can apply for funding, like our scholarships or course development grants).

More time trying to better understand the field and share our findings.

We’ve also come to prioritize “time-effectiveness” over “cost-effectiveness” in most cases (that is, aiming to achieve our goals while conserving EA time/labor, even if that means spending more money).

I think we should have made those changes faster than we did, and see it as a mistake that I didn’t (a) hire more quickly and (b) advocate more forcefully for certain opportunities that were promising, but difficult to evaluate.

For our future grantmaking on our team, I’m concerned about avoiding measurability bias (prioritizing grants that come with impressive numbers/credentials attached) and certain forms of motivated reasoning.

There are many kinds of projects we hope to fund in the future that could allow us to sharply scale up our total grantmaking.

For much more detail on all of this, see the rest of the post.

Introduction

This post is a report and update on the Open Philanthropy Longtermist Effective Altruism Movement-Building team’s thinking and goals. It’s written by me, Claire, and mostly represents my perspective. I’m writing this in my role as an Open Phil staff member, but I take sole responsibility for the angsty commentary near the bottom.

Our team currently consists of me, Asya Bergal, Bastian Stern, and Eli Rose. We are supported by Open Phil’s “longtermist budget” (funding to support projects motivated by the longtermist view), but unlike the other longtermist cause areas, we aren’t aiming to make progress on longtermist priorities directly. Instead, our goal is to grow and support the pool of people motivated and well-positioned to work on longtermist priority projects (e.g. reducing existential risk and aiming to improve the far future).

I think our team’s grantmaking has high expected value, because (1) in my experience, most of the relevant object-level longtermist projects are bottlenecked by the dearth of aligned people who are good fits (so our goals are aimed at a core problem), (2) there’s a lot of funding to direct (which is bottlenecked by the number of grantmakers working to direct it), and (3) we have a reasonably high number of potential grantmaking projects we are working as fast as we can to implement (i.e. there’s a feeling of traction) and could implement faster if we had more capacity on the team. Not coincidentally, we are currently hiring for several roles.

What’s happened so far

I took over this area from Nick Beckstead (who was working on it part-time) in early 2019. Bastian started working with me and Eli joined in 2020, and Asya joined in 2021.

I think a lot has changed since then, and a lot of imp...