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: Transforming Democracy: A Unique Funding Opportunity for US Federal Approval Voting, published by aaronhamlin on June 30, 2023 on The Effective Altruism Forum.

I'm excited to share a special opportunity to create a systemic impact: a statewide approval voting ballot initiative in Missouri. This would affect all elections throughout the state including federal and presidential. Approval voting favors consensus candidates and a more accurate representation of the public's support. This is critical if we want a government to behave in our interests on policies that concern our well-being.

The organization leading this charge is Show Me Integrity, where I'm currently doing a fellowship and assisting with fundraising efforts. Show Me Integrity has successfully passed a ballot initiative before, showing their ability to succeed on this kind of scale. They also successfully ran the ballot initiative for approval voting in St. Louis.

Why is this important?

Approval voting is a method that allows voters to select as many candidates as they want; still, most votes wins. Approval voting, an easy-to-implement system, can greatly improve our current plurality-based approach to electing Federal and state-level positions. If you’ve read my writing on this before, you’ve seen me make that case. And this is much more effective and lasting than putting money behind individual candidates. This opportunity may not come around again.

The Impact

This initiative is not just about changing the voting method; it's about transforming how we elect individuals to government office, from local to federal positions, in the 19th largest state of over 6 million people. This includes influencing presidential electoral votes. This is the first statewide ballot initiative for approval voting, making it a pioneering effort with potentially far-reaching implications.

The Ask

We are currently in the signature-gathering phase, a crucial step that requires initial funding. The cost for signature gathering is around $4M, with an additional $9M needed later for campaign execution. Yes, it's expensive, but the potential impact justifies the investment.

About Show Me Integrity

Show Me Integrity has a history of successfully implementing ballot initiatives, including a statewide initiative and passing approval voting in St. Louis. Their experience and proven success make them an ideal organization to lead this initiative. There is no better opportunity.

Logistics and Challenges

There are some challenges to be aware of. The competition for signature gathering means that costs could increase if we can't secure initial funding soon. This could also necessitate the use of different firms, which may not have the same quality. With timely funding, we can overcome these hurdles. Additionally, while the initial polling is over 60% even with opposition messaging, this support can change. Ballot measures can be risky.

A Matching Opportunity

To encourage donations, a generous donor is currently offering a match of $600K. This match may end soon, but there's a possibility of an extension. This is a true match, meaning the donor will only match funds that can meaningfully kickstart the signature-gathering process. $100K of this match is already being met by other donors.

Your Chance to Make a Difference

This is a rare opportunity to contribute to a significant societal change. If you, someone you know, or an institution is interested in supporting this initiative, please reach out to me at [email protected]. [Please do not use my old CES email.]

Thank you for considering this opportunity. Your support could help transform our democracy and create a lasting impact.

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: Agency from a causal perspective, published by Tom Everitt on June 30, 2023 on The AI Alignment Forum.

Post 3 of Towards Causal Foundations of Safe AGI, preceded by Post 1: Introduction and Post 2: Causality.

By Matt MacDermott, James Fox, Rhys Ward, Jonathan Richens, and Tom Everitt representing the Causal Incentives Working Group. Thanks also to Ryan Carey, Toby Shevlane, and Aliya Ahmad.

The purpose of this post is twofold: to lay the foundation for subsequent posts by exploring what agency means from a causal perspective, and to sketch a research program for a deeper understanding of agency.

The Importance of Understanding Agency

Agency is a complex concept that has been studied from multiple perspectives, including social science, philosophy, and AI research. Broadly it refers to a system able to act autonomously. For the purposes of this blog post, we interpret agency as goal-directedness, i.e. acting as if trying to direct the world in some particular direction.

There are strong incentives to create more agentic AI systems. Such systems could potentially do many tasks humans are currently needed for, such as independently researching topics, or even run their own companies. However, making systems more agentic comes with an additional set of potential dangers and harms, as goal-directed AI systems could become capable adversaries if their goals are misaligned with human interest.

A better understanding of agency may let us:

Understand dangers and harms from powerful machine learning systems.

Evaluate whether a particular ML model is dangerously agentic.

Design systems that are not agentic, such as AGI scientists or oracles, or which are agentic in a safe way.

Lay a foundation for progress on other AGI safety topics, such as interpretability, incentives, and generalisation.

Preserve human agency, e.g. through a better understanding of the conditions under which agency is enhanced or diminished.

Degrees of freedom

(Goal-directed) agents come in all shapes and sizes – from bacteria to humans, from football teams to governments, and from RL policies to LLM simulacra – but they share some fundamental features.

First, an agent needs the freedom to choose between a set of options. We don’t need to assume that this decision is free from causal influence, or that we can’t make any prediction about it in advance – but there does need to be a sense in which it could either go one way or another. Dennett calls this degrees of freedom.

For example, Mr Jones can choose to turn his sprinkler on or not. We can model his decision as a random variable with “watering” and “not watering” as possible outcomes:

Freedom comes in degrees. A thermostat can only choose heater output, while most humans have access to a range of physical and verbal actions.

Influence

Second, in order to be relevant, an agent’s behaviour must have consequences. Mr Jones decision to turn on the sprinkler affects how green his grass becomes:

The amount of influence varies between different agents. For example, a language model’s influence will heavily depend on whether it only interacts with its own developers, or with millions of users through a public API. Suggested measures of influence include (causal) channel capacity, performative power, and power in Markov decision processes.

Adaptation

Third, and most importantly, goal-directed agents do things for reasons. That is, (they act as if) they have preferences about the world, and these preferences drive their behaviour: Mr Jones turns on the sprinkler because it makes the grass green. If the grass didn’t need water, then Mr Jones likely wouldn’t water it. The consequences drive the behaviour.

This feedback loop, or backwards causality, can be represented by adding a so-called mechanism node to each object-level node in the original graph. The mechanism n...

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: CE: Rigorously prioritizing the top health security (biosecurity) ideas, published by CE on June 30, 2023 on The Effective Altruism Forum.

Every year at Charity Entrepreneurship (CE), we try to find the best interventions for impact-focused charities to launch through our Incubation Program. Our research in 2022 focused on health security (which includes biosecurity) and on large-scale global health.

This post outlines:

An introduction to health security at CE. What CE means by health security exactly, some of CE’s views on preventing catastrophic and existential risks, and why we picked “health security” as a topic.

Our prioritization of the top health security interventions. Why rigorous prioritization of ideas is needed to be effective in the world (especially for longtermist cause areas), and CE’s approach to this within health security.

Our results. The long list of nearly 200 ideas, our top six shortlisted ideas, and our final recommendations.

If you’d like to learn how to conduct similar prioritization research, and how to evaluate the results that come out of starting high-impact interventions, we have just launched our Research Training Program that you can apply to by July 17, 2023.

Introduction to health security at CE

What we mean by health security

We took a fairly broad interpretation of health security. Our initial definition was: anything that minimizes the danger of acute public health events.

This included several broad categories:

Pandemic prevention

Pandemic preparedness

General risk preparedness for acute public health events

Antimicrobial resistance

Health system strengthening

Other / meta ideas (e.g., better biosecurity forecasting)

This included looking at both natural pandemics and engineered pandemics (which could be due to accidental or deliberate misuse of biotech).

How much does CE value preventing catastrophic/existential risks?

This may seem like a bit of a cop-out answer, but there are a wide range of views on this topic among CE staff. There is no clear overarching organizational view or policy. This section gives a rough sense of how “CE staff” might tend to think about these topics, but probably doesn’t accurately reflect any individual staff member’s stance.

In general, CE’s staff tend to apply a very high level of empirical rigor to decision making. Staff tend to trust empirical evidence (e.g., RCTs, M&E of existing charities, base-rate forecasting, etc.) above other kinds of evidence, particularly valuing such evidence above theoretical reasoned arguments. That said, staff tend to accept that making good decisions requires robust/cluster thinking and look for cases where many kinds of evidence align, including theoretical reasoning. Along the same lines, staff are likely to think that doing good is really difficult and having some ongoing measurable evidence of impact is probably required.

In general, CE staff believe that preventing extinction is a worthwhile endeavor. However, given the above, staff are likely to be skeptical about:

The ability to know what the biggest future risks are, especially where risk estimates rely on reasoned speculation about future technologies.

The success of any organization that doesn’t have a clear path to measure the impact of its activities.

We took all these views into consideration and chose to focus on health security, including biorisks. This focus allows us (and future CE charities) to explore risk areas that have at least some chance of being globally catastrophic, but also have clear historical precedents and evidence of less extreme catastrophes. Additionally, we expected to find a number of options within biorisk that could be impactful in addressing catastrophic biorisks while also demonstrating health impacts in the short run, meaning a new charity should be able to track and demonst...

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: Introducing EffiSciences’ AI Safety Unit, published by WCargo on June 30, 2023 on LessWrong.

This post was written by Léo Dana, Charbel-Raphaël Ségerie, and Florent Berthet, with the help of Siméon Campos, Quentin Didier, Jérémy Andréoletti, Anouk Hannot, Angélina Gentaz, and Tom David.

In this post, you will learn what were EffiSciences’ most successful field-building activities as well as our advice, reflections, and takeaways to field-builders. We also include our roadmap for the next year. Voilà.

What is EffiSciences?

EffiSciences is a non-profit based in France whose mission is to mobilize scientific research to overcome the most pressing issues of the century and ensure a desirable future for generations to come.

EffiSciences was founded in January 2022 and is now a team of ~20 volunteers and 4 employees.

At the moment, we are focusing on 3 topics: AI Safety, biorisks, and climate change. In the rest of this post, we will only present our AI safety unit and their results.

TL;DR: In one year, EffiSciences created and held several AIS bootcamps (ML4Good), taught accredited courses in universities, organized hackathons and conferences in France’s top research universities. We reached 700 students, 30 of whom are already orienting their careers into AIS research or field building. Our impact was found to come as much from kickstarting students as from upskilling them. And we are in a good position to become an important stakeholder in French universities on those key topics.

Field-building programsMachine Learning for Good bootcamp (ML4G)Parts of the MLAB and AGISF condensed in a 10-day bootcamp (very intense)2 ML4G, 36 participants, 16 are now highly involved.

This program was reproduced in Switzerland and Germany with the help of EffiSciences.Turing Seminar AGISF-adapted accredited course that has been taught with talks, workshops, and exercises.3 courses in France’s top 3 universities: 40 students attended, 5 are now looking to upskill, and 2 will teach the course next year.AIS Training DaysThe Turing Seminar compressed in a single day (new format)3 iterations, 45 studentsEffiSciences’ educational hackathonsA hackathon to introduce robustness to distribution change and goal misgeneralization.2 hackathons, 150 students, 3 are now highly involvedApart Research’s hackathonsWe hosted several Apart Research hackathons, mostly with people already onboarded4 hackathons hosted, 3 prizes won by EffiSciences’ teamsConferencesIntroductions to AI risks250 students were reached, ~10 are still in contact with usLovelace programSelf-study groups on AIS4 groups of 5 people each, which did not work well for upskilling.

TL;DR of the content

Results

In order to assess the effectiveness of our programs, we have estimated how many people have become highly engaged thanks to each program, using a single metric that we call “counterfactual full-time equivalent”. This is our estimate of how many full-time equivalent these people will engage in AI safety in the coming months (thanks to us, counterfactually). Note that some of these programs have instrumental value that is not reflected in the following numbers.

ActivityNumber of eventsBy occurrenceFounding the AI safety unit (founders & volunteers)166,0French ML4Good bootcamp27,43,7Word-of-mouth outreach12,92,9Training Day21,91,0Hackathon42,50,6Turing Seminars (AGISF adaptations)31,30,4Research groups in uni40,40,1Frid'AI (coworking on Fridays)50,50,1Conference50,10,0Total3023,0

Counterfactual

full time equivalent

In total, those numbers represent the aggregation of 43 people who are highly engaged, i.e. that have been convinced by the problem and are working on solving it through upskilling, writing blog posts, facilitating AIS courses, doing AIS internships, attending to SERI MATS, doing policy work in various orgs, etc. The time spen...

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 Think Eliezer Should Go on Glenn Beck, published by Lao Mein on June 30, 2023 on LessWrong.

Glenn Beck is the only popular mainstream news host who takes AI safety seriously. I am being entirely serious. For those of you who don't know, Glenn Beck is one of the most trusted and well-known news sources by American conservatives.

Over the past month, he has produced two hour-long segments, one of which was an interview with AI ethicist Tristan Harris. At no point in any of this does he express incredulity at the ideas of AGI, ASI, takeover, extinction risk, or transhumanism. He says things that are far out of the normie Overton Window, with no attempt to equivocate or hedge his bets. "We're going to cure cancer, and we're to do it right before we kill all humans on planet Earth". He just says things like this, with complete sincerity. I don't think anyone else even comes close. I think he is taking these ideas seriously, which isn't something I expected to see from anyone in mainstream media.

According to Glenn, he has been trying to get interviews with people like Geoffery Hinton, but they have declined for political reasons. An obvious low-hanging fruit, if MIRI is willing to send him an email.

He seems on board with a unilateral AI pause from the US for national security reasons.

He's also personally cited statements from Eliezer to convey the dangers of ASI. I think an interview between the two could be a way for AI alignment to be taken more seriously by retirees who will vote and write their congressman. I dislike Eliezer on a personal level, but I think he is the only person who will actually go on the show and express how truly dire the situation is, with full sincerity.

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: MHFC Spring Grants Round, published by wtroy on June 30, 2023 on The Effective Altruism Forum.

The Mental Health Funding Circle recently completed our second round of funding. We are very happy to support great organizations working on highly impactful mental health interventions. Our next open funding round will be held in the Fall - applications will be due October 1st, with final decisions made in early November. For more information, visit our website.

This round, MHFC members disbursed $789,833 to the following organizations, institutions and individuals:

$70,000 to Fine Mind for their direct service work on depression in northern Uganda.

$78,225 to Phlourish for guided self-help in the Philippines.

$50,000 to Happier Lives Institute for continued research on subjective wellbeing metrics and general operating support.

$65,000 to Overcome to support their work offering free online therapy.

$153,000 to Eggshells for their work on digital guided self-help.

$120,000 to Children’s Hospital of Philadelphia to station task-shifting youth therapy practitioners in HIV clinics in Botswana.

$24,000 to School of Hard Knocks to support lay practitioner interpersonal therapy for youth in South Africa.

$30,000 to CEARCH for meta research into effective mental health interventions.

$99,360 to Columbia University and Makerere University for capacity building for interpersonal therapy in primary care in Uganda.

$30,000 to Tata Institute of Social Sciences to support research on interpersonal therapy in primary schools in India.

$70,248 to Swiss Tropical and Public Health Institute for research on suicidality and suicide prevention in urban settings in Zambia.

All funding decisions are made personally by individual circle members and do not necessarily reflect the priorities of the circle. For any information about funding or membership, please reach out!

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: The Case for Overconfidence is Overstated, published by Kevin Dorst on June 28, 2023 on LessWrong.

(Written with Matthew Mandelkern.)

TLDR: When asked to make interval estimates, people appear radically overconfident—far more so than when their estimates are elicited in other ways. But an accuracy-informativity tradeoff can explain this: asking for intervals incentivizes precision in a way that the other methods don’t.

Pencils ready! For each of the following quantities, name the narrowest interval that you’re 90%-confident contains the true value:

The population of the United Kingdom in 2020.

The distance from San Francisco to Boston.

The proportion of Americans who believe in God.

The height of an aircraft carrier.

Your 90%-confidence intervals are calibrated if 90% of them contain the true value. They are overprecise if less than 90% contain it (they are too narrow), and underprecise if more than 90% do.

We bet that at least one of your intervals failed to contain the true value. If so, then at most 75% of your intervals contained the correct answer, making you overprecise on this test.

You’re in good company. Overprecision is one of the most robust findings in judgment and decision-making: asked to give confidence intervals for unknown quantities, people are almost always overprecise.

The standard interpretation? People are systematically overconfident: more confident than they rationally should be. Overconfidence is blamed for many societal ills—from market failures to polarization to wars. Daniel Kahneman summed it up bluntly: “What would I eliminate if I had a magic wand? Overconfidence.”

But this is too quick. There are good reasons to think that when people answer questions under uncertainty, they form guesses that are sensitive to an accuracy-informativity tradeoff: they want their guess to be broad enough to be accurate, but narrow enough to be informative.

When asked who’s going to win the Republican nomination, “Trump” is quite informative, but “Trump or DeSantis” is more likely to be accurate. Which you guess depends on how you trade off accuracy and informativity: if you just want to say something true, you’ll guess “Trump or Desantis”, while if you want to say some thing informative, you’ll guess “Trump”.

So accuracy and informativity compete. This competition is especially stark if you’re giving an interval estimate. A wide interval (“the population of the UK is between 1 million and 1 billion”) is certain to contain the correct answer, but isn’t very informative. A narrow interval (“between 60 and 80 million”) is much more informative, but much less likely to contain the correct answer.

We think this simple observation explains many of the puzzling empirical findings on overprecision.

The Empirical Puzzle

The basic empirical finding is that people are overprecise, usually to a stark degree. Standardly, their 90%-confidence intervals contain the true value only 40–60% of the time. They are also almost never under-precise.

In itself, that might simply signal irrationality or overconfidence. But digging into the details, the findings get much more puzzling:

1) Intervals are insensitive to confidence level

You can ask for intervals at various levels of confidence. A 90%-confidence interval is the narrowest interval that you’re 90%-confident contains the true value; a 50%-confidence interval is the narrowest band that you’re 50%-confident contains it, etc.

A standard finding is that people give intervals of very similar width regardless of whether they’re asked for 50%-, 90%-, or even 99%-confidence intervals. For example, this study found that the widths of 99%-confidence intervals were statistically indistinguishable from those of 75%-confidence intervals.

This is puzzling. If people are giving genuine confidence intervals, this implies an extremely im...

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: AI #18: The Great Debate Debate, published by Zvi on June 29, 2023 on LessWrong.

Is debate worthwhile? That is a matter of some, ahem, disagreement.

A lot of that was mainstream discussions of the whole Rogan-RFK Jr fiasco. In the context of AI, we had a real debate on whether AI is an existential threat that by all accounts didn’t go great, and a Nature article that didn’t even pretend to debate and which Tyler Cowen claimed was evidence that the debate had been utterly lost, with a plea to shift to formal scientific debate instead.

It would be so nice to discuss and explore and figure things out together, instead of debating. Debates, like Democracy, are highly frustrating, deeply flawed institutions that do a terrible job converging on truth, except what alterative do you have that isn’t even worse?

Some noteworthy new releases this week. MidJourney 5.2 is getting rave reviews. Generative AI for GMail, Google Docs and Google Sheets is online, I keep expecting to get use out of it but so far no luck, ideas welcome. The Chinese claim a new very strong language model, ChatGLM, on which I am withholding judgment given the track records involved.

Mostly I sense that we picked the lowest hanging fruit for mundane utility from generative AI, and now our knowledge of, skills at using and integration into our workflows and lives of existing tools is going to be where it is at for a while, modulo the full release of Microsoft Copilot and the fully functional Google suite of generative AI offerings. That doesn’t mean that we won’t see technical advances at a breakneck pace versus almost anywhere else, we totally will, but we’re now seeing things like a $1.3 billion investment in Inflection AI in the hopes of building something better over time. Which takes a while. I don’t doubt that some efforts will get there, but we are now less likely to be talking days or weeks, and more likely to be talking months or years.

Table of Contents

Table of Contents.

Language Models Offer Mundane Utility. AI system. Making copies.

Fun With Image Generation. A few trifles.

Deepfaketown and Botpocalypse Soon. Deepfake detector sighting.

They Took Our Jobs. They will never take our jobs, or freedom?

They Created Our Terrible Job. Work it while it lasts. Need input.

Introducing. GMail generative AI, ChatGLM, get your compute.

In Other AI News. Demis Hassabis says next big thing is coming.

Asterisk Magazine Issue on AI. Outstanding issue all around.

Quiet Speculations. Beware jumping to conclusions.

The Ask. Eliezer Yudkowsky makes the case in clear fashion.

No, No One Is Actually Proposing a Global Surveillance State. Really.

The Quest for Sane Regulation. Debating forms of compute controls.

The Week in Audio. Some very good stuff.

Can LLMs Improve Democracy? Can humans?

Nature’s Op-Ed and Tyler Cowen’s Advice. What debate?

Debate Debate. Oh no, I didn’t mean that debate.

Rhetorical Innovation. New paper on risks, various other things.

Safely Aligning a Superintelligent AI is Difficult. Test for benevolence?

People Are Worried About AI Killing Everyone. Why yes, yes they are.

What Does Hinton Believe? Why is he emphasizing particular points?

Other People Are Not As Worried About AI Killing Everyone. As always.

The Wit and Wisdom of Sam Altman. Billionaire cage matches are fun.

The Lighter Side. Too many jokes ended up elsewhere this week.

Language Models Offer Mundane Utility

Predict hit songs with 97% accuracy using machine learning on brain reactions. This does not yet imply the ability to create new hit songs, but it would mean one could A/B test different versions. What makes me most suspicious is that this level of accuracy implies that hit songs are deterministic. What about all the things such a test can’t possibly pick up on? All the promotion decisions, the cultural developments, the hap...

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: Anthropically Blind: the anthropic shadow is reflectively inconsistent, published by Christopher King on June 29, 2023 on LessWrong.

For the purposes of this post, the anthropic shadow is the type of inference found in How Many LHC Failures Is Too Many?.

"Anthropic principle! If the LHC had worked, it would have produced a black hole or strangelet or vacuum failure, and we wouldn't be here!"

In other words, since we are "blind" to situations in which we don't exist, we must adjust how we do bayesian updating. Although it has many bizarre conclusions, it is more intuitive than you think and quite useful!

There are many similar applications of anthropics, such as Nuclear close calls and Anthropic signature: strange anti-correlations.

This actually has implications for effective altruism. Since we are so early into humanity's existence, we can infer from the anthropic shadow that humans will probably soon die out. Also see The Hero With A Thousand Chances.

More practically, the anthropic shadow should give us useful advice on how to reason about personally risky activities like driving or perhaps even aging. I have not actually seen any advice based on this principle, but theoretically there should be some conclusions you could draw.

The problem, as you probably deduced from the title, is that it is reflexively inconsistent.

Central Claim: Someone using the anthropic shadow should update their decision making to no longer use it. This can be justified with their current decision making procedure.

(This also suggests that if you used it in the past, that was probably the wrong thing to do.)

A weaker (and obvious) claim that is also sometimes called the anthropic shadow is that we do not have experience with situations in which we have died. I agree with this version, but isn't what I will be arguing against.

Note that I am not the first to notice paradoxes with the anthropic shadow. See No Anthropic Evidence for example. I have not yet seen the result about reflexive inconsistency though, hence why I am making this post.

I also introduce the concepts of "Anthropic undeath", "Anthropic angel" (how you would explain an absurdly large number of weird coincidences having to do with death), "Fedora shadow", and apply the central claim to a couple examples. To my knowledge, these contributions are novel.

Anthropically undead: ghosts are as good as gone

This section contains the most general form of the argument. (This could be mathematically formalized; I just haven't gotten around to doing it.) If it seems strange to you, a generalized version of this section might also work.

First, we establish the basic framing of how we will check if something is reflexively consistent. Imagine yourself before a catastrophe potentially happens. You are an expected utility maximizer (as all good agents should be, although this assumption can probably be weakened). You are trying to come up with a policy for your future-self to follow.

Consider the following scenario: in any situation that you would die, imagine instead that you become an agent with only one choice each time it takes an action: "do nothing". This state is still given the same utility as before (including the utility change from physically dying (a reinforcement learner would stop getting rewards, for example)), but as an agent you never stop existing. Call this unreal scenario "anthropic undeath".

Optimizing the utility of the real scenario is the same as optimizing utility in anthropic undeath, because the agent choosing "do nothing" in the anthropic undeath scenario has the same physical effect as what actually happens in the real scenario when the agent is dead. I call this the "ghosts are as good as gone" principle.

The anthropic undeath scenario has no anthropic shadow, because the agent never stops existing. Thus, the ...

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 Reflections on EA Strategy Fortnight, published by Ben West on June 29, 2023 on The Effective Altruism Forum.

I stated “I see this mostly as an experiment into whether having a simple “event” can cause people to publish more stuff” and I feel like the answer is conclusively “yes”. I put a little bit of effort into pitching people, and I’m sure that my title and personal connections didn’t hurt, but this really was not a terribly heavy lift.

Thanks to the fortnight I have a post I can reference for EA being a do-ocracy! I would encourage other people to try to organize things like this.

I noticed an interesting phenomenon: contributors who are less visible in EA wanted to participate because they thought it would give their writing more attention, and people who are more visible in EA wanted to participate because they thought it would give their writing less attention.

I think the average EA might underestimate the extent to which being visible in EA (e.g. speaking at EAG) is seen as a burden rather than an opportunity.

This feels like an important problem to solve, though outside the bounds of this project.

3. Part of my goal was to get conversations that are happening in private into a more public venue. I think this basically worked, at least measured by “conversations that I personally have been having in private”. There are some ways in which karma did not reflect what I personally thought was most important though:

I’ve started to worry that it might be important to get digital sentience work (e.g. legal protection for digital beings) before we get transformative AI, and EA’s seem like approximately the only people who could realistically do this in the next ~5 years. So I would have liked to have seen more grappling with this post, although in fairness Jeff wasn’t making a strong pitch for prioritizing AI welfare.

I also find myself making the points that Arden raised here pretty regularly, and wish there was more engagement with them.

4. When doing a “special event” on the forum, I always wonder whether the event will add to the forum’s engagement or just cannibalize existing engagement. I think the strategy fortnight was mostly additive, although it’s pretty hard to know the counterfactual.

5. Some events I would be interested in someone trying to organize on the Forum

“Everyone change your job week” – opportunity for people to think seriously about whether they should change their jobs, write up a post about it, and then get feedback from other Forum users

Rotblat day – Joseph Rotblat was a physicist on the Manhattan project who was originally motivated by wanting to defeat Nazi Germany, but withdrew once he realized the project was actually motivated by wanting to defeat the USSR. On Rotblat day, people post what signs they would look for to determine if their work was being counterproductive.

“Should we have an AI moratorium?” debate week – basically the comments here, except you recruit people to give longer form takes.

Video week – people create and post video versions of forum articles (or other EA content)

Or more specifically: they would be seen as one of many voices, rather than someone whose opinions should receive special attention/deference

Largely because no one else cares.

This and the Rotblat idea come from Sydney Von Arx

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