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: [Repost] Non-Nashian Game Theory: A Normal-Form Primer, published by Ghislain Fourny on June 1, 2022 on LessWrong.

This is a slightly edited repost of a quick introduction to non-Nashian game theory that I once posted on another platform. I saw that reposts seem to be allowed practice on lesswrong so I hope that this is fine. Specifically, this is a generalization of superrationality to arbitrary games (the only assumption is no ties) and it has, since I wrote the post, been peer-reviewed and published in the Journal of Mathematical Psychology.

For now almost 18 years, I have been researching with my colleagues Jean-Pierre Dupuy and Stéphane Reiche what happens with alternate assumptions in game theory; I like to call this Non-Nashian Game Theory, from what I understand it is close to what is called "acausal trade" on lesswrong, so that I hope it can contribute to the ongoing conversation taking place here. Our first results (back in 2004) were for games in extensive form (trees) in which players perfectly predict each other and the solution concept is mostly based on the avoidance of Grandfather's paradoxes. I suspected back then a relationship of some kind with superrationality, except the latter was defined for games in normal form, as opposed to extensive form.

In 2017, I was finally able to generalize our results to games in normal form, and I would like to share some of the insights below and explain what exactly differentiates it from Nash equilibria (spoiler: this has to do with a definition of free will).

The high-level economic summary of this area of research is very simple to understand:

Rational agents who have integrity and wholesomeness, in the sense that they believe that they are an open book, when playing against each other, achieve better -- namely, Pareto-optimal -- results than those who do not.

Put your Vulcan ears on, because you will need a lot of purely logical thinking. Performing mind meld is also very welcome and aligned with what we are doing here, but optional.

Our example of normal-form game

There is nothing better than a concrete example to show what is going on. Let us consider the following game. In normal form, a game is represented as a matrix. We have a row player that can pick strategy A, B or C, and a column player that can pick strategy D, E or F.

When both players pick a strategy, say for example A and D, then this jointly selects a cell (AD) and this gives us payoffs. The left number (5) is what the row player gets ($5), and the right number is what the column player gets ($1).

We assume that both players would like to gain as much as possible. This is known by economists as "utility maximization", in plain words, rationality. We also assume that both players know what game they are playing. We also assume a couple of fancy properties such as common knowledge of what we have just said, impressive logical and reasoning skills, etc.

A model that tells us what players do, or should do, is called a solution concept. A mainstream solution concept for this kind of games is that of Nash equilibria. The one I would like to present here is called Perfectly Transparent Equilibrium.

The Nash Equilibria

Let us start with what John Nash suggested, back in the 1950s.

A very important assumption is that players can pick their strategies independently. Intuitively, you can think of the players' being in separate rooms.

The row player can reason as follows: if the column player picks D, then what should I pick, in other words: what is my best response to D? Well, if I pick A I get $5, if I pick B $7 and if I pick C $6. So my best response to D is B. Likewise, my best response to E is A, and my best response to F is A.

The column player does the same thing and finds out that her best response to A is F (gets her $4), her best response to B is ...

Welcome to The Nonlinear Library, where we use Text-to-Speech software to convert the best writing from the Rationalist and EA communities into audio. This is: My slack budget: 3 surprise problems per week., published by Raemon on June 1, 2022 on The Effective Altruism Forum.

(Previously posted on LW)

Previously: Slack

In a couple earlier articles I urged people to adopt strategies that reliably maintain a margin of "30% slack." I've seen lots of people burn out badly (myself included), and preserving a margin of resources such that you don't risk burning out seems quite important to me.

But I realized a) "30% slack" isn't very clear, and b) this is an important enough concept it should really have a top-level post.

So, to be a bit more obvious:

Maintain enough slack that you can absorb 3 surprise problems happening to you in a week, without dipping into reserves.

"Surprise problems" can take multiple forms, and cost different types of reserves. These can be financial expenses you didn't know about (whoops, I needed to buy some medicine), or cognitive attention (whoops, I need to figure out what medicine to buy) or stress (whoops, I'm sick, and now I need to talk to a bunch of doctors while being kinda exhausted).

These can be problems happening to you, or problems happening to friends that you care about.

Why "3 surprises", and not just one? Because at least a couple times a year I personally run into 3-surprises-in-a-week. And sometimes I get hit with much bigger things require me to burn my reserves, and if I didn't have a habit of ensuring my reserves I just wouldn't be able to do those things at all.

I was motivated to write this today because, last week, four problems came up. I had recently taken on two major projects; a community institution was in trouble, and a friend was hurt and needed help. I only had bandwidth to deal with 3 of those. I realized that not only was this a particularly bad week, but I had let too many ongoing responsibilities accumulate.

The weekend came 'round, right as I hit exactly-zero-slack. I needed time to recover, but I had made some time-sensitive commitments to help with some of the above things, and I ended up having to spend the weekend doing a more constrained version of those outstanding obligations, while carefully recovering.

Being Pro-Social Requires Slack

I think an important part of being a good friend, community member or effective altruist, is taking care of yourself. I think if you can't absorb 3 surprise problems per week, it probably make sense to prioritize fixing that above most other things.

"But my friend is in trouble!"

That's legitimately sad, but if you can't absorb 3 surprise problems per week, you are probably not going to be able to help your friend next week. Moreover, you might burn out, then you will need to be asking other friends for help, and you might need to ask that at a time when they were also hurting.

Take care of yourself.

"But the world is in trouble!"

The world is always in trouble. There is no end to the things you might hypothetically do to help it. It is virtuous to help. It is not virtuous to help in a way that runs the risk of you becoming one of the people who need help and are adding to the problem.

"But maybe I can do this and it'll be fine?"

One of the legitimately tricky things is that, sure, often you can roll the dice and come out fine this time. Many extra tasks you take on are usually okay, but have, say, a 10% chance of turning out to involve much bigger responsibilities than you thought you were committing to. You can skirt by a few times and be okay.

But, well, if you do that 10 times, one of the times you will turn out not to be okay.

Developing intuitions around how risky actions are, and developing policies for how often to do them, is pretty important.

"But I made commitments!"

I take commitments pretty seriously. Trust is one of the sources of slack. So if you've suddenly realized you've overcommitted, often the solution is no...

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: Experiment in Retroactive Funding: An EA Forum Prize Contest, published by DonyChristie on June 1, 2022 on The Effective Altruism Forum.

Coauthors: Denis Drescher, Matt Brooks

Background

TL;DR: We want to buy the impact of quality EA Forum posts.

We’re working on building impact markets to make public goods tradeable by retroactively funding the labor of the people who create them.

Our overall goal: make the nonprofit space work more like for-profits. Pay for outcomes, not for promises.

We’re doing this to increase funding for public goods and make it easier for large funders to find and fund valuable projects among other benefits. (See below for more detail on our vision for these markets.) We are receiving a grant via the Future Fund Regranting Program and are excited to start launching experiments to iterate towards a large open impact market. You can check out our new informational website and join our community Discord.

Our first “minimal viable product” is this contest. We are purchasing the impact of EA Forum posts in a centralized and moderated way to create a controlled sandbox in which we can observe the consequences and course-correct with feedback from the community. This way, we plan to mitigate potential risks and downsides to impact markets.

If it goes well, we plan to progressively expand both the prize pool amount for each contest and the scope of the market to things beyond the EA Forum. More broadly, we envision a world where markets for public and common goods (with mechanisms such as impact certificates, impact stock, retroactive funding, quadratic funding, dominant assurance contracts, Harberger taxes, etc.) play a significant role in the creation and funding of these goods.

Below are the instructions for the contest. If you feel confused, there are examples and analogies below the instructions.

Instructions

Please submit your post on app.impactmarkets.io.

If you want to save time, you can leave the justification (i.e. description) of your impact certificate empty and add it only if and when someone indicates interest in buying some fraction of it. Without the description it should only take a minute or two.

EA Forum posts from May and June are eligible. Posts submitted after June 30th will not be considered for this particular contest but may be eligible for future contests.

We value forum posts in proportion to how much we consider them morally good, positive-sum, and non-risky. (More on our criteria below.)

We have, for almost a year, thought a lot about risks from contests like these and from the markets that they may turn into. You can read more about our thinking in the post Toward Impact Markets.

We welcome any feedback you might have, positive or negative! Here is an anonymous feedback form that you can use. Comments on this post are of course welcome too.

Please only submit EA Forum posts. For example, if your post discusses a project that you’ve launched, then please make it clear in your certificate text that the certificate is for the Forum post only (a similar purview to copyright) and does not extend to your project.

If we buy some percentage of the impact of your post you will be required to edit your EA forum post and put a note at the bottom linking to your certificate on our website (for future tracking purposes).

You can give some percentage of the impact (and potential sale price) to someone else you collaborated with (coauthors, editors, proofreaders, etc.). Make sure to discuss and agree on the shared percentage before submitting your forum post for sale. We will not facilitate this sharing, you will have to coordinate between yourselves.

Don’t take advance monetary investments or make promises of future profits or you might get in trouble with your local securities law. Ask us if you're unclear whether this is a concern 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: Resource for criticisms and red teaming, published by Lizka on June 1, 2022 on The Effective Altruism Forum.

tl;dr: We've put together this post to share resources and tips that might be useful for people writing criticism or working on a red team of existing work.

We’re releasing this as a companion document to the announcement of the Criticism and Red Teaming Contest, but we hope this resource will be useful beyond that specific occasion.

We’re also hoping to grow and improve this resource after posting it, so please feel free to suggest additions and changes.

Why criticisms?

In brief, we think:

The people involved in the project of “effective altruism” are probably getting many things wrong. Noticing those errors (and course-correcting) matters for the project of actually doing the most good we can.

There are reasons to expect good criticism to be under-supplied relative to its value.

Many people feel tense and stressed about writing something critical, and we hope this resource will help them.

You can see more of our rationale here.

Existing resources

Note that some of the links here appear again below. If you know of other resources, please let us know!

Guides and collections

Relevant topics on the Forum

Red teaming

And other useful topic pages: Epistemic deference, Criticism of effective altruism, Criticism of longtermism and existential risk studies

And on LessWrong: Epistemic Review, Epistemic Spot Check, Disagreement

Red Teaming Handbook, 3rd Edition

A particularly useful excerpt starts on page 71: “Argument Mapping”

Outline

Chapters 1-2, 7: Introduction

Chapters 3-6: Mindset

Chapters 8-9: Tools

Note: we’re happy to pay a $200 bounty for notes on the Red Teaming Handbook that we’d like to add to this post.

Minimal-trust Investigations (Holden Karnofsky)

Scout Mindset (Julia Galef — you can get a free copy here, find a review here, and see a short clip discussing the core points here)

Slides from a presentation that Lizka, Charlotte Siegmann, Joshua Monrad, and Fin Moorhouse gave at EA Global: London 2022.

Assorted other writing

Some writing on criticism that we appreciate.

Arguing without warning (Dynomight)

I think it's great to have your work criticized by strangers online (Andrew Gelman)

Negativity (when applied with rigor) requires more care than positivity (Andrew Gelman)

Six Ways To Get Along With People Who Are Totally Wrong (Rob Wiblin)

Challenges of Transparency (Open Philanthropy), from Notable Lessons

Butterfly Ideas (Elizabeth van Nostrand)

Epistemic Legibility (Elizabeth van Nostrand)

The motivated reasoning critique of effective altruism (Linch Zhang)

Noticing the skulls, longtermism edition (David Manheim)

Liars (Kelsey Piper)

Models and types of criticisms

Some useful resources:

Four categories of effective altruism critiques (Joshua Monrad)

Minimal-trust Investigations (Holden Karnofsky)

Epistemic spot checks (Elizabeth van Nostrand)

Minimal trust investigations

A minimal trust investigation (MTI) involves suspending your trust in others' judgments and trying to understand the case for and against some claim yourself, ideally to the point where you can keep up with experts and form an independent judgment.

In his post about MTIs, Holden Karnofsky explains that they are, “not the same as taking a class, or even reading and thinking about both sides of a debate”. They are more thorough and skeptical, interrogating claims as far down as possible.

MTIs are useful when you identify some claim in EA that seems crucial to you, but hasn’t been scrutinized enough, and you’re genuinely unsure about its truth. When you embark on an MTI, you shouldn’t feel confident about the outcome: that’s the point!

For the contest: If you conduct something like an MTI focused on a claim you’re unsure about, and you end up agreeing with the claim, that still counts a...

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: Power dynamics between people in EA, published by Julia Wise on June 1, 2022 on The Effective Altruism Forum.

In any field, including EA, power differences affect how people interact.

Sometimes power differences are obvious, like between a manager and their report or a grantmaker and grantee. Organizations should have policies around some things here, like on conflicts of interest within an organization.

But there are a bunch of smaller power differences that subtly affect a lot of interactions. Formal policies would be too heavy-handed here, but we can all adjust how we handle these situations. I've tried to lay out some situations to be aware of, and steps that might help.

This post is meant for people who have relatively more power within EA — though who that is changes with the circumstances!

EA is complicated

EA is both a social community and a professional field. In the words of one HR person at an EA organization about handling this dynamic, “Boy, is it complicated and strange.”

For many of us who work in EA, it will be well worth it to have friendships, housemates, and/or partners within the community, not just our professional contacts. But that also introduces more complexity, and we should expect to need to tread carefully.

Some factors in how people influence each other

Funding

Being an actual grantmaker

Having grantmakers in your network who listen to you

Being a major donor to an organization

In organizations

If you work at or are on the board of an EA organization, you can affect the outcome of other people’s interactions with the organization. This affects hiring, grantmaking, etc even if you aren’t directly involved in hiring or funding. Your impression of a colleague can matter for promotion and other decisions, even if you’re not directly managing them.

Network

Who you can introduce people to, who you’re friends with, etc.

Gatekeeping

Ability to get people in or keep them out of events, retreats, office spaces, parties, etc.

Age

This won’t be true of everyone, but my impression is that people under about 25 often weigh others’ opinions more heavily and have a harder time expressing clearly when they’re uncomfortable with something.

History

If they looked up to you when they were first getting involved in EA, you may still have a kind of influence over them even if you’re now peers in other ways.

Demographics and culture

Some of this is big obvious categories like race and gender, but some is subtler things like accent, sharing more background knowledge with the cool kids, etc. People from cultures with higher power distance may be more reluctant to question authority figures.

How this affects things

Relationships that feel equal aren’t always equal

If I ask a community member for a favor, I don’t feel like I’m pulling rank or anything. But because of various advantages I have over them, their perception of the interaction will probably be slanted by those things. So maybe they’ll go along with things that are more costly for them than I intended.

Possible steps: Ask fewer favors from people who might feel uncomfortable refusing. Try to make clear when things are truly optional, and provide easy ways to decline.

It’s harder for junior people to give their real opinion

Opinions that junior people express will often be shaped (maybe unintentionally and unknowingly) by what they perceive senior people’s opinions to be or what they think the senior people want to hear.

Possible steps:

Ask junior people to share thoughts before you give yours. Listen openly and show interest.

Give them more time and encouragement to lay out messy thoughts.

Set the stage for welcoming messy / unformed thoughts by sharing some of your own.

Give encouragement and appreciation when they offer opinions and especially when they disagree with you - make it a good experience 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: Announcing a contest: EA Criticism and Red Teaming, published by Lizka on June 1, 2022 on The Effective Altruism Forum.

Introduction

tl;dr: We're running a writing contest for critically engaging with theory or work in effective altruism (EA).

Submissions can be in a range of formats (from fact-checking to philosophical critiques or major project evaluations); and can focus on a range of subject matters (from assessing empirical or normative claims to evaluating organizations and practices).

We plan on distributing $100,000, and we may end up awarding more than this amount if we get many excellent submissions.

The deadline is September 1, 2022. You can find the submission instructions below. Neither formal nor significant affiliation with effective altruism is required to enter into the contest.

We are: Lizka Vaintrob (the Content Specialist at the Centre for Effective Altruism), Fin Moorhouse (researcher at the Future of Humanity Institute), and Joshua Teperowski Monrad (biosecurity program associate at Effective Giving). The contest is funded via the FTX Future Fund Regranting Program, with organizational support from the Centre for Effective Altruism.

We ‘pre-announced’ this contest in March.

The rest of this post gives more details, outlines the kinds of critical work we think are especially valuable, and explains our rationale. We’re also sharing a companion resource for criticisms and red teams.

How to apply

Submit by posting on the EA Forum and tagging the post with the contest’s tag, or by filling out this form.

If you post on the Forum, you don't need to do anything except tag your post with the “Criticism and Red Teaming Contest” topic, and we’ll consider your post for the contest. If you’d prefer to post your writing outside the Forum, you can submit it via this form — we’d still encourage you to cross-post it to the Forum (although please be mindful of copyright issues).

We also encourage you to refer other people’s work to the contest if you think more people should know about it. To refer someone else’s work, please submit it via this form. If it wins, we may reward you for this — please see an explanation below.

The deadline is September 1, 2022.

Please contact us with any questions. You can also comment here.

Prizes

We have $100,000 currently set aside for prizes, which we plan on fully distributing.

Prizes will fall under three main tiers:

Winners: $20,000

Runners up: $5,000 each

Honourable mentions: $1,000 each

In addition, we may award a prize of $100,000 for outstanding work that looks likely to cause a very significant course adjustment in effective altruism.

Therefore, we’re prepared to award (perhaps significantly) more than $100,000 if we’re impressed by the quality and volume of submissions.

We’re also offering a bounty for referring winning submissions: if you refer a winning submission (if you’re the first person to refer it, and the author never entered the contest themselves), you’ll get a referral bounty of 5% of the award.

We will also consider helping you find proactive funding for your work if you require the security of guaranteed financial support to enable a large project (though we may deduct proactive funding from prize money if you are awarded one). See the FAQ for more details.

Submissions must be posted or submitted no later than 11:59 pm BST on September 1st, and we’ll announce winners by the end of September.

Criteria

Overall, we want to reward critical work according to a question like: “to what extent did this cause me to change my mind about something important?” — where “change my mind” can mean “change my best guess about whether some claim is true”, or just “become significantly more or less confident in this important thing.”

Below are some virtues of the kind of work we expect to be most valuable. We’ll look out 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: What YouTube channels do you watch?, published by Bella on June 1, 2022 on The Effective Altruism Forum.

TL;DR Please fill in this form! (<2mins)

At 80,000 Hours, we’re trying to reach more people with our content.

One way we’ve tried doing this is by partnering with YouTube creators and getting them to talk a bit about us on their channel. You can see examples of this here, here and here.

The main question we have when doing this is ‘which YouTube creators should we partner with?’

An important consideration when answering that question is how similar their audience is to our target audience — are they likely to be the kinds of people who find 80,000 Hours’ content useful, or not?

If a YouTube creator is being watched by people who we know are in 80,000 Hours’ target audience, then that’s a pretty good indicator that their audience in general is likely to be a good fit. If you’re reading this, you’re most likely in 80,000 Hours’ target audience.

So: what YouTube creators do you watch? Whose content do you especially like, or think is very high quality?

Don’t try to filter in advance for something like ‘does 80,000 Hours want to work with this creator’ — it’d be most useful if you just tell me who you in fact watch, the more regularly the better. Thank you!

(The form also asks about other media you’re into, like podcasts you listen to, or email newsletters you receive. Info on that would be helpful too, but I’m primarily interested in YouTube right now!)

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: Stuff I buy and use: a listicle to boost your consumer surplus and productivity, published by Aaron Bergman on June 1, 2022 on The Effective Altruism Forum.

Note: this post was originally written for my blog, which may be why some parts might not appear to have been written for the Forum. I may publish a more anecdote-rich, freewheeling version there soon, in which case I'll make this a linkpost.

Thanks to Martin Glusker for his detailed, very helpful thoughts and feedback.

Intro

This post is my attempted contribution to the blogosphere’s “things I recommend you buy and use” micro-genre.

To give due credit to my intellectual forefathers: Sam Bowman's archetypal founding text, published in 2017, has since catalyzed iterations by Rob Wiblin, Michelle Hutchinson, Megan McArdle, Arden Koehler, Arden again, Julia Wise, James Aung, Rosie Campbell, Michelle Hutchinson, David Megans-Nicholas, Ben Schifman, Yuriy Akopov, Yuriy again, Sam himself again, and surely many others too in the five years since.

Those articles are generally cogent, succinct, easy to read, and well-organized. This one is not. While writing, some of my “recommendations” metastasized into needlessly elaborate product reviews or gratuitously heterodox hot takes. It's also a bit heavier on the dietary supplements than is typical.

So, without further ado, sit back and enjoy this hypertrophied smorgasbord of “product recommendations.”

Part 1: Non-consumables

2021 MacBook Pro

(An accidental mini-review)

14" model starts at $1,999.

16" model (which I have) starts at $2,499.

Good condition used versions go for $100-$500 below retail on eBay, Craigslist, and Facebook Marketplace.

After experiencing some buyer's remorse with the smaller 14" screen, I up selling this laptop and buying a used 16" version, each for ~$500 below retail.

You can also spend up to 1 Trillion USD $6,599 for all the bells and whistles.

I know, an expensive device purchased by around 0.3% of all humans on Earth in 2021 isn’t really what the “stuff you should buy” micro-genre is all about.

Well that's too bad, because I really like this computer and three of its features in particular:

1) Bright screen

This is the feature I care about and appreciate most. Apparently some people are perfectly satisfied with dimmer screens. If that were me, I probably wouldn't have gotten this computer. All 2021 MacBook Pros reach 500 nits (brightness/surface area) for everything, and 1,000 or more for some content. For reference:

Older Mac laptops max out at 400 nits (brightness/surface area), and I regularly wished my 2017 Air would get brighter.

Most inexpensive (and some pricier) monitors reach 250.

Most name-brand but less expensive models do even worse, like this HP at 220.

New Dell laptops (e.g. this one for $1,449) can reach 500, but at this point you're not saving a huge amount of money compared to Apple.

Honestly, I’m not sure why Apple doesn’t make a bigger deal out of this; it’s arguably the only feature that matters every second you’re using the device and the new line of MacBooks are significantly brighter than almost every other make and model. To self-plagiarize some Tweets from a few months back:

And ~95% of the time I keep the screen on full blast, churning electrons into a delightfully vociferous stream of electrons.

2) A ~luxurious~ amount of storage

For an additional $2,400, you can get a Pro with 8TB of storage, enough to contain God Himself hold many large files and applications. But you can probably save your money for the other items on this list, because the base 512GB capacity is really quite a lot.

For reference, this was my storage summary after importing every file accumulated throughout high school and college and adding a bunch more:

3) Good battery life

In short, it’s nice to be be able to carry the thing around more or less al...

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 Brain That Builds Itself, published by Jan on May 31, 2022 on LessWrong.

Previously in this series: How to build a mind - neuroscience edition, Serendipitous connections: applying explanations from AI to the brain, The Unreasonable Feasibility Of Playing Chess Under The Influence, A Brief Excursion Into Molecular Neuroscience.

TL;DR: Shameless advertisement for a paper some colleagues and I wrote. But also some pretty pictures of brain development, and some first principle arguments for why it's difficult to build a brain.

Constraints make creative

Originally, my motivation for getting into neuroscience was: "Building artificial intelligence is difficult, so perhaps I can learn something from looking at how the biological brain is built. Once I've figured that out, AI will be easy".

(When I read these words I can't help but add a mental "duuuuh".)

Since those innocent days, I have come to appreciate that building a biological brain is a ridiculously overconstrained problem.

First off - even before you're done building, the brain must make itself useful. And it's not just "occasionally helping with an arithmetic problem" - I'm talking about "breathing", "eating", "coordinating limbs"... If the environment is dangerous, then also "running away" or "fighting" is on the menu. And you need to do these things reliably - lapsing means death.

Second - every neuron only has access to local information. In particular, every neuron only receives input from a limited number of other neurons (usually < 10,000 out of ~80 billion). From that information, a neuron needs to infer what its role in the developing brain ought to be. Consequently, every solution you come up with must be "symmetric" in the sense that you can swap two neurons, and the end result must still work. (It turns out the two neurons don't even have to be from the same species).

Third - there is no clear separation between "the thing you're building" and "the thing that is doing the building". The brain is making itself with only rough guidance from genes and the environment (see this post I wrote on the topic a while ago).

In contrast, the solution for AI appears to be (for now) "just stack more layers":

In AI, "the thing being built" and "the thing that's doing the building" is cleanly separated into "model" and "optimizer". Global information like the learning rate is readily available to all units. And even after training, there are no predators that punish lack of robustness.

But constraints make creative, and it seems like the aforementioned constraints turn out to have their benefits:

First - An adversarial environment forces the brain to be robust against perturbations. Both theory and practice tell us that with robustness comes the ability to generalize to novel domains. (Intense adversarial dynamics might, in fact, explain the huge gap between monkeys and humans in terms of generality.)

Second - A neat consequence of locality is modularity. When you try to use as little "wire" as possible to connect your neurons, locally interconnected modules tend to emerge. Beyond making things slightly more interpretable, modularity can also increase some types of robustness.

Third - ... This one beats me.

What is the benefit of having to build yourself? I'm not sure. Let's zoom in on that.

Bootstrapping brains

What does it mean for the brain to be building itself?

As the brain is growing, it is dynamically responding to cues. Removing input to one eye and the visual cortex balances the loss by upregulating input from the other. Adding an additional eye results in... well, this:

When the brain does this type of rearranging, it is both the thing that "notices" that something is wrong, and it's also the thing that "is being modified".

Third-eye shenanigans are only one (illustrative) example of the dynamic nature of ...

Welcome to The Nonlinear Library, where we use Text-to-Speech software to convert the best writing from the Rationalist and EA communities into audio. This is: New Yorkers: take <5m to help reimburse kidney donors!, published by abierohrig on June 1, 2022 on The Effective Altruism Forum.

Background:

The Living Donor Support Act is currently being considered by the New York State Assembly, and even though it has universal support, it might not be brought to the floor for a vote. If (and only if) you’re a New Yorker, emailing your assembly-member now can encourage them to vote on the bill before the assembly session closes this Thursday.

The bill would reimburse kidney donors for non-medical costs associated with surgery such as wage reimbursement — the median costs are as high as $4,000. There is good reason to believe that removing this disincentive would cause a non-trivial increase in kidney donations (see this study, esp. Table 2). If it increases living donations in New York by 20% (which is what a similar policy accomplished in New Zealand), that would mean over 100 additional donations per year. See here for a detailed breakdown of the benefits of receiving a kidney here for the recipient (though some of the figures are a bit out of date, living donor transplants probably last longer now).

In addition to being directly good for the world, this policy change would be a great win for the effective altruism community — Waitlist Zero, which is supported by Open Philanthropy, has led the advocacy for this bill, and others in the EA community helped it pass the NY Senate.

Call to action:

Find your assembly-member here

Click “Visit Assembly-member’s Page”

Click “Contact”

Send them an edited version of this email, the more personalized the better but even something very formulaic is helpful.

If you want to go above and beyond, you can also call your assembly-member — this is even better than emailing. The email template provides all the background you need for a phone call.

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