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: Cause Area: UK Housing Policy, published by GMcGowan on January 11, 2022 on The Effective Altruism Forum.

Summary

I allocated just under £30k of my 2021 giving to improving UK housing policy and plan to give more in 2022.

TL;DR of my case for this:

It seems plausible that better housing policy could boost the UK's GDP by more than a fifth, with all the attendant benefits of extra economic growth.

The current government has some interest in doing this, but needs more pressure to do it and good policy ideas for how to do it.

There are a couple of organisations I think are doing particularly good work in this area in that they're focusing on what I think the key problem is (housing supply) and have good ideas for how to solve the problem.

I think that a few tens of thousands of pounds could make a big difference here

Introduction

We should have a high bar for funding interventions to help people in rich countries in the short to medium term. Being wealthy, their citizens can afford much higher rates of consumption and their governments can spend at much higher rates, so many of their strongest needs are met. Giving money directly to these relatively wealthy people is not as effective as giving to people who are roughly x100 poorer. Further, there are interventions ~ x10 as good as giving money to these poorer people. This leads to Open Philanthropy arguing that EA interventions in rich countries should seek to meet the x1000 bar.

It is difficult to find causes that meet this standard (though the x100 bar for simple cash transfers is easier to meet and arguably more relevant for the long term).

I think changing housing policy in rich countries is a plausible cause that meets this bar, particularly in areas with the highest costs of housing and highest productivity. Here is a rough cost-effectiveness model that suggests housing policy could meet the bar.

I think that the UK’s housing policy is among the worst of all rich countries, and to date has not received the funding or attention that US housing policy has. Founders Pledge commissioned a great report on this area, focussing on the UK. In contrast, Open Philanthropy has already put several million dollars into funding this cause in the US.

The problem

Others have summarised this far better than me. Skip to “One Solution” if you read any of those links, otherwise read on for my take.

House prices have been rising substantially in many developed countries:

For the headline price of a house, the proximate cause of this is easy access to credit caused by low interest rates. The basic story is that low interest rates allow people to pay higher prices for goods by borrowing more, as the cost of that borrowing is low.

Interest rates are not the full story. We do not see the prices of other goods rising substantially with greater access to credit. Cars for example have not seen the same level of price increase because as prices increase, more cars can be manufactured to meet the level of demand. So interest rates do lead to increases in price, but only if supply of a good is limited. For housing, increasing prices have not led to corresponding increases in supply due to remarkably low price elasticity in many countries. Interest rates also do not account for the variation in the rate of house price growth between regions in the same country.

A better metric for housing costs is simply rent price. Here too we see large increases in real terms (graph UK only):

Again, this is not what we would expect if interest rates were the full story. Lower cost of credit should make it cheaper for landlords to pay their mortgages, which should lower rents.

As we see in the figure above, and as argued most forcefully here, as a percentage of income, rents are decreasing, so in some sense things are getting more affordable. I think this is tr...

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: Calibration proverbs, published by Elmer of Malmesbury on January 11, 2022 on LessWrong.

Cross-post from Telescopic Turnip.

Every problem is a calibration problem. That’s why most advice is basically useless: it tells you what to do, but doesn’t tell you when to stop. Therefore, the best pieces of advice are calibration advice. What we need is a metric to know if we are doing well, or if we should change our habits (and in which direction). For example, here are two poor pieces of advice:

"You should spend more time reading blogs, because compared to traditional media, bloggers have more freedom to communicate in original ways, and are more accountable when they say something false."

"You should spend less time reading blogs, because keeping up with a big pile of subscriptions takes a lot of time and makes you anxious."

Neither of these is really useful. Instead, we can produce one piece of calibration advice:

"After catching up with your backlog of blog posts, ask yourself what you remember out of it. If you can’t say what half of the posts were about, you should probably clean up your subscriptions."

Of course, that doesn’t sound like the timeless advice my grandmother received from her grandmother. To make it compelling, we need to make it rhyme:

"Before you subscribe to another blog, check what you took home from your backlog."

Now that sounds like ancient wisdom.

Thereafter is my humble attempt at writing calibration proverbs. As a non-native, English’s pronunciation is still a mystery to me, so the rhymes might be a bit wonky. Anyways, may these proverbs contribute to making 2022 a better year than 2021.

"He who thinks a claim by evidence is backed, can he predict the effect size just from the abstract?"

"Only when you leave a long conversation, you can tell if it warranted your apprehension."

"If you don’t enjoy the taste of tea, seek how to infuse it properly."

"Always check for bikes when you get out of your car. Always."

"When checking the source teaches you something new, the source of the source you should check too."

"One hour with your kids is worth two hours online, according to Han et al twenty twenty one."

"Man whose roommates wear t-shirts in winter, would better calm down with the goddamn heater."

What other calibration proverbs should we transmit to our grandchildren?

Thanks to Justis for the proofreading and rhyme consulting.

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: Sustained effort, potential, and obligation, published by AronM on January 10, 2022 on The Effective Altruism Forum.

This post is inspired by [Aiming for the minimum of self-care is dangerous] written by the EA forum user tessa. I felt very seen. With this post I attempt to add to its ideas. I recommend you read tessa’s post first but I don't think it is necessary to understand what I am exploring here.

Summary: This post describes a few of my thoughts around effort, optimization and obligation. It starts with a simple runner analogy that introduces the concept of facilitative stress. Then I share some personal considerations and anecdotes. I expect most value to be in the chapter “Problems that arise when you aim at the facilitative stress zone”.

After drafting this post I have checked and found that Lynette Bye has actually done research and has written about very adjacent ideas here and here. At the end of the post I have included a list of more relevant readings.

Disclaimer: I am no expert on mental health. These are not recommendations. I expect my thinking about this topic to change over time. Additionally, things I describe here might not relate to you at all.

Analogy of a runner

Consider a hypothetical me, that is in relatively good shape and who can run 5 km every day.

One day I think: hey, maybe I can achieve more. I will now start running 7 km every day.

And it works. The next day and the day after that I run 7 km. But on the third day I feel tired and exhausted. My body sends me signals of distress or fatigue, such as muscle pain, problems while breathing etc.

So I realize I probably pushed myself a bit too far and I decide to run just my usual 5 km today. Because I can totally do them - right? I am running and I am struggling already at km 2. I manage km 3 and have to really push myself. At km 4 I run out of energy. I stop.

I am dumbfounded. What is happening? Usually I can always do 5 km. Why not today? And I start reflecting on what kind of signals I have been getting from my body. And I realize that on day 2 of running 7 km there were already some of the signals (muscle pains and challenges while breathing) but I just ignored them.

I start tracking my distances and signals. I am careful and systematic this time. So after a rest day and being back to 5km a day I start increasing the distance in 0.5 km increments and I intensely monitor my body for signals. And after 2-3 weeks of experimentation I find out that I am actually a 5 km runner on average. For every day that I run, say, 6 km I will have to run 4 km the next day if I want to do 5 km again the day after that. This is a sustainable level of effort for me.

I also discover that if I run 7 km in one day (which I am capable of!) my total average over the week actually decreases! This is because when I push myself to run 7 km in one day, I need more rest to recover than I do on days where I stick closer to my daily average of 5 km. This extra rest is enough that the total distance (productivity) decreases. I have extra potential that comes at a cost. Good to know. An important data point!

There seems to be an elastic range above my average that doesn't affect my total weekly kms (35 km total (7 days x 5 km per day)) as long as I include the elasticity in my planning over the next few days. Say, if I go 6 km on day 1, I can do 4.5 km and 4.5 km on day 2 and 3.

Sustainable Effort Diagram

Breaking up the analogy

By “daily distance” I mean anything that requires effort. That is, not just your day job but also other energy drains in your free time, such as researching, coding, active networking, cooking, doing taxes, cleaning your house, maintaining relationships, learning, studying etc. I know certain tasks are way harder and cost more energy and some tasks are super easy, barely an inconvenience. Some t...

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: Concrete Biosecurity Projects (some of which could be big), published by eca on January 11, 2022 on The Effective Altruism Forum.

Andrew Snyder-Beattie and Ethan Alley

This is a list of longtermist biosecurity projects. We think most of them could reduce catastrophic biorisk by more than 1% or so on the current margin (in relative[1] terms). While we are confident there is important work to be done within each of these areas, our confidence in specific pathways varies widely and the particulars of each idea have not been investigated thoroughly.

Still, we see these areas as critical parts of biosecurity infrastructure, and would like to see progress building them out. If you’d like to be kept up to date about opportunities to get involved, please fill out this Google Form.

Early Detection Center

Early detection of a biothreat increases the amount of time we have to respond (e.g. designing tailored countermeasures, using protective equipment, heading to bunkers, etc). The current approach for early warning of novel pathogens is severely lacking—it typically relies on a particularly astute doctor realizing that something is strange combined with negative tests for everything else. Existing systems are also almost exclusively focused on known pathogens, and we could do a lot better using pathogen-agnostic systems that can pick up unknown pathogens.

One concrete goal would be something simple where a small team of people collects samples from volunteer travelers around the world and then does a full metagenomic scan for anything that could be dangerous.[2] Even collecting and analyzing only 100 random samples per day could make a big difference in some scenarios, since that would mean we would still expect to catch things before they infect too large a fraction of the global population. We think that with the right team, this could be done with close-to-existing technology for less than $50 million per year.[3]

There are a handful of bottlenecks and a number of ways to decompose this problem. To get started on subproblems, one of us (Ethan) is working on a list of suggestions, which we will backlink here.

Super PPE

Most personal protective equipment (PPE) is not good enough. Things like masks and suits require training to fit properly, lack reusability, and are generally designed for routine uses rather than for the most extreme events. The small minority of PPE that is designed for extreme use cases (e.g. BSL4 suits or military-grade PPE) is bulky, highly restrictive, and insufficiently abundant—not the kind of thing you could easily put millions of healthcare/pharma/essential workers into if needed. It seems plausible that with good materials science and product design we could come up with next-generation PPE that is simultaneously highly effective in extreme cases, easy to use, reliable over long periods of time, and cheap/abundant.

One concrete commercial goal would be to produce a suit (and accompanying system) that is designed for severely immunocompromised people to lead relatively normal lives, at a cost low enough to convince the US government to acquire 100 million units for the Strategic National Stockpile.[4] Another goal would be for the suit to simultaneously meet military-grade specifications, e.g. protecting against a direct hit of anthrax.

PPE has the advantage of being truly ‘pathogen-agnostic’—we can stockpile it in advance of knowing what the threat is, in contrast to vaccines or many medical countermeasures. It is also ‘defensively stable’ in that physical barriers can’t be easily bypassed using pathogen engineering techniques (whereas many medical countermeasures might be defeated with some creative tinkering). See Carl Shulman’s post here for more on this.

To get started on subproblems within PPE, one of us (Ethan) will publish a PPE deeper dive 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: Activated Charcoal for Hangover Prevention: Way more than you wanted to know, published by Maxwell Peterson on January 10, 2022 on LessWrong.

Activated charcoal is charcoal that has been heated up a whole lot and then washed really hard, by water or acid. This process is what "activates" it, and activating it increases the surface area by a lot. I think it prevents hangovers.

I won't spend too much time in this post arguing that activated charcoal works. Instead, I'll mostly take for granted that it does, then take a tour of the various things that might cause hangovers, stopping at each to look at why activated charcoal might or might not be likely to reduce them. So even if you're sure activated charcoal is just a hokey health craze thing I've tricked myself into believing, this post still has lots of interesting hangover science you might not have heard about.

I used to suffer vicious hangovers pretty often - usually a headache and sometimes nausea - after nights of drinking with friends. This past spring, I started taking two 520mg capsules of activated charcoal before going to bed on these nights. Since then, I've had maybe 26 nights out with the level of drinking I'd expect to result in hangover, and had a headache twice and nausea just once. (Tiredness is still common). Based on past experience, without activated charcoal I would have had headaches 12-18 of these times and nausea 4-10 times.

I recently convinced a friend to try a capsule before going to bed after a late night, after she'd insisted for months that there was no way this would work. The next morning she sent me a rave review, saying she'd expected to feel terrible after the number of drinks she'd had, but just felt a little tired. She's tried it a couple more times since and it seems like it worked those times too. I had another friend try it once and he said he thinks there was a 60% chance it helped; another friend's husband says it worked twice and didn't work once. I originally tried the stuff after I found this article, by a woman who went to a bachelorette party. They were split into two hotel rooms, and she gave activated charcoal to the women staying in her room, but not to anyone in the other one. The next day, those in the other room were in much worse shape.

The idea that activated charcoal helps hangovers doesn't have much support in the first few pages of google results. If I'd stopped at those, I wouldn't think it does anything. But my personal experience with it is striking. So I'm confused. Let's look into it.

(A quick warning: Some sellers of activated charcoal say to take it every day. I don't think there's any reason to do this, and some people say it's bad to. Also, if you take medications, activated charcoal will make them less effective; this could be pretty bad, depending on how important the medications are to you. Don't take activated charcoal every day.)

Why might activated charcoal prevent hangovers?

I mentioned activated charcoal (charcoal for short) has a lot of surface area. From US Poison Control:

Activated charcoal is a special form of carbon that can bind other substances on its surface (adsorption). That is why it is used for water filtration. It is also used to adsorb drugs in the gut so the drugs don't enter the body. Activated charcoal is made by burning carbon-rich materials, such as wood, at very high temperatures to create charcoal. The resulting product is a black, odorless powder. The charcoal is then "activated" through a special process that makes lots of holes and crevices on the charcoal particles to increase surface area and available binding sites. One teaspoonful of activated charcoal has about the same total surface area as a football field!

They also share a story, back from the arsenic-drinking days of 1813, of a chemist drinking 5 grams (read: ...

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 IHME Report, published by Zvi on January 10, 2022 on LessWrong.

An IHME report came out over the weekend. I was pleasantly surprised by it contents. Its conclusions and projections broadly match what I said in the last Omicron post, with only relatively minor obvious nonsense, and seemed worth their own post to break down.

Headline Takeaways

Our models for the United States suggest that infections may be currently peaking at 6 million a day, with varied peaks by state. We expect more than 50% of the US population will be infected with Omicron in the next 6 weeks. The infection-detection rate (IDR) is declining, although shortages of testing may lead to more rapid decreases in the IDR than we currently model. Nevertheless, we expect reported cases to exceed 1 million a day, peak by the third week of January, and then decline sharply.

I too expect reported cases to peak during the third week of January (to be clear, when I say we will peak on 19 January, I mean cases detected on 19 January, not infections that happen on 19 January. If counting from initial infections, I’d have said 16 January, but it’s hard to ever know what the actual lag was there.)

Once the peak arrives, a sharp decline is to be expected shortly thereafter, although I still expect cases to remain elevated (above pre-Omicron levels) for several months.

Note that ‘more than 50%’ sounds like the kind of thing one says when one does not want to make a precise prediction, but there are graphs later on.

Hospitalizations will increase to a peak that may be twice as high as last winter. These figures, however, include incidental admissions. Because the prevalence of Omicron infection is so high, many individuals hospitalized for other conditions will have asymptomatic infections. Incidental admissions may exceed 50% of total COVID-19 admissions in some states. Daily deaths will increase but remain below 2,000 a day, peaking by the end of the month. This means the peak of deaths will remain below the summer Delta surge and well below the peak last winter.

That sounds like a best case scenario. If hospital admissions peak at twice last winter, but half of them are incidental, then things won’t be substantially worse than last winter. We survived last winter. Deaths peaking below 2,000 per day means they won’t even double once, and the peak will come within a few weeks. Expecting them to peak by month’s end seems wrong given the historical lag times, but it’s only a difference of a few weeks.

Words of high wisdom follow, noting that all of this is baked in.

Our alternative policy scenarios, including more rapid scale-up of boosters to all who have been previously vaccinated, increasing mask use to 80%, and vaccinating the partially hesitant, have only a small impact on the trajectory over the next 4 months. The speed of the epidemic is so fast that policy interventions will have little impact. In previous waves, the control strategy has been to control infection and thus reduce hospitalization and death. Given that there is little prospect of controlling infection, strategies need to focus on reducing harm in the vulnerable and minimizing disruption.

It is essentially too late to do anything but mitigation. If the baseline scenario is as they describe (and roughly as I predict as well) then only those with a strong desire to avoid infection should do much. So this, very much this:

Given the massive numbers of infections in the community, testing and quarantining asymptomatic individuals may not be helpful. There appears to be no prospect for controlling transmission and considerable prospect for disruption of schools and essential services due to screening. States may need to consider revisions to their testing and quarantine strategies.

Quite right. Doing constant testing of asymptomatic schoolchildren ma...

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: Bryan Caplan on EA groups, published by Aaron Gertler on January 10, 2022 on The Effective Altruism Forum.

Crossposting because it seems good to share sincere reviews of our community. If the post had been about how terrible EA groups were, I'd have crossposted it anyway.

Kudos to whichever student groups actually interacted with Bryan and inspired this post!

As a professor and public speaker, I’ve spoken to a wide range of student groups. On reflection, my very favorite turns out to be: Effective Altruism. Indeed, I’ve had positive experiences with 100% of the EA groups I’ve encountered.

What’s so great about the Effective Altruists? They combine high knowledge, high curiosity, and high iconoclasm. When I ask EAs if they’ve heard of signaling, or the Non-Identity Problem, or pollution taxes, most of them say Yes. The ones who say No are eager to get up to speed. And if I defend a view that would shock a normal audience, EAs are more likely to be amused than defensive or hostile. They’re genuinely open to reasoned argument.

Though you might expect EAs to be self-righteous, they’re not. EA is a chill movement. While ethical vegans are greatly overrepresented in EA, they’re the kind of ethical vegans who seek dialogue on the ethical treatment of animals, not the kind of ethical vegans who seek to bite your head off.

Most EAs are official utilitarians. If they were consistent, they’d be Singerian robots who spent every surplus minute helping strangers. But fortunately for me, these self-styled utilitarians severely bend their own rules. In practice, the typical EA is roughly 20% philanthropist, 80% armchair intellectual. They care enough to try make the world a better place, but EA clubs are basically debating societies. Debating societies plus volleyball. That’s utilitarianism I can live with.

Why do I prefer EA to, say, libertarian student clubs? First and foremost, libertarian student clubs don’t attract enough members. Since their numbers are small, it’s simply hard to get a vibrant discussion going. EA has much broader appeal. Anyone who likes the idea of “figuring out how to do the most good” fits in. Furthermore, to be blunt, EAs are friendlier than libertarians – and as I keep saying, friendliness works.

Furthermore, while the best libertarian students hold their own against the best EA students, medians tell a different story. The median EA student, like the median libertarian student, like almost any young intellectual, needs more curiosity and less dogmatism. But the median EA’s curiosity deficit and dogmatism surplus is less severe.

The good news is that most EA clubs already contain some libertarians. And the best way to improve both movements is for the libertarians to regularly attend EA meetings. It’s a great chance to spread superior libertarian logos while absorbing superior EA ethos.

When I last spoke at the University of Chicago, one student defended education as a crucial promoter of social justice. In response, I argued that Effective Altruism is what the social justice movement ought to be. EAs know that before you can make the world a better place, you must first figure out how to make the world a better place. This in turn requires you to prioritize the world’s problems – and calmly assess how much human action can remedy each of them. Social justice activists imagine that these questions are easy – and as a result their movement has become one of the world’s major problems. Probably like the twentieth-worst problem on Earth, but still.

Perhaps the main reason why I get along so well with EAs is that their whole movement rests on a bunch of my favorite heresies. First and foremost: Good intentions often lead to bad results. EA exists because many good things sound bad, and many bad things sound good. The very existence of their movement is an 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: Rowing and Steering the Effective Altruism Movement, published by jtm on January 9, 2022 on The Effective Altruism Forum.

Acknowledgements

lI am grateful to Charlotte Siegmann, Jessica McCurdy, Carla Zoe Cremer, Janvi Ahuja, Jonas Sandbrink and Kaleem Ahmid for conversations and feedback that significantly improved this post. Any errors, of fact or judgment, remain my entirely own.

Introduction

In a recent blog post, Open Philanthropy Co-CEO Holden Karnofsky introduces the analogy of “the-world-as-a-ship” as a framework for thinking about efforts to make the world better.

Under this analogy, if you are ‘rowing’ the ship, you are trying to “help the ship reach its current destination faster.” By contrast, if you are ‘steering’ the ship, you aim to “navigate to a better destination than the current one.” Alternative options include anchoring, equity, and mutiny. It’s a terrific post introducing a practical framework for thinking about world-improving efforts such as those undertaken by the effective altruism and longtermism communities.

I think we can apply the analogy of the ship in a narrower sense to effective altruism and longtermist movement-building efforts. In particular, I find that the notions of rowing and steering can elucidate some key differences in beliefs that I’ve recently observed in the community.

In this post, I briefly contrast what I see as the rowing and steering approaches to building the effective altruism movement and discuss success and failure from each perspective.

I present three arguments for why ‘steering’ is more important than is currently recognised: i) Given the momentum of the effective altruism and longtermism movements, steering is currently more neglected than rowing; ii) steering is more tractable if implemented sooner rather than later; and, iii) steering is valuable for community health.

I then offer concrete proposals for one form of steering that I’d like to see more of, namely good-faith critiques of the effective altruism philosophy and movement. Specifically, I suggest commissioning research on effective altruism and longtermism failure modes, offering Effective Altruism Forum prizes for good-faith critiques, funding, creating, and managing effective altruism ‘red teams’, and committing to reviewing and acting on valid critiques.

Defining rowing and steering

I’ll define ‘rowing for the effective altruism community’ roughly as ‘efforts to increase the power of the effective altruism movement.’ This can be achieved by recruiting or converting more people to support the cause, increasing the engagement of existing community members, obtaining more funds committed to the movement, creating powerful allies in government and industry, and so on.

By contrast, I’ll define ‘steering for the effective altruism community’ roughly as ‘efforts to improve the movement’s aims and/or to ensure that the movement executes on those specific aims.’ This can be achieved by researching global priorities, evaluating competing causes and interventions, critically examining and improving current cause-prioritisation methodologies, critiquing the direction of existing efforts, creating institutions and norms to enhance procedures and decision-making, and so on. Crucially, on my definition, steering isn’t just about pulling the ship in the direction you currently think is best. It’s also about questioning whether your navigational approach, crew composition, and equipment are adequate for the task. It’s about ensuring that, in the future, one can steer even better than the current crew, and the future crew is not hindered from pulling the ship in the direction they think is best.

For example, it’s not just about doing global priorities research; it’s also about continually and rigorously asking whether your way of doing such research has any blind s...

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: Use Normal Predictions, published by Jan Christian Refsgaard on January 9, 2022 on LessWrong.

Making predictions is a good practice, writing them down is even better.

However, we often make binary predictions when it is not necessary, such as

Biden win popular vote: 91%

Danish COVID deaths above 10.000 by January 1. 2022: 84%

Alternatively, we could make predictions from a normal distribution, such as ('~' means ‘comes from’):

Biden’s popular vote ~ N(0.54, 0.03)

Danish COVID deaths by January 1. 2022 ~ N(15,000, 5,000)

While making "Normal" predictions seems complicated, this post should be enough to get you started, and more importantly to get you a method for tracking your calibration, which is much harder with dichotomous predictions.

The key points are these:

Predicting from a normal is surprisingly easy.

Getting an actionable number for how over/under confident you are requires only simple math!

The normal distribution carries more information than the Bernoulli (binary outcome such as coins) and will therefore give you more information to act on!

Things this post will answer:

How do I make a normal prediction?

Why do I want to do this?

How do I track my calibration?

Quick recap about the normal distribution

The normal distribution is usually written as N(μ,σ) has 2 parameters:

a location parameter μ (pronounced mu) which is both the most likely and the average value

a scale parameter σ (pronounced sigma) which captures uncertainty, high σ implying high uncertainty

the 68-95-99.7 rule states that:

68% of your predictions should fall in μ±1σ

95% of your predictions should fall in μ±2σ

99.7% of your predictions should fall in μ±3σ

Finally 50% of the predictions should fall within 0.674≈23σ, which can be used as a quick spot check.

How to make predictions

To make a prediction, there are two steps. Step 1 is predicting μ. Step 2 is using the 68-95-99.7 rule to capture your uncertainty in μ.

I tried to predict Biden’s national vote share in the 2020 election. From the polls, I got 54% as a point estimate, so that seemed like a good guess for μ. For σ I used the 68-95-99.7 rule and tried to see what that would imply for different values of σ. Here is a table for 2-5%

Intervals. 68% 95% 99.7% σ=2. 52-56% 50-58% 48-60% σ=3. 51-57% 48-60% 45-63% σ=4. 50-58% 46-62% 42-66% σ=5. 49-59% 44-64% 39-69%

σ=2 implies a 97.5% (95% interval + half a tail) chance that Biden would get more than 50% of the votes; I was not that confident. σ=4 implies a 84% chance that Biden would get more than 50% of the votes (68% + 32%/2), and a 16% chance Trump wins, I likewise found this too high, so I settled on σ=3.

Why do I want to do this

Biden Got 52% of the vote share, which was within 1 sigma of my prediction. There are two weak lessons that I drew from this ONE data point.

The pollsters screwed up, so I should have regressed μ towards the mean (50%), such as predicting 53% instead of 54%

The prediction was exactly 23σ from μ, so the σ was on the 50%/50% boundary just as expected. This was lucky, but it's weak evidence that the σ was well chosen.

Imagine I instead had predicted Biden wins (the popular vote) 91%, well guess what he won, so I was right... and that is it. Thinking I should have predicted 80% because the pollsters screwed up seems weird, as that is a weaker prediction and the bold one was right! I would need to predict a lot of other elections to see whether I am over or under confident.

How to track your calibration

Note: In the previous section we used μ and σ for predictions. In this section we will use μi and σi where i is the index (prediction 1, prediction 2... prediction N). We will use ^σz for the calibration point estimate; this means that ^σz is a number such as 1.73. In the next post in this series, we will use σz for the calibration distribution, this means that σz...

Welcome to The Nonlinear Library, where we use Text-to-Speech software to convert the best writing from the Rationalist and EA communities into audio. This is: How to make Slack workspaces welcoming and valuable, published by Alex Berezhnoy on January 9, 2022 on The Effective Altruism Forum.

I’m a community builder at EA Anywhere and I use Slack daily as a communication tool. I helped to set up multiple EA workspaces in Slack, and I believe it can create a lot of value for your group members. In recent times, it has become an especially popular choice for professional and affiliation groups within EA as their number grows. But virtual workspaces should be set up wisely. Sometimes people underestimate all the time costs and responsibilities involved, and in some cases, I argue against creating them.

Most of the advice covered here is applicable if you’re using other tools like Discord, Zulip, or Mattermost.

What are the benefits your group members can gain from a Slack workspace?

I think a good workspace combines all of the following:

It's a place where people keep updated about the group's activities and ways to get involved in the EA. If you run a professional group, it makes sense to share opportunities within your field (e.g. Entrepreneurship, Mental Health, etc). You can share information about volunteering, internships, and post job adverts from EA-aligned organizations.

It's a place where group members can receive support and celebrate their successes, no matter how big or small they are. It can empower them to take more actions based on EA principles and feel better. But people are usually uncomfortable asking for help if other group members are complete strangers to them. That's why creating a friendly environment and fostering connections are important.

It's a place where people can share knowledge and learn from each other. Nothing engages people more with the group than the feeling they are creating value for others. So make sure that it's easy for everyone to contribute and encourage group members to do it.

It's a place where group members can connect with each other. It's especially challenging to foster bonds if the group operates only virtually, like many professional groups with members from all over the world. Fortunately, there are solutions that can help strengthen personal connections and build human relationships.

Concrete steps you can take

Create a welcoming message for everyone who joins your workspace

Encourage people to introduce themselves in the #introductions channel, share their background and interests.

Promote good discussion norms and include community guidelines into the welcoming message, but don’t make it too long.

Welcome message can be implemented with Slack Workflow, GreetBot, or “Get Started” page (Slack Settings & Administration → Customize → Get Started page → Craft a greeting).

Give group members the opportunity to connect for virtual coffee

It would increase the number of serendipitous connections made, let people meet a range of other EAs and swap ideas.

Multiple apps are created for this purpose, like Donut, Shuffl, and RandomCoffee. You can check them and figure out what works better for you, but an app doesn't matter as much as good promotion & clear instructions on how to use it.

Don't auto-add everyone to the channel. A high proportion of inactive members would discourage others from using the app. Make sure that only people who actually can meet are available for virtual coffee, encourage members to pause intros when they are not available.

Share an invite link wisely

Especially in the beginning, as the first workspace members will have a significant impact on the inner culture.

Invite people with good epistemic norms. It's better to invite a few people who share EA values which increase trust, rather than a dozen who will be inactive.

As Julia Wise mentions in her notes for moderating online spaces, it's easier to prevent someone from joining than to deal with misbehavior ...