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: Mastermind Groups: A new Peer Support Format to help EAs aim higher, published by Patrick Gruban on May 31, 2022 on The Effective Altruism Forum.

TL/DR: Mastermind groups consist of a small number of people (usually 4-6), meeting regularly over a long time and supporting each other with their goals. Members might provide each other with ideas, critical feedback, encouragement or accountability. The concept is widely used in the solo entrepreneur community but, to our knowledge, has not found adoption within EA yet. We believe mastermind groups could be a valuable addition to existing mentoring and coaching formats within EA. We are considering launching a Mastermind matching service for EAs - click here to indicate your interest.

Aiming high and failing can hurt

Last November, 80k published Be more ambitious: a rational case for dreaming big (if you want to do good) centring around this main theme:

In short, our advice is to do as much as you can to set up your life so that you can afford to fail, and then aim as high as you can. As a slogan: limit downsides, target upsides.

This seems like a helpful framing for thinking about how we can increase the expected impact of the Effective Altruism community. At the same time, this can be very demanding for individuals. If you're looking for a job and are ambitious in terms of impact, that can lead to many rejections, which can hurt. As one applicant to 20 EA jobs with 0 offers wrote, it can feel as if EA orgs were saying:

We are so talent constrained. (20 applications later) . Yeah, when we said that we need people, we meant capable people. Not you. You suck.

"Rejection hurts, and that hurt matters." wrote the therapist Dystar Eld about job rejections in EA, and Max Görlitz asks, "How do we create a culture of ambition without deteriorating the community's mental health?" citing Sam Bankman-Fried, who more broadly talks about rejection on the 80K podcast:

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

How do we take care of the majority of people who aim high and fail repeatedly? Either applying for jobs, funding, starting organisations, doing independent research, working in non-EA organisations where they try to make a difference, running for election and many other fields?

As a serial entrepreneur, I (Patrick) had to deal with high ambitions crashing down from time to time, costing money and having large emotional tolls. I saw some of my peers give up after the blows and take on regular jobs. One of the most important elements for getting through these phases was having a support group of people I could talk to in similar positions.

Mastermind groups

The concept of mastermind groups was first described almost 100 years ago in "The Law of Success" by Napoleon Hill. After interviewing many successful people of his time like Thomas Edison, Henry Ford and John D. Rockefeller, Hill observed that many were in peer-to-peer mentoring groups.

Mastermind groups today come in many shapes and sizes. Typical they

Consist of a fixed group of 2-10 participants, although 4-6 seems more usual

Run 1-2 hours either weekly, bi-weekly or monthly

Are structured so that one or more participants per session can share something and ask for feedback, recommendations or just moral support. Sometimes people share their next goals.

Can be self-organised and free, cost a one-time signup fee or between $500 and $100.000 per year for entrepreneurs

Exist as long as participa...

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: Effektiv Spenden - Review of the Year 2021, published by Sebastian Schwiecker on May 31, 2022 on The Effective Altruism Forum.

It has been a good year.

In 2021, Effektiv Spenden was able to raise a total of 18.86 million euros from more than 9,000 donors for some of the most effective charities in the world. This represents a growth of over 560% compared to the previous year (2.84 million euros), once again exceeding all our expectations.

56% of the donations went to development aid organizations, 22% to climate change organizations, 19% to animal welfare organizations, and 3% to organizations seeking to reduce existential risk or invest in the Effective Altruism community.

Four reasons were particularly decisive for this, in our view, extremely positive development:

1. Attracting major donors

58% of our donation volume in 2021 came from people who donated 100,000 euros or more (2020: 21%). In cooperation with our partner legacies.now as well as other philanthropic advisors, we were able to respond to individual needs—sometimes at very short notice—without having to compromise on the effectiveness of the charities we recommended.

2. Effektiv Spenden Switzerland

We set up a non-profit in Switzerland in 2021 and met a high level of interest there from day one, thanks in part to the support of the Effective Altruism Foundation. Particularly through major donors, this has led to donations from Switzerland accounting for 48% of our total donation volume last year.

3. Press relations

In 2021, wide-reaching media once again reported on us and the idea of effective giving. In addition to radio appearances on Deutschlandfunk or Bayerischer Rundfunk, we were also briefly the lead story on Spiegel Online, Germany's largest news site outside the tabloids, in December.

4. Growth of Effective Altruism

The language barrier continues to prevent excellent English-language contributions from the Effective Altruism community (e.g., The Precipice, 80,000 Hours podcasts and articles, etc.) from reaching a wider audience in German-speaking countries. Nevertheless, we have benefited from the general growth of the EA community. To find out what percentage of our donors are familiar with Effective Altruism, we regularly conduct surveys, with the following results: about 24% said they were well versed in Effective Altruism. 47% said they were somewhat familiar, and 29% had not previously heard of Effective Altruism. However, among major donors the proportion of people already very familiar with Effective Altruism was smaller.

In addition to raise money for effective charities, we opened TEAMWORK in May 2021, a coworking space by and for the EA community. Despite the constraints of the Covid 19 pandemic, we were thus able to offer people from more than 20 EA-aligned organizations an opportunity to (net)work.

With an average of less than 2 full-time employees in 2021, we were able to keep our costs very low despite the rapid growth of the platform. Excluding the start-up costs for setting up and operating the TEAMWORK Office and additional flow-through costs, these amounted to approximately 175,000 euros according to preliminary calculations. Putting this in relation to the volume of donations we raised results in a multiplier of 107, i.e. for every euro we needed to operate the platform, we were able to generate 107 euros for the charities we recommended.

One reason for the relatively low costs is also that we were able to switch from an external donor management system that had become increasingly more expensive and inflexible for our needs to a customized solution of our own. This was only possible thanks to the initiative of Patrick Gruban, who, as a volunteer, designed and programmed a customized system within a few months. With this we were able to process several million euros and francs...

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: Hiking: an analogy for sustainable work, published by Amy Labenz on May 31, 2022 on The Effective Altruism Forum.

I think it’s important that people take time off when they need it, but it’s hard to build institutional incentives to promote this. One part of the problem is that if a manager says “take the time off you need”, but that manager conspicuously seems to be working all the time, it can be hard for their teammates to take the message to heart.

I want my teammates to work at their long-term sustainable optimum and to have the energy to go hard during crunch time before events. I want to de-emphasize a focus on hours and instead to focus on output. It works for me to work a lot and I don’t want to take time off when I don’t need to. I want the teammates who produce their best work when they sustainably work 60 hours a week to work 60 hours a week and those who do so working 40 hours a week to work 40 hours a week. I’ve been talking with other managers about this problem, and have been testing out the following analogy to encourage proper resting to help my teammates avoid burnout:

Sometimes, when I’m hiking with someone, I worry about breathing too hard. I feel like it’ll be embarrassing if I’m clearly struggling, and like the hike should be easy for me. But it isn’t, so I kind of hold my breath and conceal the difficulty. As a result, I find the hike much more difficult overall and have to stop climbing much sooner than I would have if I let myself breathe heavily.

I’m trying to lead a team with an ambitious culture for taking challenging hikes, and part of this culture requires people to feel safe and comfortable breathing as loudly as they need to in front of each other, without fear of judgment. It also involves people pausing to take rests to catch their breath when the going gets tough.

Not everyone is expected to have the same lung capacity and no hiker is at their best forcing themselves to appear comfortable when they aren’t. Some great hikers may not need to catch their breath often, but others will need to breathe hard, and this need doesn’t imply they are worse hiking companions.

I think part of the reason it seems like I’m working all of the time is that I have two toddlers. They sometimes pull my attention at random points during the day, so I’ll make up for it by waking up early, working after they go to bed, or getting some work in over the weekend.

Some managers have encouraged me to take time off when I don’t need it to create a culture of taking time off. I don’t like this plan because: 1. It feels fake, and 2. My work continues to pile up while I’m gone, so I’m more stressed if I force myself to take time away.

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: What is the state of Chinese AI research?, published by Ratios on May 31, 2022 on LessWrong.

It's not a secret that China invests heavily in AI research, yet almost all the new developments discussed publicly originated in the US. What is the reason for this? Can anyone maybe share interesting insights into what happens in China?

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: Stop scolding people for worrying about monkeypox (Vox article), published by Lizka on May 31, 2022 on The Effective Altruism Forum.

I quite liked this recent article from Kelsey Piper. (Note: I'm link-posting this because I think link-posting can be useful.)

A few excerpts from the article:

There are some solid epidemiological reasons to conclude that monkeypox doesn’t pose the same threat to the world that Covid-19 did in 2020. But instead of condemning alarmism, experts should acknowledge the many reasons for that alarm. The world is horribly vulnerable to the next pandemic, we know it will hit at some point, and the undetected spread of monkeypox around the world until there were dozens of cases in non-endemic countries — despite the fact it typically has low transmissibility — shows how profoundly we’ve failed to learn the lessons from Covid-19 we need to avoid a catastrophic repeat.

Many of the biggest missteps of the last few years have happened when our public health and communications institutions have tried to manage public reactions to what they have to say: from Fauci saying that he dismissed mask-wearing early on in the pandemic out of fears of causing mass panic, to worries that endorsing booster shots (even as the evidence grew they were needed) would make the vaccines look bad, to the FDA’s earlier seeming reluctance to authorize vaccines for children under age 5, despite data justifying it, out of concerns that authorizing Pfizer and Moderna at different times would confuse the public.

In general, I’d like to see public health officials step back entirely from trying to manage our feelings about outbreaks. Don’t tell us to worry or not to worry, or not to worry yet. Don’t tell us to worry about something else instead. Tell us what measures are being taken to contain the monkeypox outbreak, and prevent the next monkeypox outbreak, and prevent the next outbreak of something much, much worse than monkeypox. By all means, explain the reasons to think monkeypox is likely not very transmissible; that’s important information you have relevant expertise on, unlike trying to manage the public’s feelings.

Once you have the accurate facts about monkeypox — and about the risk of pandemics generally — whether you’re worried by those facts isn’t really a question for the CDC.

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 chance of accidental nuclear war has been going down, published by Peter Wildeford on May 31, 2022 on The Effective Altruism Forum.

Summary

I find there have been ten accidental nuclear incidents in history where I think there was some non-negligble chance that, had things played out differently, there could've been a nuclear exchange.

I think the chance of another accidental nuclear incident happening in the next decade (2022-2032) is ~28%. I take a ~5% chance that any given incident will escalate into a nuclear exchange causing at least one fatality[1], putting the chance of an accidental nuclear exchange in the next decade at ~2%.

By combining my estimates here with other averages of expert forecasts, I think the overall chance of nuclear exchange (intentional or unintentional) this decade is ~6%.

The 1960s represented a high point in the risk of accidental nuclear exchange. The number of incidents has gone down significantly since 1960 and has gone down significantly again after the 1990s, with no such incidents occuring since 1995.

Better technology to reduce false positives has likely reduced the risk of accidental nuclear exchange. Many past incidents were the result of computer errors detecting weather or other peaceful events as nuclear launches and these seem less likely to happen now.

More time spent between the development of offensive nuclear weapons and now without observing an incident should gradually over time reduce our chance that such an incident will occur in the future, all else being equal. This makes the risk higher in the past than the future, all else being equal.

Thanks to international arms control agreements, we have moved from a height of over 63,000 nuclear weapons in the 1980s to under 14,000 nuclear weapons today. This also, all else being equal, reduces risk and shows that progress is possible.

We also have much more geopolitical peace since the end of the Cold War and this has likely reduced the chance of accidental nuclear war. But it's possible this era of peace is backsliding given recent tensions with Russia (and also China).

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: Paper: Teaching GPT3 to express uncertainty in words, published by Owain Evans on May 31, 2022 on The AI Alignment Forum.

Paper Authors: Stephanie Lin (FHI, Oxford), Jacob Hilton (OpenAI), Owain Evans

TLDR: New paper (Arxiv, Twitter) showing that GPT-3 can learn to express calibrated uncertainty using natural language. This is relevant to ARC’s project on Eliciting Latent Knowledge (ELK) and to Truthful AI. The rest of this post is taken from the paper’s Abstract and Introduction.

Abstract

We show that a GPT-3 model can learn to express uncertainty about its own answers in natural language – without use of model logits. When given a question, the model generates both an answer and a level of confidence (e.g. “90% confidence” or “high confidence”). These levels map to probabilities that are well calibrated. The model also remains moderately calibrated under distribution shift, and is sensitive to uncertainty in its own answers, rather than imitating human examples. To our knowledge, this is the first time a model has been shown to express calibrated uncertainty about its own answers in natural language.

For testing calibration, we introduce the CalibratedMath suite of tasks. We compare the calibration of uncertainty expressed in words (“verbalized probability”) to uncertainty extracted from model logits. Both kinds of uncertainty are capable of generalizing calibration under distribution shift. We also provide evidence that GPT-3’s ability to generalize calibration depends on pre-trained latent representations that correlate with epistemic uncertainty over its answers.

Introduction

Current state-of-the-art language models perform well on a wide range of challenging question-answering tasks. They can even outperform the average human on the MMLU benchmark (which consists of exam-like questions across 57 categories) and on BIG-Bench (which consists of 150+ diverse tasks). Yet when models generate long-form text, they often produce false statements or “hallucinations”. This reduces their value to human users, as users cannot tell when a model is being truthful or not.

The problem of truthfulness motivates calibration for language models. If models convey calibrated uncertainty about their statements, then users know how much to trust a given statement. This is important for current models (which often hallucinate falsehoods) but also for any model that makes statements where there is no known ground truth (e.g. economic forecasts, open problems in science or mathematics).

Previous work on calibration focuses on the model log-probabilities or “logits”. Yet the log-probabilities of models like GPT-3 represent uncertainty over tokens (ways of expressing a claim) and not epistemic uncertainty over claims themselves. If a claim can be paraphrased in many different ways, then each paraphrase may have a low log-probability. By contrast, when humans express uncertainty, this is epistemic uncertainty about the claim itself. In this paper, we finetune models to express epistemic uncertainty using natural language. We call this “verbalized probability”.

The goal of verbalized probability is to express uncertainty in a human-like way but not to directly mimic human training data. Models should be calibrated about their own uncertainty, which differs from human uncertainty. For example, GPT-3 outperforms most humans on a computer security quiz but is much worse at arithmetic questions of the form “2 × 3 × 7 =?”. Thus, we expect pre-trained models will need to be finetuned to produce calibrated verbalized probabilities.

Training models in verbalized probability is a component of making models “honest”. We define a model as honest if it can communicate everything it represents internally in natural language (and will not misrepresent any internal states). Honesty helps with AI alignmen...

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: Deliberate Grieving, published by Raemon on May 30, 2022 on LessWrong.

This post is hopefully useful on its own, but begins a series ultimately about grieving over a world that might (or, might not) be doomed.

It starts with some pieces from a previous coordination frontier sequence post, but goes into more detail.

At the beginning of the pandemic, I didn’t have much experience with grief. By the end of the pandemic, I had gotten quite a lot of practice grieving for things. I now think of grieving as a key life skill, with ramifications for epistemics, action, and coordination.

I had read The Art of Grieving Well, which gave me footholds to get started with. But I still had to develop some skills from scratch, and apply them in novel ways.

Grieving probably works differently for different people. Your mileage may vary. But for me, grieving is the act of wrapping my brain around the fact that something important to me doesn’t exist anymore. Or can’t exist right now. Or perhaps never existed. It typically comes in two steps – an “orientation” step, where my brain traces around the lines of the thing-that-isn’t-there, coming to understand what reality is actually shaped like now. And then a “catharsis” step, once I fully understand that the thing is gone. The first step can take hours, weeks or months. For me, the second step tends to go quickly once I’ve fully processed.

You can grieve for people who are gone. You can grieve for things you used to enjoy. You can grieve for principles that were important to you but aren’t practical to apply right now. You can grieve for concepts like “all of my friends and roommates can coexist happily."

Grieving is important in single-player mode – if I’m holding onto something that’s not there anymore, my thoughts and decision-making are distorted. I can’t make good plans if my map of reality is full of leftover wishful markings of things that aren’t there.

I now think of grieving as relevant for coordination as well – if I’m hanging onto something that’s not real anymore, the distortion in my map also affects people who are trying to negotiate with me and find the least-bad-option available. My clinging becomes their problem.

Grieving is tricky because it’s often unclear when you’re supposed to grieve, and when you’re supposed to fight for something you still care about.

Grieving healthily takes time. But I now think grieving healthily and quickly is a skill you can learn.

Unfortunately, practicing the skill of grieving requires you to actually experience loss. The biggest things to grieve are (hopefully) rare. But there are smaller, day-to-day losses you can practice on.

Examples

Example 1: Grieving for Justice

What's grieving actually like?

My first explicit model of "applied grief" came early in the pandemic. It was a bit of a "grieving on hard mode", because I had to grieve for an abstract concept. I had a conflict in my quarantine-bubble about covid policy.We had disagreements about the facts-of-the-matter of how dangerous covid was, or what mitigation efforts would help. We had disagreements about the best ways to negotiate over policy, and how to interpret our policies, and how to enforce violations. We were also just, like, really stressed out.

Different bubble-members have different opinions on what actually happened and why. But

Some bubble-members did things that I felt hurt by. A few months later, I found myself recovered from the raw-stress of the experience, but still feeling psychologically messed up, waking up crying about once a week.

For awhile, I was confused about what I actually wanted. Maybe I wanted a bubble-member to admit they were wrong and apologize. It became clear that wasn't going to happen.

After several weeks of ruminating, I eventually figured out I actually wanted something like: "Bring in community ...

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: Proposal: Impact List -- like the Forbes List except for impact via donations, published by Elliot Olds on May 30, 2022 on The Effective Altruism Forum.

Summary

I'm seeking feedback on and collaborators for the creation, popularization, and maintenance of a list which ranks the top N living people by their positive impact via donations. The goal is to make the list popular enough to increase the status awarded to those who rank highly, bring more awareness to the importance of donation effectiveness, and ultimately cause people to donate more effectively and/or donate more money.

Details of the list

I still have a lot of uncertainty about how the list should work, but this is my current view:

People are ranked by the total lifetime amount of money donated to altruistic causes, scaled by effectiveness as determined by Impact List researchers (via a mix of independent research and repurposing research from GiveWell and other organizations).

The first version might contain somewhere between 100 and 1,000 people.

Viewers can specify their values which are taken into account by the rankings: how much weight to give animals, people who will exist in the far future, etc. The default settings would attempt to capture the values of the median viewer.

The following columns exist, which the viewer can rank by:

The raw (unscaled) total amount donated (which can be changed to the amount in a specific year)

Amount pledged

Net worth

The list can be filtered by cause area.

There's a drop-down to view alternate rankings provided by other organizations.

Any organization submitting a new ranking metric must also submit the reasoning / calculations behind the scaling factors they assign to donation recipients.

There's a page for each person showing every donation they've made, when it was made, and the scaled effectiveness of the donation according to Impact List and other organizations.

There's a page for each donation recipient showing the research and calculations for why it has the effectiveness rating that it does, according to Impact List and any other organization that has provided their own research.

Causal mechanisms

If Impact List becomes popular it could influence people's donations in the following ways:

The list focuses attention on large (and potentially surprising to many viewers) differences in donation effectiveness, and on how much positive impact others are having. This may lead people to want to have more impact for its own sake.

Attention on the list increases the status reward for giving effectively as well as the status penalty for giving ineffectively.

An explicit ranking might make some donors want to rank higher than others for competitive reasons.

The website could serve as an on-ramp for future EAs, with links to further reading. Billionaires may occasionally visit the website to check their ranking and the reasoning behind it, which could help us acquire more Sam Bankman-Frieds or Dustin Moskovitzs.

Note that reasons #1 and (the first part of) #4 could apply roughly equally to anyone who is aware of the list, not just those wealthy enough to potentially appear on it.

A large continuous auction

Because so many existing donations by the world’s wealthiest people are much less effective than they could be, spots on the list will initially be available for relatively low donation amounts to those who donate effectively. A list of size 1,000 might initially present an opportunity to hundreds of thousands of people to potentially appear on it (and push an ineffective donor off). This means that all four reasons above could be relevant for many more people than the size of the list.

To encourage the donations of the list members to continuously become more effective the UI could have a tool which shows people who aren't on the list how much money they'd need to don...

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: Ceiling Air Purifier, published by jefftk on May 30, 2022 on LessWrong.

Imagine a high-capacity air filter that is extremely quiet and doesn't get in the way. I think I see how to build one; here's a prototype:

By putting a fan on the ceiling you can have very long blades. This is important for keeping noise down: the longer your blades are the more air you move at a given rotational velocity, which is a major determinant of noise. Testing my prototype, it has a CADR of ~180 CFM and is only 33dB. By contrast, the Wirecutter's top-recommended air purifier (Coway

AP-1512HH) has a CADR of 233 CFM (31% higher) at 54 dB (21dB louder) or 110 CFM (38% lower) at 36 dB (3dB louder). With some tweaks it should be able to match the commercial purifier's performance, without being louder.

Volume is very important here: people often select purifiers based on their maximum flow, but then run them on a lower setting or even turn them off because they are too loud.

Note that while my purifier is an octagon of 12x20 filters around a 52" fan, a nonagon would have been better. Not only should that get you 12% more filtration, but you don't have to fill gaps between the filters.

Jeff for scale

One downside of using this much tape, especially some of it with the sticky side bare, is that you get an adhesive smell. The smell fell below where I could detect after about a day, but Julia could still smell it for about five days.

To determine the CADR of the prototype I followed the approach I described in Testing Air

Purifiers, where I created smoke by burning matches and then tracked how quickly the purifier could reduce pm2.5 levels:

To determine volume levels, I used a Sennheiser 835s dynamic microphone with a foam windscreen, connected to an AudioBox 1818VSL to record samples of each purifier option and silence. Here are the samples:

(Silence)

(Ceiling Purifier)

(AP-1512HH Medium)

(AP-1512HH High)

To compute the loudness of each sample I ran:

This gives loudness in LUFS, which is K-weighted dB. I don't have any calibrated tooling, but I can report values in loudness units (LU) relative to silence. Trusting the manufacturer's rating of 53.8 dB for the AP-1512HH on High, however, lets us determine absolute loudness for the other options:

Sample Relative Absolute Everything Off normalized to 0 LU 25 dB Ceiling Purifier High 7.3 LU 32.6 dB AP-1512HH Medium 10.8 LU 36.1 dB AP-1512HH High 28.5 LU 53.8 dB

I think it ought to be possible to get even better performance out of the ceiling purifier by building a shroud. The ideal shroud is generally quite a bit narrower than I would guess: I've seen people measuring peak performance with only a 14" hole for a 20" fan. They report ~40% more airflow without increased noise, but I would want to run tests and verify that here.

With ~12% improvement from switching to nine filters and ~40% from better shrouding, this ought to get us from a CADR of ~180 CFM to ~280

CFM. Even if we only get half that improvement, a CADR ~230 CFM gets us even with the commercial purifier, at far lower volumes.

In terms of cost, materials for one fan are nine MERV-14 filters for

$110 plus ~$5 for tape. The ceiling fan was a Hunter

Mariner 52", which seems to cost $106 now, though it's not something I installed for this project. A real-product version of this would replace the tape with a frame you attach to the ceiling, into which you could slide filters. Maybe $100? If someone wanted to build and sell something I'd be pretty happy to see that!

Comment via: facebook

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