Part of a series for My Cause Selection 2016. For background, see my writings on cause selection for 2015 and my series on quantitative models.

The last time I wrote about values spreading, I primarily listed reasons why we might expect existential risk reduction to have a greater impact. Now I’m going to look at why values spreading—and animal advocacy in particular—may look better.

When I developed a quantitative model for cause prioritization, the model claimed that effective animal advocacy has a greater expected impact than AI safety research. Let’s look at some qualitative reasons why the model produces this result:

• Animal advocacy has lower variance—we’re more confident that it will do a lot of good, especially in the short to medium term.
• Animal advocacy is more robustly positive—it seems unlikely to do lots of harm¹, whereas the current focus of AI safety research could plausibly do harm. (This is really another way of saying that AI safety interventions have high variance.)
• The effects of animal advocacy on the far future arguably have better feedback loops.
• Animal advocacy is more robust against overconfidence in speculative arguments. I believe we ought to discount the arguments for AI safety somewhat because they rely on hard-to-measure claims about the future. We could similarly say that we shouldn’t be too confident what effect animal advocacy will have on the far future, but it also has immediate benefits. Some people put a lot of weight on this sort of argument; I don’t give it tons of weight, but I’m still wary given that people have a history of making overconfident claims about what the future will look like.

Continue reading

Let’s look at how we use frameworks to prioritize causes. We’ll start by looking at the commonly-used importance/neglectedness tractability framework and see why it often works well and why it doesn’t match reality. Then we’ll consider an alternative approach.

Importance/neglectedness/tractability framework

When people do high-level cause prioritization, they often use an importance/neglectedness/tractability framework where they assess causes along three dimensions:

1. Importance: How big is the problem?
2. Neglectedness: How much work is being done on the problem already?
3. Tractability: Can we make progress on the problem?

This framework acts as a useful guide to cause prioritization. Let’s look at some of its benefits and problems.

Continue reading

Part of a series on quantitative models for cause selection.

Introduction

In the past I’ve written qualitatively about what sorts of interventions likely have the best far-future effects. But qualitative analysis is maybe not the best way to decide this sort of thing, so let’s build some quantitative models.

I have constructed a model of various interventions and put them in a spreadsheet. This essay describes how I came up with the formulas to estimate the value of each intervention and makes a rough attempt at estimating the inputs to the formulas. For each input, I give either a mean and σ¹ or an 80% confidence interval (which can be converted into a mean and σ). Then I combine them to get a mean and σ for the estimated value of the intervention.

This essay acts as a supplement to my explanation of my quantitative model. The other post explains how the model works; this one goes into the nitty-gritty details of why I set up the inputs the way I did.

Note: All the confidence intervals here are rough first attempts and don’t represent my current best estimates; my main goal is to explain how I developed the presented series of models. I use dozens of different confidence intervals in this essay, so for the sake of time I have not revised them as I changed them. To see my up-to-date estimates, see my final model. I’m happy to hear things you think I should change, and I’ll edit my final model to incorporate feedback. And if you want to change the numbers, you can download the spreadsheet and mess around with it. This describes how to use the spreadsheet.

Continue reading

Part of a series on quantitative models for cause selection.

Update: There’s now a web app that can do everything the spreadsheet could and more.

Quantitative models offer a superior approach in determining which interventions to support. However, naive cost-effectiveness estimates have big problems. In particular:

1. They don’t give stronger consideration to more robust estimates.
2. They don’t always account for all relevant factors.

We can fix the first problem by starting from a prior distribution and updating based on evidence–more robust evidence will cause a bigger probability update. And we can fix the second problem by carefully considering the most important effects of interventions and developing a strong quantitative estimate that incorporates all of them.

So that’s what I did. I developed a quantitative model for comparing interventions and wrote a spreadsheet to implement it.

Continue reading

Update 2016-12-14: GiveWell’s 2016 cost-effectiveness analysis has updated the way it handles population ethics. It now explicitly takes the value of saving a 5-year old’s life as input and no longer assumes that it’s worth 36 life-years.

Update 2018-08-14: I recently revisited GiveWell’s 2018 cost-effectiveness analysis. Although the analysis spreadsheet no longer enforces the “GiveWell view” described in this essay, most GiveWell employees still implicitly adopt it. As a result, I believe GiveWell is still substantially mis-estimating the cost-effectiveness of the Against Malaria Foundation.

GiveWell claims that the Against Malaria Foundation (AMF) is about 10 times as cost-effective as GiveDirectly. This entails unusual claims about population ethics that I believe many people would reject, and according to other plausible views of population ethics, AMF looks less cost-effective than the other GiveWell top charities.

GiveWell’s Implicit Assumptions

A GiveWell-commissioned report suggests that population will hardly change as a result of AMF saving lives. GiveWell’s cost-effectiveness model for AMF assumes that saving one life creates about 35 quality-adjusted life years (QALYs), and uses this to assign a quantitative value to the benefits of saving a life. But if AMF causes populations to decline, that means it’s actually removing (human) QALYs from the world; so you can’t justify AMF’s purported cost-effectiveness by saying it creates more happy human life, because it doesn’t.

You could instead justify AMF’s life-saving effects by saying it’s inherently good to save a life, in which case GiveWell’s cost-effectiveness model shouldn’t interpret the value of lives saved in terms of QALYs created/destroyed, and should include a term for the inherent value of saving a life.

GiveWell claims that AMF is about 10 times more cost-effective than GiveDirectly, and GiveWell ranks AMF as its top charity partially on this basis (see “Summary of key considerations for top charities” in the linked article). This claim depends on the assumption that saving a life creates 35 QALYs.

To justify GiveWell’s cost-effectiveness analysis, you could say that it is good to cause existing people to live longer, but it is not bad to prevent people from existing. (Sean Conley of GiveWell says he and many other GiveWell staffers believe this.)

In particular, you’d have to assume that:

Continue reading

Part of a series on quantitative models for cause selection.

Introduction

One major reason that effective altruists disagree about which causes to support is that they have different opinions on how strong an evidence base an intervention should have. Previously, I wrote about how we can build a formal model to calculate expected value estimates for interventions. You start with a prior belief about how effective interventions tend to be, and then adjust your naive cost-effectiveness estimates based on the strength of the evidence behind them. If an intervention has stronger evidence behind it, you can be more confident that it’s better than your prior estimate.

For a model like this to be effective, we need to choose a good prior belief. We start with a prior probability distribution P where P(x) gives the probability that a randomly chosen intervention¹ has utility x (for whatever metric of utility we’re using, e.g. lives saved). To determine the posterior expected value of an intervention, we combine this prior distribution with our evidence about how much good the distribution does.

For this to work, we need to know what the prior distribution looks like. In this essay, I attempt to determine what shape the prior distribution has and then estimate the values of its parameters.

Continue reading

Introduction

The Open Philanthropy Project has made some grants that look substantially less impactful than some of its others, and some people have questioned the choice. I want to discuss some reasons why these sorts of grants might plausibly be a good idea, and why I ultimately disagree.

I believe Open Phil’s grants on criminal justice and land use reform are much less effective in expectation¹ than its grants on animal advocacy and global catastrophic risks. This would naively suggest that Open Phil should spend all its resources on these more effective causes, and none on the less effective ones. (Alternatively, if you believe that the grants on US policy do much more good than the grants on global catastrophic risk, then perhaps Open Phil should focus exclusively on the former.) There are some reasons to question this, but I believe that the naive approach is correct in the end.

Why give grants in cause areas that look much less effective than others? Why give grants to lots of cause areas rather than just a few? Let’s look at some possible explanations for these questions.

Continue reading

Part of a series on quantitative models for cause selection.

Introduction

Last time, I wrote about the most likely far future scenarios and how good they would probably be. But my last post wasn’t precise enough, so I’m updating it to present more quantitative evidence.

Particularly for determining the value of existential risk reduction, we need to approximate the probability of various far future scenarios to estimate how good the far future will be.

I’m going to ignore unknowns here–they obviously exist but I don’t know what they’ll look like (you know, because they’re unknowns), so I’ll assume they don’t change significantly the outcome in expectation.

Here are the scenarios I listed before and estimates of their likelihood, conditional on non-extinction:

*not mutually exclusive events

(Kind of hard to read; sorry, but I spent two hours trying to get flowcharts to work so this is gonna have to do. You can see the full-size image here or by clicking on the image.)

I explain my reasoning on how I arrived at these probabilities in my previous post. I didn’t explicitly give my probability estimates, but I explained most of the reasoning that led to the estimates I share here.

Some of the calculations I use make certain controversial assumptions about the moral value of non-human animals or computer simulations. I feel comfortable making these assumptions because I believe they are well-founded. At the same time, I recognize that a lot of people disagree, and if you use your own numbers in these calculations, you might get substantially different results.

Continue reading

Part of a series on quantitative models for cause selection.

Alternate title: Excessive Pessimism About Far Future Causes

In my post on cause selection, I wrote that I was roughly indifferent between $1 to MIRI, $5 to The Humane League (THL), and $10 to AMF. I based my estimate for THL on the evidence and cost-effectiveness estimates for veg ads and leafleting. Our best estimates suggested that these are conservatively 10 times as cost-effective as malaria nets, but the evidence was fairly weak. Based on intuition, I decided to adjust this 10x difference down to 2x, but I didn’t have a strong justification for the choice.

Corporate outreach has a lower burden of proof (the causal chain is much clearer) and estimates suggest that it may be ten times more effective than ACE top charities’ aggregate activities¹. So does that mean I should be indifferent between $5 to ACE top charities and $0.50 to corporate campaigns? Or perhaps even less, because the evidence for corporate campaigns is stronger? But I wouldn’t expect this 10x difference to make corporate campaigns look better than AI safety, so I can’t say both that corporate campaigns are ten times better than ACE top charities and also that AI safety is only five times better. My previous model, in which I took expected value estimates and adjusted them based on my intuition, was clearly inadequate. How do I resolve this? In general, how can we quantify the value of robust, moderately cost effective interventions against non-robust but (ostensibly) highly cost effective interventions?

To answer that question, we have to get more abstract.

Continue reading

Update 2016-05-18: I no longer entirely agree with this post. In particular, I believe GiveWell employees are more replaceable than this post suggests. I may write about my updated beliefs in the future.

Edited 2016-03-11 because I’ve adjusted my estimate of the value of global poverty charities downward, which makes working at GiveWell look worse.

Edited 2016-03-11 to add a new section.

Edited 2016-02-16 to update the model based on feedback I’ve received. Temporal replaceability doesn’t apply so I was underestimating the value of research analysts.

Summary: The value of working as a research analyst¹ at GiveWell is determined by:

• Temporal replaceability of employees
• How good you are relative to the counterfactual employee
• How much good GiveWell money moved does relative to where you could donate earnings
  • A lot if you care most about global poverty, not as much if you care about other cause areas
• How directly more employees translate into better recommendations and more money moved
  • This relationship looks strong for Open Phil and weak for GiveWell Classic

If you believe GiveWell top charities are the best place to donate, working at GiveWell is probably a really strong career option; if you believe other charities are substantially better (as I do) and you have good earning potential, earning to give is probably better.

Continue reading