After working on WIT, I’ve grown a lot more comfortable producing provisional answers to deep questions. In similar academic work, there are strong incentives to only try to answer questions in ways that are fully defensible: if there is some other way of going about it that gives a different result, you need to explain why your way is better. For giant nebulous questions, this means we will make very slow progress on finding a solution. Since these questions can be very important, it is better to come up with some imperfect answers rather than just working on simpler problems. WIT tries to tackle big important nebulous problems, and we have to sometimes make questionable assumptions to do so. The longer I’ve spent here, the more worthwhile our approach feels to me.

Excellent question, Ian! At a high level, I’d say that moral uncertainty has made me much more inclined to care about having an overlapping consensus of reasons for any important decision. Equivalently, I want a diverse set of considerations to point in the same direction before I’m inclined to make a big change. That’s how I got into animal work in the first place. It’s good for the animals, good for human health, good for long-term food security, good for the environment, etc. There are probably lots of other impacts too, but that’s the first one that comes to mind!

A few things come to mind. First, I’ve been really struck by how robust animal welfare work is across lots of kinds of uncertainties. It has some of the virtues of both GHD (a high probability of actually making a difference) and x-risk work (huge scales). Second, when working with the Moral Parliament tool, it is really striking how much of a difference different aggregation methods make. If we use approval voting to navigate moral uncertainty, we get really different recommendations than if we give every worldview control over a share of the pie or if we maximize expected choiceworthiness. For me, figuring out which method we should use turns on what kind of community we want to be and which (or whether!) democratic ideals should govern our decision-making. This seems like an issue we can make headway on, even if there are empirical or moral uncertainties that prove less tractable.

One of the big prioritization changes I’ve taken away from our tools is within longtermism. Playing around with our Cross-Cause Cost-Effectiveness Model, it was clear to me that so much of the expected value of the long-term future comes from the direction we expect it to take, rather than just whether it happens at all. If you can shift that direction a little bit, it makes a huge difference to overall value. I no longer think that extinction risk work is the best kind of intervention if you’re worried about the long-term future. I tend to think that AI (non-safety) policy work is more impactful in expectation, if we worked through all of the details.

I was personally struck by how sensitive portfolios are to even modest levels of risk aversion. I don’t know what “correct” level of risk aversion is, or what the optimal decision procedure is in practice (even though most of my theoretical sympathies lie with expected value maximisation). Even so, seeing how introducing bits of risk aversion, even when using parameters relatively generous towards x-risk, still points towards spending most resources on animals (and sometimes global health) has led me to believe that type of work is robustly better than I used to think. There are many uncertainties and I don't think EA should be reduced to any one of its cause-areas but, especially given this update, I would be sad to see the animal space shrink in relative size any more than it has.

I’m a “chickens and children” EA, having come to the movement through Singer’s arguments about animals and global poverty. I still find EA most compelling both philosophically and emotionally when it focuses on areas where it’s clear that we can make a difference. However, the more I grapple with the many uncertainties associated with resource allocation, the more sympathetic I become to diversification, to include significant resources for work that doesn’t appeal to me at all personally. So you probably won’t catch me pivoting to AI governance anytime soon, but I’m glad others are doing it. 

Great questions, Vaipan. To your first question, the short answer is “Yes.” We can’t say much about our efforts in that direction, but we’re certainly keen to have broad influence! To your second question, we’ve received positive reviews of our work from several people at OP and have had lots of productive discussions about the place of risk aversion in resource allocation. However, we can’t comment on OP’s internal decision-making. 

Good question! Re: the Moral Weight Project, perhaps the biggest area of impact has been on animal welfare economics, where having a method to make interspecies comparisons is crucial for benefit-cost analysis. Many individuals and organizations have also reported to us that our work was an update on the importance of animals and on invertebrates specifically. We’ve seen something similar with the CCM tool, with results ranging from positive feedback and enthusiasm to more concrete updates in their decisions. There’s more we can say privately than publicly, however, so please feel free to get in touch if you’d like to chat! 

Thanks for raising this point. We think that choosing the right decision theory that can handle imprecise probabilities is a complex issue that has not been adequately resolved. We take the point that Mogensen’s conclusions have radical implications for the EA community at large and we haven’t formulated a compelling story about where Mogensen goes wrong. However, we also believe that there are likely to be solutions that will most likely avoid those radical implications, and so we don’t need to bracket all of the cause prioritization work until we find them. Our tools may only be useful to those who see there to be work done on cause prioritization.

As a practical point, our Cross-Cause Cost-Effectiveness Model handles precise probabilities with Monte Carlo methods by randomly selecting individual values for parameters in each outcome from a distribution. We noted hesitance about enforcing a specific distribution over our range of radical uncertainty, but we stand behind this as a reasonable choice given our pragmatic aims. If the alternative is not to try to calculate relative expected values, we think that would be a loss, even if our own results have methodological doubts still attached to them.

Thanks for these questions, Toby!

• Re: fitting products to our audience, that’s one reason we release them on the Forum! All our tools are in beta; the feedback we receive here is one of the  important ways we identify necessary refinements. As time and funding permit, we hope to improve our tools so that they better serve individuals and organizations trying to do as much good as they can. That being said, we also did a lot of user testing in advance, soliciting feedback on many iterations of each tool to improve usability and accessibility.
• Re: how we hope people will use the Moral Parliament Tool, we have two main goals. First, we hope that people will use it to have more transparent conversations about their disagreements. For instance, when people are debating the merits of a particular intervention, is the crux the probability that an intervention will backfire, how bad they think backfiring will be, their relative aversions to backfiring, or the way they think they should navigate uncertainty given all their other commitments? The tool forces people to make these kinds of differences explicit and think through their implications. Second, we hope that people will use the Moral Parliament Tool to explore the implications of even modest levels of uncertainty. The tool makes it obvious that changes to parameter values, credences in moral theories, and aggregation methods have big consequences for overall allocations!

• What are selfish lifestyle reasons to work on the WIT team?
  • It’s fun to talk to smart people! Remote work is great. It’s a privilege to be able to think about big problems that are both philosophically complicated and practically important. 
• Is it fair to say the work WIT does is unusual outside of academia? What are closely related organizations that tackle similar problems?
  • Yes, what we do is very unusual outside of academia—and inside it too. Re: other groups that do global priorities research, the most prominent ones are GPI, PWI, and the cause prio teams at OP.
• How does your team define "good enough" for a sequence? What adjustments do you make when you fall behind schedule? Cutting individual posts? Shortening posts? Spending more time?
  • That’s a hard one and we’re still trying to figure it out. There are a lot of variables here, many of which are linked to whether we have the funding to linger on a particular project. In general, however, our job isn’t to produce academic research: it’s to inform decisions. So, if we think we’ve done enough to help people who need to make decisions, then that’s a good sign that we should wrap up the project soon.
• How much does the direction of a sequence change as you're writing it? It seems like you have a vision in mind when starting out, but you also mention being surprised by some results.
  • The general structure tends not to change much—we plan out posts together and have a general sense of the research we want to do—but the narrative certainly evolves as we learn more about the topic we’re investigating. The conclusions definitely aren’t set from the beginning!
• Can you tell us more about the structure of research meetings? How frequently do individual authors chat with each other and for what reason? In particular, the CURVE sequence feels very intentionally like a celebration of different "EA methodologies". Most of the posts feel individual before converging on a big cost-effectiveness analysis.
  • We’re in touch all the time, brainstorming new ideas, reviewing drafts, and figuring out solutions to problems. The whole team meets once or twice a week and then we individually hop on 1-1 calls more frequently to discuss specific aspects of our projects. Most of the research still has a lead who’s driving it forward, but everyone’s fingerprints tend to be on everything. 
• Much of your work feels numerical simulation over discrete choices. Have there been attempts to define "closed-form" analytical equations for your work? What are reasons to allocate resources to this versus not?
  • This ties to your previous question “How does your team define "good enough" for a sequence?”. We think analytical equations can be valuable (they are often tidier, speed up computational work, and can provide clearer insights into sensitivity analysis). For example, it’s a natural next step in our human extinction post, which we flagged in the conclusion. And indeed we’ve done some work towards this already but not polished it enough for it to be in a shareable state. Back to your question “when is a piece of research good enough to wrap up?” We don’t know for sure, but we’ve found that running computational simulations that we’re sufficiently confident in gives us approximations that are perfectly suitable to learn about the models we’re interested in. We hear you, closed-form solutions are mathematically satisfying. But, once we’ve learned the main headlines, it’s hard to justify spending the extra time working through closed-form solutions for everything, especially for some of the more complex models with several moving parts.
• What are the main constraints the WIT team faces?
  • The standard ones: we’re funding- and capacity-constrained. We could do a lot more with additional resources!

Here’s one method that we’ve found helpful when presenting our work. To get a feel for how the tools work, we set challenges for the group: find a set of assumptions that gives all resources to animal welfare; find how risk averse you’d have to be to favor GHD over x-risk; what moral views best favor longtermist causes? Then, have the group discuss whether and why these assumptions would support those conclusions. Our accompanying reports are often designed to address these very questions, so that might be a way to find the posts that really matter to you. 

Thanks for looking through our work and for your comment, Deborah. We recognise that different parts of our models are often interrelated in practice. In particular, we’re concerned about the problem of correlations between interventions too, as we flag here. This is an important area for further work. That being said, it isn’t clear that the cases you have in mind are problems for our tools. If you think, for instance, that environmental interventions are particularly good because they have additional (quantifiable or non-quantifiable) benefits, you can update the tool inputs (including the cause or project name) to reflect that and increase the estimated impact of that particular cause area. We certainly don't mean to imply that climate change is an unimportant issue.