Hi Bill. Thanks for engaging with our work. I agree with you that the project had lots of limitations. And as I’ve said many times, no one should put much stock in the specific estimates we offered. A few additional thoughts:
We don’t say that 14 bees = 1 human. We say that if the one and only thing you care about is the intensity of valenced experience (which is not the only thing that most people care about), then, probably, you should think of the value of a life year of bee suffering as being within an order of magnitude of the value of a life year of human suffering. Of course, there are lots of reasons why we might be wrong about that more modest claim. Still, just wanted to clarify.
Totally fair about there being a double-counting problem, which we discuss in section 3.1 here). It’s a tough problem to tackle given the functional approach we chose. It’s also one reason why, in the work I’m doing now on this topic, I’m interested in aggregating over very different kinds of estimation strategies rather than developing a functional approach in more detail. That being said, I do think it’s pretty interesting if it turns out that there’s a lot of clustering of intensity-relevant traits. That would be some evidence that welfare ranges don’t differ that much, at least in my view.
Fair point re: the impacts of your priors. Of course, it’s an open question whether you ought to have such a low prior in insect sentience. FWIW, I think 0.01% is probably overconfident, given how poorly we understand consciousness generally and sentience specifically. (For some additional thoughts on this, see the “Studying sentience” section of this article.) I’ll also mention that there’s a methodological disagreement here about how to approach this kind of problem. I say just a bit about it in this comment.
Hi Bill. Thanks for engaging with our work. I agree with you that the project had lots of limitations. And as I’ve said many times, no one should put much stock in the specific estimates we offered. A few additional thoughts:
We don’t say that 14 bees = 1 human. We say that if the one and only thing you care about is the intensity of valenced experience (which is not the only thing that most people care about), then, probably, you should think of the value of a life year of bee suffering as being within an order of magnitude of the value of a life year of human suffering. Of course, there are lots of reasons why we might be wrong about that more modest claim. Still, just wanted to clarify.
Totally fair about there being a double-counting problem, which we discuss in section 3.1 here). It’s a tough problem to tackle given the functional approach we chose. It’s also one reason why, in the work I’m doing now on this topic, I’m interested in aggregating over very different kinds of estimation strategies rather than developing a functional approach in more detail. That being said, I do think it’s pretty interesting if it turns out that there’s a lot of clustering of intensity-relevant traits. That would be some evidence that welfare ranges don’t differ that much, at least in my view.
Fair point re: the impacts of your priors. Of course, it’s an open question whether you ought to have such a low prior in insect sentience. FWIW, I think 0.01% is probably overconfident, given how poorly we understand consciousness generally and sentience specifically. (For some additional thoughts on this, see the “Studying sentience” section of this article.) I’ll also mention that there’s a methodological disagreement here about how to approach this kind of problem. I say just a bit about it in this comment.