This is a challenging question, since new interventions in our space typically lack any good data on cost-effectiveness. But in general I’d set a much lower bar than ~10X cage-free campaigns. Instead I think a promising intervention is worth trying if it has an expected value at least as good as cage-free campaigns. E.g. If we think there’s a 50% chance a new intervention will fail and a 50% chance it will be 2X cage-free campaigns, I think it’s worth a shot.
It may be worth considering even interventions that seem less cost-effective than marginal cage-free campaigns, say because:
You can gather evidence on their cost-effectiveness and build capacity for the future, when cage-free campaigns are less cost-effective.
If the upside is high enough and feedback loops are good enough, you could scale it up if it seems successful or shut it down if not. For example, if it has a 5% chance of being 10x cage-free campaigns is worthless otherwise, then the EV is only 50% cage-free. After a pilot, if you become confident that it will succeed and scale, then you now have a 10x intervention, which would be great. If instead you become confident that it isn’t cost-effective, hopefully you didn’t spent too much to find that out, and then you can stop funding it.
Diversifying across intervention types, regions, or species might be instrumentally useful (e.g. for capacity building) or otherwise valuable if you’re somewhat difference-making risk averse or difference-making ambiguity averse, which I assume most who prioritize animal welfare are.
I suppose for most of these, more careful cost-effectiveness modelling can actually capture these benefits. For 2, AIM/Charity Entrepreneurship often models the expected benefits and expected costs, taking into account different benefits and costs conditional on success/scale up and separately conditional on failure/shutdown (with good enough feedback loops). You can also think of this like value of information. For 1 and 3, you can also just include the value of indirect benefits like capacity building and value of information.
Thanks Michael. Yeah I agree with those three categories. In practice we support a lot of interventions with much worse short-term cost-effectiveness than cage-free campaigns, in part for information value, in part so we can scale them up if they do work out, and in part for diversification purposes.
This is a challenging question, since new interventions in our space typically lack any good data on cost-effectiveness. But in general I’d set a much lower bar than ~10X cage-free campaigns. Instead I think a promising intervention is worth trying if it has an expected value at least as good as cage-free campaigns. E.g. If we think there’s a 50% chance a new intervention will fail and a 50% chance it will be 2X cage-free campaigns, I think it’s worth a shot.
It may be worth considering even interventions that seem less cost-effective than marginal cage-free campaigns, say because:
You can gather evidence on their cost-effectiveness and build capacity for the future, when cage-free campaigns are less cost-effective.
If the upside is high enough and feedback loops are good enough, you could scale it up if it seems successful or shut it down if not. For example, if it has a 5% chance of being 10x cage-free campaigns is worthless otherwise, then the EV is only 50% cage-free. After a pilot, if you become confident that it will succeed and scale, then you now have a 10x intervention, which would be great. If instead you become confident that it isn’t cost-effective, hopefully you didn’t spent too much to find that out, and then you can stop funding it.
Diversifying across intervention types, regions, or species might be instrumentally useful (e.g. for capacity building) or otherwise valuable if you’re somewhat difference-making risk averse or difference-making ambiguity averse, which I assume most who prioritize animal welfare are.
I suppose for most of these, more careful cost-effectiveness modelling can actually capture these benefits. For 2, AIM/Charity Entrepreneurship often models the expected benefits and expected costs, taking into account different benefits and costs conditional on success/scale up and separately conditional on failure/shutdown (with good enough feedback loops). You can also think of this like value of information. For 1 and 3, you can also just include the value of indirect benefits like capacity building and value of information.
Thanks Michael. Yeah I agree with those three categories. In practice we support a lot of interventions with much worse short-term cost-effectiveness than cage-free campaigns, in part for information value, in part so we can scale them up if they do work out, and in part for diversification purposes.