It might be worth adding weighting by the number of neurons in sensory-associative structures (or functions of these) as an option, if the data is available.
Did you include probabilities of sentience? This would be separate from brain weighting. Charity Entrepreneurship used some numbers, I think mostly borrowed from Luke Muehlhauser at Open Phil, but they aren’t already included in their welfare points, so you need to multiply by them yourself. Rethink Priorities has more recent estimates.
I think a weighted additive model would be preferable to a weighted product model, mostly just because if we were to combine the harms, we would just add them. A cost-benefit analysis would put prices on CO2e and suffering and add them up, and if we wanted to offset the harms, we would think of them additively. Then,
You could mention the units of food (2000 kcal calories) on the graph.
You could add units to the graph at least for CO2e emissions, since those are pretty interpretable.
You could make up units for suffering (say relative to some reference per day), and include those units, too. Basically, you would be looking at labelling the axis by “CO2e (kg) + weight * suffering”, and then just allow weight to go from 0 to infinity (at which point CO2e emissions are ignored). I think this would be more interpretable than the current percentage weights, too.
If you’re including CO2e emissions, it might be worth adding some plants (and insects?). (People may complain that harms to wild animals aren’t included, but to be fair, we would also want to include benefits to wild animals for each product, too, and this would be messy and not super tractable.)
Thank you for the feedback, MichaelStJules! I added all of your ideas to my todo list.
I definitely should have added probability of sentience to the model. I looked at Brian Tomasik’s model which included sentience multipliers and I have read the OPP report you linked so I don’t know why I didn’t consider it.
Jason Schukraft’s “Differences in the Intensity of Valenced Experience across Species” was great and I will be happy to study his other research. Thank you for linking to it.
I wish I was aware of kbog’s post before I started. I managed to find multiple analyses of suffering but I didn’t know someone had already devised a combined model!
This is very cool! Good job!
Some thoughts:
It might be worth adding weighting by the number of neurons in sensory-associative structures (or functions of these) as an option, if the data is available.
Did you include probabilities of sentience? This would be separate from brain weighting. Charity Entrepreneurship used some numbers, I think mostly borrowed from Luke Muehlhauser at Open Phil, but they aren’t already included in their welfare points, so you need to multiply by them yourself. Rethink Priorities has more recent estimates.
I think a weighted additive model would be preferable to a weighted product model, mostly just because if we were to combine the harms, we would just add them. A cost-benefit analysis would put prices on CO2e and suffering and add them up, and if we wanted to offset the harms, we would think of them additively. Then,
You could mention the units of food (2000 kcal calories) on the graph.
You could add units to the graph at least for CO2e emissions, since those are pretty interpretable.
You could make up units for suffering (say relative to some reference per day), and include those units, too. Basically, you would be looking at labelling the axis by “CO2e (kg) + weight * suffering”, and then just allow weight to go from 0 to infinity (at which point CO2e emissions are ignored). I think this would be more interpretable than the current percentage weights, too.
If you’re including CO2e emissions, it might be worth adding some plants (and insects?). (People may complain that harms to wild animals aren’t included, but to be fair, we would also want to include benefits to wild animals for each product, too, and this would be messy and not super tractable.)
You could also consider weighting by critical flicker-fusion frequency to track the subjective experience of time. Jason Schukraft wrote 3 posts on this topic. Here’s the summary, which also links to the other two posts. Consider reaching out to Jason to see if this makes sense.
Also, it might be worth taking a look at this model, too, for more ideas.
Thank you for the feedback, MichaelStJules! I added all of your ideas to my todo list.
I definitely should have added probability of sentience to the model. I looked at Brian Tomasik’s model which included sentience multipliers and I have read the OPP report you linked so I don’t know why I didn’t consider it.
Jason Schukraft’s “Differences in the Intensity of Valenced Experience across Species” was great and I will be happy to study his other research. Thank you for linking to it.
I wish I was aware of kbog’s post before I started. I managed to find multiple analyses of suffering but I didn’t know someone had already devised a combined model!