Thanks for this, I’m trying to at least get a couple of numbers to get in the right ballpark to help people calibrate. Epistemic status: uncertain, a couple of hours of research. Takeaway: Probably billions of fish killed.
To get a higher end of the estimate of total fish killed as a result of mosquito net fishing, the appendix of the paper you link to gives an example of a single large-scale fishery in Madagascar where 68.7 million fish were killed in 2018-2019 by 75 families.
Unless I’m misreading, this looks like 80-90% of the total fish numbers were caught through mosquito net trawler (27.7% by weight, but much smaller fish on average, so a higher number).
This other paper in Mozambique provides another example—they two fishers catch around 35kg of (mostly tiny, < 20 mm, so 1-10g?) fish a day, so I guess around 10,000 fish a day? Over 2 million a year (so similar ballpark to Madagascar).
These are the locations at which mosquito net fishing had been observed in this study (data mostly from 2015), with a rapidly rising trajectory.
I can’t seem to find any better data on the scale of fishing at all of these locations, so I have no idea whether it’s closer to 50 or 10000 locations where mosquito nets are used for fishing to a similar extent to those in these articles, but 25 million fish * 50 locations would be 1.25 billion fish.
I think it’s safe to assume that at least 1 billion fish, possibly orders of magnitude higher, are killed yearly through mosquito net fishing, most of whom wouldn’t be killed if it weren’t for the distribution of mosquito nets (due to the small mesh sizes).
Of course, to make a reasonable welfare calculation, you shouldn’t just look at the number of animals killed and how painful their deaths must have been. You also have to consider how long these fish would have lived for counterfactually, the quality of their lives and suffering involved in their counterfactual deaths, as well as broader ecosystem effects etc.
But @Dylan Matthews’s claim that “There is little research on what fishing with these nets actually does to fish or people — but also little reason to think the magnitudes of these effects are remotely near the number of lives saved by nets” strikes me as indefensible.
Causing billions of fish to experience a painful death is not just a rounding error, and it may well be worth reevaluating whether bednets have had a positive net impact or not.
Thanks so much for this! I’d only been thinking about the potential harms to people with fish welfare as a side note. You’re absolutely right that we can get a decent estimate on the added burden of fish suffering here, which will be relevant to the calculations of many EAs
Fish welfare would be a big factor in my model about the net utility of free malaria nets. However, it seems really hard to calculate because we don’t know the average net qualia of fish in these regions. If it is negative, then overfishing may be a positive thing for their welfare.
Late to the party, but isn’t the relevant thing for AMF donors the counterfactual number of fish killed by mosquito nets distributed by AMF? It seems like AMF has higher rates of nets being used properly than other charities.
I couldn’t track down comparative data on whose bednets seem to be the most responsible—presumably it’s more whoever distributes the most nets in areas close to major fisheries, rather than % of appropriate use.
I also don’t know whether the marginal bednet will increase this kind of fishing much (there might already be a glut of bednets).
But these are all important questions that I don’t think GiveWell or AMF have ever taken seriously.
Thanks for this, I’m trying to at least get a couple of numbers to get in the right ballpark to help people calibrate. Epistemic status: uncertain, a couple of hours of research. Takeaway: Probably billions of fish killed.
To get a higher end of the estimate of total fish killed as a result of mosquito net fishing, the appendix of the paper you link to gives an example of a single large-scale fishery in Madagascar where 68.7 million fish were killed in 2018-2019 by 75 families.
Unless I’m misreading, this looks like 80-90% of the total fish numbers were caught through mosquito net trawler (27.7% by weight, but much smaller fish on average, so a higher number).
This other paper in Mozambique provides another example—
theytwo fishers catch around 35kg of (mostly tiny, < 20 mm, so 1-10g?) fish a day, so I guess around 10,000 fish a day? Over 2 million a year (so similar ballpark to Madagascar).These are the locations at which mosquito net fishing had been observed in this study (data mostly from 2015), with a rapidly rising trajectory.
I can’t seem to find any better data on the scale of fishing at all of these locations, so I have no idea whether it’s closer to 50 or 10000 locations where mosquito nets are used for fishing to a similar extent to those in these articles, but 25 million fish * 50 locations would be 1.25 billion fish.
I think it’s safe to assume that at least 1 billion fish, possibly orders of magnitude higher, are killed yearly through mosquito net fishing, most of whom wouldn’t be killed if it weren’t for the distribution of mosquito nets (due to the small mesh sizes).
Of course, to make a reasonable welfare calculation, you shouldn’t just look at the number of animals killed and how painful their deaths must have been. You also have to consider how long these fish would have lived for counterfactually, the quality of their lives and suffering involved in their counterfactual deaths, as well as broader ecosystem effects etc.
But @Dylan Matthews’s claim that “There is little research on what fishing with these nets actually does to fish or people — but also little reason to think the magnitudes of these effects are remotely near the number of lives saved by nets” strikes me as indefensible.
Causing billions of fish to experience a painful death is not just a rounding error, and it may well be worth reevaluating whether bednets have had a positive net impact or not.
Thanks so much for this! I’d only been thinking about the potential harms to people with fish welfare as a side note. You’re absolutely right that we can get a decent estimate on the added burden of fish suffering here, which will be relevant to the calculations of many EAs
Fish welfare would be a big factor in my model about the net utility of free malaria nets. However, it seems really hard to calculate because we don’t know the average net qualia of fish in these regions. If it is negative, then overfishing may be a positive thing for their welfare.
Late to the party, but isn’t the relevant thing for AMF donors the counterfactual number of fish killed by mosquito nets distributed by AMF? It seems like AMF has higher rates of nets being used properly than other charities.
I couldn’t track down comparative data on whose bednets seem to be the most responsible—presumably it’s more whoever distributes the most nets in areas close to major fisheries, rather than % of appropriate use.
I also don’t know whether the marginal bednet will increase this kind of fishing much (there might already be a glut of bednets).
But these are all important questions that I don’t think GiveWell or AMF have ever taken seriously.