Summary

I estimated the badness of eating farmed animals:
- As a fraction of the goodness of the human living time supported by their calories.
- In terms of decreasing human life expectancy per energy provided by their calories.
- In terms of smoked cigarettes per energy provided by their calories.
For example, my guess is that eating:
- A given mass/energy of crickets causes 1.03 k times as much suffering to crickets as the happiness of the human living time supported by such mass/energy (1 d for 2250 kcal).
- 100 kcal of shrimps causes as much suffering to shrimps as decreasing human living time by 43.6 d, which is associated with smoking 4.32 k cigarettes.
My estimates involve major uncertainty.
Nevertheless, I believe one can conclude that eating cows is the best / least bad.

Methods

I calculated the badness of eating farmed animals as a fraction of the goodness of the human living time supported by their calories from the product between:

Edible mass of farmed animals needed to satisfy the human daily caloric requirement (kg/d), which I computed from the ratio between:
- Human daily caloric requirement, which I set to 2250 kcal/d^[1].
- Calorie density of the edible mass of farmed animals (kcal/kg), which I retrieved from the United States Department of Agriculture’s FoodData Central, except for the 127 kcal/(100 g) of crickets.
Living time of farmed animals per edible mass (d/kg), which I estimated from the ratio between:
- Lifespan of the farmed animal (d), which I took from these data collected by Dominik Peters for ethical.diet, except for the 6 weeks of crickets.
- Edible mass of the farmed animal (kg), which I got from the data mentioned just above, except for the 0.280 g of crickets.
Badness per unit time of the lives of farmed animals as a fraction of the goodness per unit time of human lives. This is the major driver for the uncertainty, and I obtained it here (see “Intensity of the mean experience as a fraction of that of humans”) based on:
- Rethink Priorities’ median welfare range estimates. For beef and dairy cows, and lambs, I used pigs as the reference. For turkeys and ducks, chickens. For crickets, silkworms.
- The badness per unit time as a fraction of the median welfare range of the lives of farmed animals equals that of broilers in a reformed scenario. I did this for simplicity, but I believe there are significant differences across farmed animals (e.g. I think cows have better lives than what I assumed).
- The time broilers experience each level of pain defined here (search for “definitions”) in a conventional and reformed scenario is given by these data (search for “pain-tracks”) from the Welfare Footprint Project (WFP).
- Excruciating pain being the worst possible experience, and 1 k times as bad as disabling pain^[2].
- Disabling pain being 100 times as bad as hurtful pain.
- Hurtful pain being 10 times as bad as annoying pain.
- Being awake being as good as hurtful pain is bad (which may underestimate the badness of broilers’ lives). This means being awake with hurtful pain is neutral, thus accounting for positive experiences.

To get a more intuitive feel of the badness of eating farmed animals, I have also determined how many smoked cigarettes would be associated with a decrease in life expectancy respecting a loss in human welfare equivalent to the harm caused to farmed animals. I calculated the badness of eating farmed animals in terms of smoked cigarettes per energy provided by their calories from the ratio between:

Badness of eating farmed animals in terms of decreasing human life expectancy per energy provided by their calories (d/kcal), which I computed from the ratio between:
- Badness of eating farmed animals as a fraction of the goodness of the human living time supported by their calories, as described in the 1st paragraph of this section.
- Human daily caloric requirement of 2250 kcal/d.
Decrease in life expectancy associated with smoking a cigarette (d), which I set to 0.0101 d (= 0.487*0.5/24). Smoking one cigarette is associated with a reduction in life expectancy of 0.487 microlives (= (10 + 9)/2/((15 + 24)/2)), in agreement with 15 to 24 cigarettes correlationally reducing the life expectancy of men by 10 microlives, and women by 9.

Note:

I have only considered nearterm direct effects. Here are my thoughts on the indirect effects.
I assumed unitary elasticity of production with respect to consumption. Relatedly, there is this in-depth discussion.

The data and calculations are here.

Results

The inputs and results are in the tables below, ordered by descending badness of eating farmed animals as a fraction of the goodness of the human living time supported by their calories.

Farmed animals	Calorie density of the edible mass of farmed animals (kcal/(100 g))	Lifespan of the farmed animal (d)	Edible mass of the farmed animal (kg)	Badness per unit time of the lives of farmed animals as a fraction of the goodness per unit time of human lives
Crickets	127	42.0	0.280 m	3.87 m
Shrimps	99.0	180	0.0250	0.0599
Broilers in a reformed scenario	149	70.0	1.83	0.642
Salmon	231	720	2.73	0.108
Turkeys	213	126	10.3	0.642
Hens in a cage-free aviary	520	504	20.0	0.642
Ducks	337	45.0	3.50	0.642
Lambs	192	180	60.0	1.00
Pigs	174	183	91.1	1.00
Beef cows	310	402	339	1.00
Dairy cows	61.0	2.01 k	50.4 k	1.00

Farmed animals	Edible mass of farmed animals needed to satisfy the human daily caloric requirement (kg/d)	Living time of farmed animals per edible mass (d/kg)	Badness of eating farmed animals as a fraction of the goodness of the human living time supported by their calories
Crickets	1.77	150 k	1.03 k
Shrimps	2.27	7.20 k	981
Broilers in a reformed scenario	1.51	38.3	37.1
Salmon	0.974	264	27.8
Turkeys	1.06	12.2	8.29
Hens in a cage-free aviary	0.433	25.2	7.00
Ducks	0.668	12.9	5.51
Lambs	1.17	3.00	3.50
Pigs	1.29	2.01	2.59
Beef cows	0.726	1.19	0.857
Dairy cows	3.69	0.0398	0.146

Farmed animals	Badness of eating farmed animals in terms of decreasing human life expectancy per energy provided by their calories (d/(100 kcal))	Badness of eating farmed animals in terms of smoked cigarettes per energy provided by their calories (1/(100 kcal))
Crickets	45.7	4.52 k
Shrimps	43.6	4.32 k
Broilers in a reformed scenario	1.65	163
Salmon	1.24	122
Turkeys	0.369	36.5
Hens in a cage-free aviary	0.311	30.8
Ducks	0.245	24.2
Lambs	0.156	15.4
Pigs	0.115	11.4
Beef cows	0.0381	3.77
Dairy cows	6.50 m	0.644

Discussion

According to my results, eating farmed crickets and shrimps causes the most suffering (among the animals I have analysed). My guess is that eating:

A given mass/energy of crickets causes 1.03 k times as much suffering to crickets as the happiness of the human living time supported by such mass/energy (1 d for 2250 kcal).
100 kcal of shrimps causes as much suffering to shrimps as decreasing human living time by 43.6 d, which is associated with smoking 4.32 k cigarettes.

So I am glad The Insect Institute and Shrimp Welfare Project exist!

Beef and dairy cows are the only farmed animals whose consumption leads to an increase in welfare when considering both humans and animals. The badness of eating them as a fraction of the goodness of the human living time supported by their calories is lower than 1. Moreover, I believe I am overestimating the badness of the living conditions of cows.

The estimates involve major uncertainty. For example, the 5th and 95th percentile smoked cigarettes per 100 kcal of shrimps are 0 and 160 k (= (4.32 k)*37.1):

Conditional on shrimps having bad lives.
Supposing the welfare range is the major driver of the uncertainty. This may not be true, given the sign uncertainty of the badness per unit time of the lives of farmed animals (not only of shrimps, but also of other animals).
Having in mind Rethink Priorities’ 5th and 95th percentile welfare range of shrimps are 0 and 37.1 (= 1.149/0.031) times their median welfare range.

Nevertheless, I believe one can conclude that eating cows is the best / least bad. Besides being associated with the best / least bad direct impact, they also:

Require the most feed and land, and therefore increase the amount of food that can be produced and directed to humans if yields drop during nuclear/volcanic/impact winters.
Have the greatest carbon footprint, but more global warming might be good to mitigate the food shocks caused by nuclear/volcanic/impact winters.
Have huge impacts on wild animals, but I guess these, unlike the ones above, do not have longterm implications.

The impact per calorie of consuming animals on their welfare and climate change is also available in Food impacts, whose methodology also relies on Rethink Priorities’ median welfare range estimates, but assumes distinct badness per unit time as a fraction of the median welfare range for different farmed animals (whereas I suppose that of broilers in a reformed scenario applies to all farmed animals). Setting the weight of the welfare impacts to 99 %, one gets a similar order of animals to mine.

^
Mean between the 2000 and 2500 kcal/d recommended by the National Health Service for women and men.
^
I encourage you to check this post from algekalipso, and this from Ren Springlea to get a sense of why I think the intensity can vary so much.

Badness of eating farmed animals in terms of smoking cigarettes

Summary

Methods

Results

Discussion