by Brian Tomasik
First written: 2007; Last update: 17 Dec. 2014

Summary

Consuming equal weights of different animal products may produce different expected amounts of direct suffering. Farmed seafood may cause the most, followed by poultry products. Pork, beef, and especially milk produce considerably less suffering in comparison. As an extreme case, creating demand for a kg of farmed catfish meat is expected to cause 20,000 times as much direct suffering as creating demand for a kg of milk.

However, direct suffering is not the only consideration. Beef has a bigger climate-change impact per calorie than the other types of meat, though milk's impact is considerably lower than beef's. If you must eat animal protein, dairy products -- milk, cheese, or whey protein (which is low in fat and carbs) -- may be a good compromise. Holden Karnofsky has developed a "Power Smoothie" with whey protein. Of course, if you just want high-density protein, consider the various plant-based protein powders: pea, hemp, spirulina, etc. More than just soy is available. (There's also rice protein powder, but I tend to avoid rice in case the impact on wild animals is worse than that of other plants. I haven't verified whether this is actually the case, though.)

By the way, vegans need non-dietary vitamin B12, and vegetarians/vegans should consider supplementing with creatine for a possible cognitive benefit.

Note: In this piece I haven't yet analyzed lobster, shrimp, prawn, or other small seafood, but given the small size of these organisms, eating them presumably causes vast amounts of expected suffering per kilogram. While there is debate about whether these animals can suffer, the probability is not too low (I'd say at least 1/3 if not >50%) and hence does little to constrain the enormous expected harm resulting from their consumption.

Introduction

It is clear that animals living on factory farms endure great amounts of suffering, enough that it is wrong to bring them into existence by creating economic demand for meat.

If readers are unconvinced on this point, I recommend the vast collection of literature available online, including "Why Vegan?", "What Came Before," and, for a more rigorous survey, HSUS white papers.

In considering the suffering of farm animals, we presumably ought to care equally about equal amounts of suffering, regardless of which animals experience that suffering. It does not follow, however, that we ought to avoid all animal-based food products with equal amounts of effort. Eating certain types of meat may cause more suffering than eating the same amount of another type of meat under otherwise identical circumstances.

Below, I investigate how much direct suffering, on average, is caused by creating demand for a kilogram of different types of animal products. By "direct" suffering, I refer only to the suffering of the animal whose flesh, milk, or eggs is part of the animal product. I'm ignoring, therefore, the suffering of male chicks in being ground up as part of the egg-production process, the suffering of calves separated from their mothers in milk production, the contribution of milk production to the veal/beef industry, the large amounts of wild fish that are caught and fed to farmed fish, the fraction of farm animals that die before reaching slaughter, the environmental destruction that meat production entails, and potential health consequences of eating certain types of meat. Readers should consider these factors, too -- perhaps incorporating them into the computations below.

Also note that I haven't weighted animals differently based on cognitive capacities. There are some arguments suggesting that bigger brains may matter more in ethical calculations, but I personally am not completely sold on them. At the very least, I think small brains matter more than their proportional number of neurons because they're more efficient, and also because they belong to a unique agent.

Results

Here is a table of data on various animal foods.

  • Column 4 computes a straightforward value for the expected number of days of life on a factory farm that purchasing a kg of an animal food causes animals of that type to endure.
  • Column 5 represents my best-guess estimates for how bad life is per day for each of those animals. For instance, since I think the suffering of hens in battery cages is perhaps 2.5 times as intense, on average, as the suffering of beef cows, I put a "1" in the beef-cow entry and "2.5" in the egg column. The overall results are relatively insensitive to these assignments, but feel free to use your own values instead.

  • Column 6 is a subjective estimate of the average pain of slaughter for each animal, expressed in terms of an equivalent number of days of regular life for that animal. For instance, I used "5" as an estimate for broiler chickens, which means I assume chickens would, on average, be indifferent between enduring the stress of slaughter and living for five more days in their usual conditions. (Update from Nov. 2013: I now think this estimate is far too low -- that the pain of slaughter might represent most of the total pain of a factory-farmed chicken's life. However, I've left the original calculations as they were, in part because most people think the pain of slaughter is less bad than I do.)
  • Finally, I computed "Suffering per kg" in column 7 as follows:

    column 7 = [(column 2) + (column 6)]*(column 5)/(column 3),

    which, dimensionally, looks like this:

    suffering / kg = [(days of life / animal) + (equivalent days of death pain / animal)] * (suffering / day) / (kg / animal).

Column 1 2 3 4 5 6 7
Animal or Food Product Average lifespan (days) Average amount of food produced per lifetime (kg) Expected days of life caused per kg of meat demanded Suffering per day of life (beef cows = 1) Number of days of life equivalent to pain of death Expected suffering caused per kg demanded
Farmed catfish 730 0.48 1521 0.9* 10 1388
Farmed salmon 730 2.73 267 0.9* 10 244**
Battery-cage eggs 365 15.3 23.9 2.5 3.6 63
Chicken 42 1.83 23.0 1.8 5 46
Turkey 126 10.3 12.2 1.8 5 23
Pork 183 91.1 2.01 1.8 5 3.7
Beef 402 339 1.19 1 9 1.2
Milk 2009 50,420 0.040 1.8 5 0.07

* I think the suffering of farmed fish may actually be greater than this, perhaps about the same as the average suffering of meat chickens.

However, while the probability is close to 1 that land farm animals can suffer, the scientific jury is still out on whether fish can as well. I use a 50% probability of fish suffering to be conservative, though my actual probability is somewhat higher, perhaps 80%.

** This analysis counts only the suffering of the farmed fish. Importantly, it ignores the many times more small fish that are killed to provide feed to these larger fish. When this is taken into account, it's possible that large farmed fish like salmon outweigh smaller fish in per-kilogram impact.

Notes

Readers should feel free to substitute their own estimates into columns 5 and 6; for instance, if they buy meat from a non-factory farm, the values in column 5 may be smaller.

It's important to remember that the above figures are per kg demanded, not per kg eaten necessarily. If you get your meat from dumpster diving, you're not causing suffering to other animals. There may be less extreme circumstances, too, in which eating animal products doesn't contribute, or is less likely to contribute, to animal suffering. (See "Does Vegetarianism Make a Difference?")

Similarly, per kg doesn't necessarily mean per serving. If your choice is between ordering a two-egg omelet (roughly 0.1 kg of egg) versus a half pound of chicken, then the latter would cause more direct expected suffering (10.4 suffering units = 0.23 kg * 46 suffering units/kg) than the former (6.3 suffering units = 0.1 kg * 63 suffering units/kg).

In addition to looking at suffering per kg, we can also look per calorie or per gram of protein. Per calorie would make sense if people ate a fixed proportion of their calories as meat, since how much people eat is mostly regulated by a calorie budget. Insofar as people eat meat for protein, suffering per gram of protein is more appropriate, because meats with higher protein density require fewer calories for the same protein load. There may also be random peculiarities of how much meat of various types is eaten in various dishes with various amounts of additional plant-based calories. Dairy's suffering per calorie will be somewhat higher than its suffering per kg because of its high water content, but the difference isn't huge, and the conclusions are basically the same either way. Julia Galef wrote an article, "Want to Kill Fewer Animals? Give Up Eggs," that did similar computations as I did here but on a per-calorie basis. (She also counted animals killed rather than days of suffering.)

Finally, it's worth noting that different animal products have different own-price elasticities. In "Comments on Compassion by the Pound," you can see that cutting consumption of eggs results in a larger reduction in egg production than reducing consumption of milk or beef results in decreases of production for those products.

Updated thoughts on this piece, May 2014

I'm somewhat less gung-ho about these numbers than when I first wrote the piece because

  • Brain complexity matters somewhat and isn't incorporated. I think 40 chickens matter many times more than a cow, but not fully 40 times, whereas these figures imply ~40 times more suffering per kg of chicken than kg of beef.
  • In practice, indirect effects dominate, though of course, they're also much harder to figure out. Beef is good if it reduces wild-animal habitat but probably net bad by contributing to climate change. On balance it may be net bad if climate change dominates.

Whatever the sign is of the indirect effects, it should be more similar across animals than the suffering figures are across animals, since cows and chickens don't differ by ~40 times in their environmental impacts. Hence, these neglected factors should tend to drive the impact estimates (much) closer together.

I think the most robust conclusion from these numbers is that milk significantly differs from other animal products because it has low direct impact even if you weigh cows more heavily than smaller animals, and its climate-change contribution isn't that bad either. This is why I'm lacto-vegetarian. If I were going to popularize anything from this analysis, it would be something like, "For people who don't want to go vegan, consider going lacto-vegetarian." I think the exercise of computing these numbers is a healthy way to explore the details of animal production and challenge vague intuitions we may have about how much harm is done by a burger vs. a chicken wing, but I also wouldn't recommend taking these figures too literally.

Non-human carnivores

Some non-humans eat more individual animals, and even more total meat, than humans do. A typical non-vegetarian US consumer directly kills a bit more than one animal per day, although the number of land animals directly killed per day is only about 1/12. US per-capita meat consumption is about 271 pounds per year. In contrast:

  • "Adult Emperor Penguins consume 2-5 kg (4.4-11 lb) of food per day except at the start of the breeding season or when they are building up their body mass in preparation for molting. Then they eat as much as six kilograms (thirteen pounds) per day." (source) "Each penguin eats about one pound of fish per day." (source) "Each of the Zoo's penguins eat[s] 2 pounds of fish per day" (source)
  • "If you do the math, you will find that one barn owl needs to eat about 79 pounds of mice a year." This amounts to about 3.5 individual mice per day. (source)

Of course, animals eaten by non-human predators aren't factory farmed, but their deaths are arguably worse than those in human slaughterhouses on average because there's not even a perfunctory attempt at stunning. Imagine being a fish swallowed live by a penguin. While human meat consumption causes lots of suffering, so does predation in nature.

Acknowledgments

This piece was inspired by Table 1, p. 583, of Human diets and animal welfare: The illogic of the larder by Matheny, G. and Chan, K.M.A. Similar figures appear in Table 1, pp. 326-27, of Farm animal welfare, legislation, and global trade by Matheny, J.G. and C. Leahy.

Calculations similar to, but much more detailed than, those in this piece are done in Bailey Norwood's Compassion by the Pound.

Links to this piece

There's a nice infographic of this analysis on Think by Numbers.

Appendix: Where the numbers came from

Here is a list, by animal, of where I got my assumptions for the values in columns 2 and 3 of the table above. In general, a good resource for this type of information is the National Agricultural Statistics Service of the US Department of Agriculture.

Unfortunately, since I compiled these figures in 2007, most of the links have since rotted away. I've replaced some, but not all, of the links to new sources that have roughly the same information.

Farmed Catfish

Catfish take 18 to 36 months to grow. I assume the average is 2 years.

The average live weight of catfish is 1.5 pounds (USDA NASS Census of Aquaculture 2002, p. 3). I assume that 70% of this weight ends up as edible meat.

Farmed Salmon

Salmon take 18 and 24 months to grow. I assume 2 years, as for catfish.

The average live weight of food salmon is 8.6 pounds (USDA NASS Census of Aquaculture 2002, p. 6). I assume that 70% of this weight ends up as edible meat.

Eggs

I assume a laying hen survives for a year on average. (Note that the exact figure is relatively unimportant, because there are more precise data on the average numbers of eggs laid per year. The lifespan of a hen is only relevant to factoring in the painfulness of death.)

In 2006, hens laid an average of 263 eggs (USDA NASS Chickens and Eggs Annual Summary 2006). I assume an egg weighs about 0.05 to 0.06 kg without a shell. (I used 0.058 kg in the calculation for convenience in correcting a small numerical error I made in the past.)

Chicken

A typical lifespan for a broiler chicken is 42 days.

In May 2007 in the US, 795,639,000 chickens were slaughtered, yielding a final edible weight of 3,213,272,000 pounds (USDA NASS Poultry Slaughter 06.29.2007).

Turkey

"Turkeys are slaughtered at between 12 and 26 weeks, depending on the size of bird being produced" (source). I assume 18 weeks.

In 2007 in the US, 22,536,000 turkeys were slaughtered, yielding a final edible weight of 509,529,000 pounds (USDA NASS Poultry Slaughter 06.29.2007).

Pork

I assume the lifespan of a typical pig is half a year.

In May 2007 in the US, 8.76 million pigs were slaughtered to produce 1.76 billion pounds of edible meat (USDA NASS Livestock Slaughter 06.22.2007).

Beef

I take a cow's lifespan to be 1.1 years, or 402 days. I count here both the six months that cattle spend in feedlots and the seven months they spend grazing (see this page and references therein). The period of grazing contributes to my relatively low estimate for suffering per day of life.

In May 2007 in the US, 2.28 billion pounds of edible beef were produced from 3.05 million head (USDA NASS Livestock Slaughter 06.22.2007).

Milk

Factory-farmed dairy cows typically live 5-6 years (I assumed 5.5).

In 2006 in the US, the average milk cow produced 20,210 pounds = 9,167 kg of milk. Multiplied by 5.5 years, this amounts to 50,420 kg per cow.